Cathy O'Neil

Key Insight: When political leaders have personal billions at stake in a financial bubble, the inevitable government intervention to prevent its collapse will be even more corrupt than past bailouts.

O'Neil draws a parallel between the 2008 financial crisis and the current cryptocurrency bubble, arguing that both involve government protection of markets for political reasons. She recounts her early skepticism of Bitcoin evangelists at Occupy Wall Street who couldn't explain how crypto would democratize banking. The post's central concern is that Trump and his family are so heavily invested in crypto that they have personal financial incentives to prevent a market crash, creating a more direct and troubling conflict of interest than even the Obama-era bailouts.

Key Insight: The AI debate should move past abstract arguments about machine cognition and focus on the concrete ways the technology is reshaping work and society—primarily to the benefit of those profiting from it.

O'Neil responds to a Matteo Wong article in The Atlantic arguing that both AI hype men and critics are missing the point. She acknowledges that dismissive critiques like calling AI 'a racist pile of linear algebra' may understate its real-world impact. The core argument is that debates about whether AI 'thinks' are beside the point when it's actually replacing workers. O'Neil advocates for taking AI seriously as a transformative technology while remaining clear-eyed that it primarily benefits those making billions from it.

Key Insight: LRMs' failure on Tower of Hanoi at 10 discs isn't about complexity scaling - it's proof they never understood the recursive solution at all, despite being trained on thousands of explanations of it.

Cathy O'Neil critiques an Apple paper that tests Large Reasoning Models (LRMs) against Large Language Models (LLMs) on problems of increasing complexity. While the paper shows both fail at complex tasks, O'Neil argues it misses a deeper point: the Tower of Hanoi puzzle isn't actually harder for more discs once you understand the recursive solution. The fact that LRMs fail at 10 discs when the solution is trivially iterative proves they never truly understood the concept. She calls for abandoning the word 'thinking' to describe these models, as they're merely predicting tokens, not achieving genuine mathematical comprehension.

Key Insight: The VA job cuts combined with proposed AI regulation bans reveal a coordinated plan to create captive government markets for unregulated AI products at the expense of veterans' care.

O'Neil argues that the deployment of LLMs in the VA represents a dangerous turning point for AI adoption. While she has been cautiously observing chatbot adoption, noting most implementations have been overhyped and error-prone, the planned 83,000 DOGE job cuts to the VA signal something more sinister. She contends this is a deliberate strategy to replace critical healthcare staff with AI, creating a captive federal customer base for Silicon Valley while simultaneously pushing for a 10-year moratorium on AI regulation that would eliminate oversight.

Key Insight: The framing of AI as inevitable—whether as helpful agent, job destroyer, or existential threat—is propaganda designed to bypass critical evaluation and force adoption of technology we should simply refuse to cede control to.

O'Neil responds to three AI articles—about AI shopping agents, AI eliminating entry-level jobs, and AI causing human extinction—by rejecting all three narratives as propaganda designed to make AI feel inevitable. She argues AI isn't actually good at these tasks, shouldn't be given control over finances, and certainly isn't conscious enough to kill humanity. Her solution is simple: don't cede control to AI in the first place, and maintain human oversight of important tasks like legal briefs and accounting.

Key Insight: A cartoon comedy special does a better job unpacking the systemic issues around obesity and weight loss drugs than years of serious journalism on the topic.

O'Neil praises a South Park special episode called 'The End of Obesity' for its satirical take on weight loss culture, fat shaming, and the pharmaceutical industry. She notes the episode skillfully critiques multiple targets simultaneously: wealthy women using drugs for vanity, the inaccessibility of medications for those who need them most, body positivity performativity, healthcare bureaucracy, and pharmaceutical profiteering. She contrasts the show's incisive commentary favorably against mainstream journalism on the topic.

Key Insight: AI researchers are projecting consciousness and moral capacity onto statistical pattern-matching systems, revealing more about human self-deception than about machine intelligence.

O'Neil argues that AI researchers are exhibiting cultish behavior by treating language models as if they possess consciousness and honesty. She critiques the AI Futures Project for making speculative predictions reminiscent of flawed COVID models and for using language that presupposes LLMs have minds that could 'lie' or be 'honest.' She emphasizes that LLMs merely predict the next word based on training data and have no notion of truth or consciousness. The post concludes that projecting consciousness onto algorithms built with Reddit data is itself a form of self-deception.

Key Insight: Democrats have become so fixated on electoral math and polling data that they've forgotten politics is fundamentally about persuading actual human beings, not optimizing turnout in purple districts.

Cathy O'Neil celebrates Bernie Sanders and AOC's 'Fighting Oligarchy Tour' through red states as a necessary revival of Occupy-era populism. She recounts an MSNBC exchange where pundits couldn't fathom why Democrats would campaign in Utah except for media attention. O'Neil argues this reveals the Democratic Party's fundamental blindness: they've replaced genuine voter engagement with poll-watching and surgical electoral strikes. The party has forgotten that changing minds requires actually talking to people, not optimizing turnout models. This strategic myopia, she contends, is why red states keep getting redder.

Key Insight: The AI industry shows classic bubble dynamics: massive capital expenditure and ubiquitous marketing for products that lose money and lack clear demand.

O'Neil observes that AI advertising has saturated public spaces like Bay Area billboards and airports, suggesting a classic bubble pattern. She argues that VC money has flooded into AI startups with the mandate to build first and find customers later, resulting in mediocre products nobody wants. She supports her skepticism with striking financial data: Big Tech spent $246 billion on AI infrastructure in 2024, OpenAI loses $2 for every $1 it makes, and trillion-dollar investments are planned despite unclear demand. She predicts this bubble will burst badly, possibly with Trump's tariffs taking the blame.

Key Insight: Accurately tracking the sun's position requires modeling Earth's elliptical orbit with iterative numerical methods rather than simple circular approximations.

This is a guest post by Sander O'Neil (likely Cathy's son) presenting a JavaScript-based clock/map/star chart that tracks the sun's position relative to Earth. The project combines orbital mechanics with geographic visualization, showing the sun's current and future positions on both a polar star chart and an equirectangular world map. The implementation uses Wikipedia-sourced orbital dynamics data and iteratively solves for Earth's position using mean anomaly calculations. The post explains each visual element and the mathematical principles behind the simulation, including handling Earth's elliptical orbit and axial tilt.

Key Insight: The utilitarian logic underlying effective altruism and techno-optimism, when combined with cost-benefit analysis, creates a framework that systematically devalues disabled lives and has historically been used to deny them basic rights and dignity.

Victor Zhenyi Wang, a data scientist with a disability, recounts attending a rationalist conference where philosopher Jonathan Anomaly presented on 'liberal eugenics' - voluntary genetic selection for traits like intelligence - to an uncritical audience. Wang challenges the utilitarian premise that disabled lives are 'worse off,' citing disability rights advocates who argue their lives are valuable and that disability-related challenges stem from societal structures, not inherent deficits. He critiques how metrics like QALYs inherit society's biases against disabled people and how cost-benefit analysis has historically been weaponized to deny disabled people basic rights and accommodations. Wang warns that techno-utopian visions built on these frameworks would systematically marginalize disabled people, returning to an era of institutionalization.

Key Insight: Effective Altruism has evolved from a practical charitable giving movement into a cult-like ideology that uses philosophical abstraction about the far future to avoid doing tangible good in the present while demanding total lifestyle commitment from young recruits.

This interview features a former Effective Altruism participant who joined through philosophy classes and the One for the World pledge organization, eventually working there after college. She describes her growing disillusionment with EA as it shifted from practical global health work toward abstract longtermism and AI safety concerns. The interviewee criticizes EA's culture of hyper-rationality, its treatment of time as an economic resource (having assistants do laundry), and its cult-like recruitment tactics targeting young students. She sees the movement as dominated by men who use philosophical questions as self-indulgent intellectual exercises while claiming moral legitimacy, and notes the deliberate pivot away from tangible short-term impact toward speculative future scenarios.

Key Insight: Fairness and diversity cannot be implemented as universal rules at scale—they require contextual judgment that tech companies are too lazy or cheap to provide.

Google's Gemini image generator produced historically absurd images like black Nazis and female popes because the company overcorrected for a known problem: datasets dominated by white male faces. Rather than thoughtfully addressing diversity in context, Google applied a blanket 'diversity at scale' approach that ignored historical and cultural specificity. O'Neil argues this reflects a fundamental misunderstanding that equity and fairness can be automated universally, when in reality these concepts require narrow, contextual application.

Key Insight: When media lets powerful tech executives be the primary voices on AI ethics and fairness, journalism has collapsed into corporate propaganda.

Cathy O'Neil critiques a CBS 'Eye on America' segment that gave uncritical airtime to Big Tech leaders like Yann LeCun to discuss AI fairness. She highlights the irony of LeCun now discussing AI fairness when he previously dismissed it as 'above his paygrade.' The piece examines how mainstream media amplifies industry talking points without skeptical voices, noting LeCun's contradictory positions on regulation and autonomous weapons. O'Neil argues this represents a collapse in AI journalism where powerful tech figures control the narrative about their own products' ethics.

Key Insight: Sometimes achieving policy change requires reframing basic concepts in jarring clinical terms to force those in power to actually confront what they're legislating.

O'Neil celebrates the progress of the tampon tax abolition movement, highlighting her friend Laura Strausfeld's successful advocacy work featured in a Washington Post op-ed. The post reflects on O'Neil's own involvement as a plaintiff in a New York lawsuit against the tampon tax years earlier. She explains how Strausfeld strategically coined the clinical term 'menstrual discharge collection device' (MDCD) to help male legislators understand these products' necessity. The movement is gaining momentum, with Texas recently eliminating its tampon tax.

Key Insight: The intersection of abortion restrictions and paternity testing technology may force men—especially wealthy ones—to confront reproductive politics as a personal financial issue rather than just a women's issue.

O'Neil argues that abortion restrictions may unexpectedly become a men's rights issue due to paternity testing technology. Historically, men could deny paternity to avoid responsibility, but DNA tests now make this impossible. With abortion access restricted in some states, men can no longer pressure partners into terminating pregnancies to avoid financial obligations. This creates a new dynamic where wealthy, powerful men who previously would have quietly supported abortions may now face court-ordered child support, potentially shifting the politics of reproductive rights.

Key Insight: Algorithmic auditing should focus on measuring real-world harm to stakeholders using actual historical data, not on achieving theoretical 'understanding' of how systems work internally.

Cathy O'Neil responds to an academic paper claiming black-box auditing is insufficient for AI systems, systematically dismantling each of their five criticisms. She argues the academics misunderstand how real-world auditing works: auditors use actual historical data rather than synthetic inputs, focus on outcomes rather than understanding internal mechanisms, and deliberately avoid the conflict of interest that comes from also being the engineers who fix problems. Her core contention is that auditing should measure harm to stakeholders, not explain how algorithms 'think.'

Key Insight: The debate over weight-loss drugs conflates cosmetic vanity with genuine medical need, and both left and right critiques ultimately perpetuate the same shame they claim to oppose or reject.

O'Neil defends her use of Wegovy after losing 60 pounds, arguing that criticism from both left and right stems from misguided notions about shame and weight. The right believes weight loss should be painful repentance, while the left argues people shouldn't want to lose weight at all. She distinguishes between cosmetic weight loss (which she dismisses) and medically necessary treatment for those at serious health risk. Drawing on Susan Sontag's work on illness stigma, she predicts anti-obesity medication will become normalized just as cancer treatment shed its moral baggage.

Key Insight: It's hypocritical to cite evidence that diets don't work while simultaneously reinforcing the narrative that weight is a personal failing to be corrected.

O'Neil critiques a Washington Post food columnist who writes about how diets statistically don't work, only to undermine her own argument by describing her personal weight loss success. The columnist acknowledges that diets are shams that exploit people's desires, yet concludes by saying she personally diets because 'being fat made me unhappy.' O'Neil argues this creates a contradictory message that implies weight is a personal choice and that dieting does work if you just try hard enough. She calls out the cognitive dissonance of simultaneously debunking diets while engaging in fat shaming.

Key Insight: Dignity-preserving social media may require sacrificing frictionless broadcasting for invitation-based spaces with shared purpose—because when people have something substantive to discuss, they're too busy to attack each other's identities.

O'Neil introduces a collaborative philosophy project exploring whether dignity can be embedded into social media design, using Donna Hicks' ten essential elements of dignity as a framework. She starts with 'Acceptance of Identity'—the ability to express authentic identity without fear of judgment—and immediately confronts the difficulty: automated hate speech detection fails, and human moderation is expensive and harmful to workers. She proposes a 'hotel room' model where spaces are private by default, require invitation, and have active moderation, contrasting it with current broadcast-friendly platforms. She acknowledges this may be impossible but points to Ravelry's knitting community as evidence that focused, purpose-driven spaces can minimize identity-based attacks.

Key Insight: Large language models can only replicate the superficial patterns of human expertise while remaining fundamentally incapable of the genuine understanding and novel thinking that makes that expertise valuable.

O'Neil argues that ChatGPT deserves neither the fear nor celebration it's receiving. The technology lacks any model for truth—it simply mimics patterns from training data without genuine understanding. While it may help students cheat, this merely exposes how weak our existing proxies for measuring learning already are. She points to Meta's Galactica as evidence that these systems are essentially 'bullshit generators' that can replicate the easy, surface-level aspects of writing while being fundamentally incapable of producing new ideas or genuine insight.

Key Insight: The negotiation over ranking formula changes exposes that algorithmic scores are not objective truths but subjective value judgments that derive false authority from mathematical presentation.

O'Neil examines the ongoing rebellion of law schools against US News & World Report rankings, sparked partly by revelations that Columbia falsified self-reported data. She traces how the rankings became the first 'Weapon of Math Destruction'—an opaque scoring system that warped higher education by incentivizing gaming rather than genuine improvement. The current negotiation between US News and law deans over formula tweaks perfectly illustrates her core thesis: algorithms are opinions embedded in math. The disagreement over what factors to weight reveals that rankings have always been subjective value judgments masquerading as objective measurement.

Key Insight: Even when ending a project, preserving the archive allows past work to continue providing value to readers.

This brief post announces that Cathy O'Neil has stopped actively blogging on mathbabe.org. The site will remain online as an archive of her previous work. She expresses hope that readers will continue to find value in the archived content.

Key Insight: Sometimes a blog post is simply a moment of celebration marking the arrival of years of work in physical form.

This is a brief announcement post where Cathy O'Neil shares her excitement about receiving physical copies of her new book 'The Shame Machine.' The post consists of a single enthusiastic sentence and a photo of the books. It serves as a personal milestone marker rather than substantive analysis. The post celebrates the tangible arrival of what would become her exploration of how shame is weaponized in society.

Key Insight: Blogging remains a uniquely valuable medium for developing ideas iteratively and building the intellectual foundation for larger works, even in an age dominated by social media.

Cathy O'Neil reflects on why blogging remains valuable in 2021, prompted by her son Aise starting his own blog. She defends blogging as an 'old fashioned' form against the dominance of Twitter and Facebook, arguing it remains uniquely suited for developing fully-formed ideas. She credits her own blog with enabling her to eventually write a book, and proudly highlights her son's early posts on economics topics like hyperinflation, housing markets, and labor reports.

Key Insight: The difference between a bubble and a shortage lies in whether home production rises or falls—and the data showing increased construction during the pandemic supports the bubble theory while contradicting the industry-promoted shortage narrative.

This guest post argues that rising home prices are driven by a speculative bubble rather than a genuine housing shortage, contrary to mainstream media narratives. The author points to historically low interest rates, white flight to suburbs, and self-sustaining speculation as the real causes. Key evidence includes rising home construction metrics (new home starts, residential investment, and new home sales) during the pandemic—data inconsistent with a shortage theory, which would predict declining production. The author suggests the 'shortage' framing is deliberately pushed by the home construction industry lobby to advocate for deregulation while keeping prices high.

Key Insight: Statistical methods that work well under normal conditions can produce misleading results when unprecedented disruptions like COVID-19 break the historical patterns those methods depend on.

This guest post argues that upcoming labor reports will overstate job growth due to how seasonal adjustments work. Seasonal adjustments are statistical techniques that account for predictable yearly patterns in economic data, but they rely on historical trends that COVID-19 has disrupted. Virtual schooling has weakened the usual summer employment drops in education and among parents (especially women) who normally leave the workforce during school breaks. Since seasonal adjustments will still correct for the historical pattern of summer job losses that won't actually occur this year, the adjusted figures will artificially inflate apparent job growth in May and June 2021.

Key Insight: High inflation readings in 2021 are a statistical artifact of comparing current prices to the anomalous COVID crash, not evidence of genuine inflationary pressure, and better measures like median CPI show no cause for alarm.

This guest post by Aise O'Neil argues that despite alarming headlines about inflation from both right-wing commentators and financial markets, the current high inflation readings are largely illusory. The author presents three reasons: standard inflation measures are distorted by comparing current prices to the anomalous crash during COVID lockdowns, more stable measures like median CPI show no concerning trends, and economic theory suggests inflation has significant inertia that prevents sudden jumps to hyperinflation. The post concludes that unemployment levels post-pandemic will be the real determinant of future inflation, and that Wall Street's inflation bets represent a temporary opportunity before markets recognize the readings as transitory.

Key Insight: The 2020 election demonstrated the decisive power of Black voters in swing states, making voter suppression through discouragement tactics less viable.

This is a brief announcement post where Cathy O'Neil shares her Bloomberg opinion piece about Black voter power in swing states following the 2020 election. She frames the piece as an uncharacteristically non-technical political commentary and a celebration of Stacey Abrams's voter mobilization work. The post serves primarily as a pointer to her external Bloomberg column rather than containing substantive analysis itself.

Key Insight: Political polling has become an emotional crutch that confirms our biases rather than a tool for genuine understanding, and we should treat it like an addiction to be broken.

O'Neil reflects on her disappointment with political polling after the 2020 election, admitting she fell for polls that told her what she wanted to hear. She argues that consuming polling data is more of an emotional addiction than genuine scientific inquiry. The post links to her Bloomberg Opinion column calling for Americans to break their dependence on political polls, suggesting that doubling down on polling methodology won't help the country understand itself better.

Key Insight: Political change alone won't solve systemic problems like big tech's unchecked power—the real work of reform begins after the election.

This brief post previews O'Neil's Bloomberg column arguing that a Biden victory would not mark an end to problems but rather the beginning of enormous catch-up work. She emphasizes that the challenges facing America extend far beyond reversing Trump-era damage, particularly regarding the unchecked power of big tech companies. The post frames the post-Trump period as a 'hangover' requiring sustained policy effort rather than celebration.

Key Insight: The principle behind healthcare's pre-existing condition protections—that people shouldn't be penalized for factors beyond their control—should be applied more broadly across society.

This brief post promotes O'Neil's Bloomberg column arguing that the concept of 'pre-existing condition' protections should be expanded beyond healthcare. She suggests the principle that people shouldn't be penalized for circumstances beyond their control has applications in many other domains of life, not just medical insurance.

Key Insight: Good news about a vaccine could become bad news if people prematurely abandon protective behaviors before the vaccine has actually made them safe.

O'Neil warns that the announcement of a safe and effective COVID-19 vaccine could paradoxically make the pandemic worse. Her argument centers on humanity's poor ability to anticipate feedback loops—people tend to overreact to good news by prematurely abandoning precautionary behaviors. She fears that vaccine announcements will trigger dramatic behavioral changes like discarding masks before the vaccine has actually been widely distributed and taken effect, undermining its potential benefits.

Key Insight: The power to manipulate recommendation algorithms is neither good nor bad—what matters is who wields that power and toward what ends.

O'Neil argues that TikTok's recommendation algorithm, like all recommendation engines, can be manipulated by whoever controls it to amplify or suppress any type of content. She points out that this capability to shape what users see is inherent to how recommendation systems work. Notably, she clarifies that the ability to manipulate these algorithms isn't inherently negative—it can be used for beneficial purposes as well as harmful ones.

Key Insight: The UK algorithmic grading failures aren't isolated incidents but early warnings of a coming wave of poorly designed algorithms that will increasingly harm students and educational systems.

Cathy O'Neil shares updates on her recent work covering algorithmic harms across multiple platforms. She announces a Bloomberg column warning that UK algorithmic grading scandals are harbingers of worse to come, a New York Times book review on gender in mathematics, and her appearance in the Netflix documentary 'The Social Dilemma.' She also highlights related work by Meredith Broussard on algorithmic grading and Yael Eisenstein on Facebook's polarization business model.

Key Insight: The dysfunction in school reopening plans reflects deliberate political choices rather than mere incompetence.

Cathy O'Neil argues that the chaotic school reopening plans during COVID-19 are not accidental failures but deliberate political choices. Based on conversations with educators, she concludes that students and teachers are being treated as pawns in a political game. The post links to her full Bloomberg column exploring how jurisdictions from New York City to Florida are mishandling reopenings. She frames the disaster as intentional rather than incompetent.

Key Insight: Trump's fixation on reducing testing to lower case counts reveals he operates like an algorithm—optimizing metrics while being indifferent to the underlying human reality those metrics represent.

O'Neil analyzes Trump's repeated claim that COVID-19 testing is the problem because it reveals too many cases. Rather than dismissing this as nonsense, she attempts to understand the mental model behind it. She concludes that Trump operates like an algorithm, treating metrics and data points as more real and important than the underlying reality they represent. This insight—that Trump prioritizes measurable outputs over human outcomes—explains his consistent focus on TV ratings and case counts over actual deaths.

Key Insight: Being wrong about a prediction doesn't mean the underlying analysis was flawed when the data itself is unreliable.

O'Neil acknowledges her prediction of 600 daily COVID deaths in Florida was wrong—the actual figure is 184—but argues the discrepancy may be due to data manipulation rather than genuine success. She points to narrow definitions of COVID deaths, lagging reporting, and changing methodologies as reasons to doubt Florida's official numbers. The post links to her Bloomberg column exploring why Florida's death toll appears lower than New York's despite high case counts.

Key Insight: When institutional responses fail, crowdsourced data collection can empower citizens to protect themselves.

O'Neil argues that the inadequate federal response to COVID-19 means citizens must take matters into their own hands. She proposes crowdsourcing CO2 level data in indoor spaces like schools, airports, and workplaces to assess ventilation quality. The post serves as a brief introduction to her longer Bloomberg piece on the topic. She frames this as a practical, citizen-driven solution to a public health information gap.

Key Insight: The pandemic exposed that governments can mobilize massive resources instantly when motivated, revealing the chronic dismissal of marginalized communities' needs as a political choice rather than an impossibility.

Cathy O'Neil reflects on her March 2020 prediction that the coronavirus pandemic would trigger a massive political uprising. While other 'thought leaders' predicted optimistic outcomes like reduced polarization and renewed faith in experts, O'Neil foresaw that the unequal response to the crisis would expose how wealthy communities would be protected while poor communities would be devastated. She argues that the pandemic revealed how massive political action is possible when causes are deemed urgent, making the chronic dismissal of marginalized populations' needs impossible to ignore. Writing in July 2020, she expresses sadness at being proven correct.

Key Insight: When algorithms secretly determine students' grades and futures without transparency or accountability, they become weapons of math destruction that can derail lives.

The post introduces a Bloomberg column about the International Baccalaureate Organization's grading algorithm that has negatively impacted students like Hadrien, who aspires to study engineering. O'Neil argues the algorithm operates in a secret, powerful, and destructive manner, making it a classic example of a 'weapon of math destruction.' The piece serves as a cautionary tale about algorithmic grading systems deployed during the pandemic.

Key Insight: Data analysis in July 2020 suggested Florida's COVID death toll trajectory could rival or exceed New York's catastrophic early pandemic numbers.

O'Neil expresses deep concern about Florida's COVID-19 trajectory in July 2020, predicting daily death counts would soon reach 600. Drawing on both personal connection to Florida and data analysis, she argues the state is particularly vulnerable to the pandemic—in some ways more so than Manhattan. The post serves as a teaser for her Bloomberg column where she explains her methodology for this grim prediction.

Key Insight: When both a trend and its acceleration are negative - quantified dread - optimism becomes dangerous rather than helpful.

O'Neil briefly updates readers on her move to Somerville, MA after 15 years, confessing to hoarding tendencies revealed by the process. The main substance is a pointer to her Bloomberg column introducing 'quantified dread' - a concept describing situations where things are getting worse and the rate of deterioration is accelerating. She applies this framework to America's COVID-19 response, arguing the country is dangerously complacent.

Key Insight: The same predictive policing algorithms that critics have long attacked for bias can now serve as a roadmap for identifying which police functions should be transferred to social services.

This brief post promotes O'Neil's Bloomberg column arguing that existing crime risk score data can be repurposed to guide police defunding efforts. She suggests that the same algorithmic tools used for predictive policing inadvertently reveal which social problems have been inappropriately delegated to law enforcement. The post positions this as making defunding practically straightforward since the analytical groundwork already exists.

Key Insight: Voluntary corporate moratoriums on facial recognition technology are meaningless without binding Congressional legislation.

This is a brief post linking to a Bloomberg column about facial recognition technology. O'Neil argues that corporate moratoriums on facial recognition, like Amazon's, are insufficient and that meaningful regulation requires Congressional action. The post serves as a pointer to her fuller argument published elsewhere.

Key Insight: The deliberate absence of police misconduct data is itself a form of institutional protection that enables continued abuse.

This brief post introduces O'Neil's Bloomberg column about missing data in police misconduct records. She frames the absence of police accountability data as a critical problem, describing missing data as 'the darkest, scariest kind of data.' The post advocates for filling these informational gaps as essential to police reform efforts.

Key Insight: The aggressive career ambition celebrated in 'Lean In' becomes a liability when it keeps leaders clinging to positions at ethically compromised companies rather than taking a moral stand.

This brief post links to O'Neil's Bloomberg column criticizing Sheryl Sandberg's leadership at Facebook during a crisis period. The post's title suggests that Sandberg's famous 'Lean In' philosophy of ambitious career advancement may be inappropriate when the company faces serious ethical problems. O'Neil implies that Facebook needs moral leadership rather than corporate ladder-climbing mentality. The timing (June 2020) places this during controversies around Facebook's handling of political content and misinformation.

Key Insight: Structural racism and mass incarceration aren't just social justice issues—they're public health infrastructure failures that amplify pandemics.

This brief post promotes O'Neil's Bloomberg column arguing that mass incarceration, inequality, and racism are root causes of pandemic vulnerability. Drawing on research from Measures for Justice, she contends that these interconnected social ills weakened community immune systems and made both the U.S. and the world more susceptible to COVID-19. The post serves as a gateway to her fuller argument that structural racism isn't just a social justice issue but a public health catastrophe.

Key Insight: The pandemic has turned college tuition into a high-stakes negotiation where institutions must finally justify whether their premium pricing delivers premium value.

This is a brief announcement post linking to O'Neil's Bloomberg column about the standoff between colleges and students/parents over tuition during COVID-19. The core argument is that colleges charging full tuition for online education creates a game of chicken scenario. O'Neil suggests that online education should be priced accordingly, and that the pandemic will force colleges to justify their value proposition.

Key Insight: When data quality is suspect, tracking the trend-based reproduction rate R(t) across multiple sources is more reliable than trusting any single dataset or complex model.

O'Neil argues that the reproduction rate R(t) is a more reliable metric for tracking Covid-19 than raw case counts or model predictions. She explains that R(t) can be estimated from multiple data sources (hospitalizations, cases, deaths) and is robust to testing biases since it only measures trends. However, she warns that deliberate political manipulation of data by states eager to reopen undermines even this approach. The piece advocates for following this single hyperparameter rather than getting lost in complex model outputs.

Key Insight: COVID-19 models vary dramatically in quality, and understanding their limitations is essential for interpreting pandemic predictions.

This is a brief pointer post directing readers to a Bloomberg column O'Neil wrote about COVID-19 epidemiological models. The column, informed by mathematician Jarod Alper's YouTube talk, examines how various COVID models work and their limitations. O'Neil signals her critical stance through the subtitle indicating that while no models are perfect, some fail more fundamentally than others.

Key Insight: Government algorithms that make life-and-death decisions without public transparency are dangerous weapons that undermine democratic accountability.

Cathy O'Neil identifies the new FEMA model used during the COVID-19 pandemic as a 'Weapon of Math Destruction' (WMD). She argues that this algorithm, which was reportedly used to pressure Arizona's governor into reopening early, meets her three criteria for a WMD: it's important (affects life-and-death decisions), secret (not open to public scrutiny), and destructive (potentially causing harm). The post links to her Bloomberg column calling for public transparency in reopening decisions.

Key Insight: Major disruptions like pandemics accelerate automation but may also create windows for workers to reshape the terms of that transition.

This brief blog post serves as a link roundup sharing two of Cathy O'Neil's Bloomberg opinion columns. The first column argues that pandemic-driven automation could present an opportunity for workers to negotiate better terms with their 'robot overlords.' The second is a personal reflection mourning the impact of COVID-19 on New York City's subway system. The post itself contains no substantive analysis, functioning purely as a pointer to her longer-form work elsewhere.

Key Insight: O'Neil consistently applies algorithmic and data-science frameworks to explain phenomena outside traditional tech contexts, from political behavior to public health policy.

This is a brief link-sharing post where Cathy O'Neil promotes two of her recent Bloomberg opinion pieces. The first draws an analogy between Trump's behavior and a malfunctioning machine learning algorithm that's 'running low on data.' The second addresses COVID-19 testing requirements for ending social distancing, arguing that test quality matters more than quantity. The post serves primarily as a pointer to her longer-form journalism rather than containing substantive analysis itself.

Key Insight: Data-driven pandemic responses risk encoding ageism into life-or-death resource allocation decisions.

This brief blog post announces a new Bloomberg column written in collaboration with ageism activist Ashton Applewhite about the dangers facing older people during the COVID-19 pandemic. The column argues that using biased pandemic data to allocate resources could perpetuate high mortality rates among the elderly. O'Neil connects her expertise in algorithmic bias with Applewhite's advocacy work to highlight how data-driven decisions during crises can discriminate against vulnerable populations.

Key Insight: The idealized 'flatten the curve' models gave the public false hope about the trajectory and duration of COVID-19 deaths.

This brief post links to O'Neil's Bloomberg column critiquing how the actual COVID-19 curve differed from the idealized 'flattened curve' models promised to the public. She frames it as an observation about the gap between models and reality, with a stark warning that deaths would continue longer than anticipated. The post serves primarily as a pointer to her longer analysis published at Bloomberg.

Key Insight: Contact tracing technology, like many algorithmic solutions, fails precisely where it matters most—protecting vulnerable populations who often lack access to the technology itself.

This brief post links to Cathy O'Neil's Bloomberg column arguing that COVID-19 contact tracing apps will be ineffective. The core critique is that such technological solutions fail to protect the most vulnerable populations. O'Neil positions this as part of her broader skeptical analysis of tech solutionism during the pandemic.

Key Insight: The daily Covid-19 statistics we rely on for understanding the pandemic are fundamentally unreliable, and we may never know the true scope of its impact.

This brief post announces a new Bloomberg column examining the reliability of Covid-19 data. O'Neil argues there are at least ten reasons to question the accuracy of the pandemic statistics being reported daily. She notes that delayed data from Easter weekend will likely compound existing data quality issues. The post suggests that the true impact of the pandemic may never be accurately measured due to systemic data collection problems.

Key Insight: Politicians promoting drugs with 'what do you have to lose' rhetoric ignore that millions of people on common medications like metformin could lose their lives due to dangerous drug interactions.

Gary Cornell, a mathematician with Type 2 diabetes, responds to Trump's promotion of hydroxychloroquine as a COVID-19 treatment by highlighting a Johns Hopkins study showing dangerous interactions with metformin—a drug taken by over 150 million people worldwide. He argues that Trump's cavalier 'what do you have to lose' rhetoric ignores serious risks for diabetics and others on common medications. The post emphasizes that hydroxychloroquine has 332 known drug interactions, 59 of them major, making blanket recommendations potentially dangerous. Cornell concludes that promoting this drug without proper safety data, especially for vulnerable populations, borders on malpractice or worse.

Key Insight: During a crisis, even data scientists turn to direct action—sometimes the most useful thing you can do is make masks and share clear instructions.

This post is a practical DIY tutorial for making cloth face masks during the early COVID-19 pandemic. O'Neil provides a step-by-step sewing guide using dishcloths, wire for nose fitting, and elastic cord. She references an existing pattern but aims to simplify the instructions for accessibility. The post includes material recommendations with links, measurements, and photos of each step. It's a departure from her usual algorithmic criticism, reflecting the urgent community response to PPE shortages in spring 2020.

Key Insight: Official pandemic statistics are so compromised by testing limitations and political incentives that back-of-envelope estimates based on hospitalization rates may be more accurate than reported numbers.

O'Neil argues that official COVID-19 statistics are systematically undercounting both cases and deaths due to inadequate testing and inconsistent reporting. She estimates actual US cases are roughly 10x the confirmed count, based on the logic that only the sickest 10% get tested. She notes that death counts are also unreliable, with nursing home deaths excluded in some countries and China's numbers particularly suspect. The post reflects the frustration of a data scientist who fundamentally distrusts the available data while trying to reason through what the true numbers might be.

Key Insight: When official metrics are corrupted by political and systemic failures, we should seek alternative proxies that are harder to manipulate and closer to ground truth.

Cathy O'Neil argues that due to inadequate COVID-19 testing in the U.S., confirmed case counts are unreliable indicators of the pandemic's severity. She proposes using hospital bed utilization as a better proxy for tracking the outbreak's true scale. While acknowledging this metric has limitations—including lag time and confounding factors like postponed elective surgeries—she argues it's more reliable because hospitals track beds accurately and the data is less susceptible to political manipulation. She calls for journalists to collect and compare hospital bed data across national chains.

Key Insight: Our phones are the weak link in pandemic hygiene because they constantly touch our hands and faces, making handwashing alone insufficient.

The post highlights an overlooked hygiene gap during the COVID-19 pandemic: while people focus on handwashing, they ignore that their constantly-touched phones accumulate and redistribute germs. O'Neil walks through a typical daily scenario showing how washing hands becomes futile when we immediately pick up a dirty phone. She notes the irony of discussing pandemic hygiene while touching her face and phone, and acknowledges that solving this problem will require new cleaning procedures that most people haven't considered.

Key Insight: The most difficult people in our lives can become our greatest teachers—not through their wisdom, but through the critical thinking and resilience we develop in response to their failures.

In this eulogy delivered at her father's memorial, Cathy O'Neil reflects on the complicated legacy of her difficult father through ten 'silver linings.' Rather than offering sanitized praise, she honestly catalogues his flaws—intellectual arrogance, sexism, eugenic beliefs, cruelty—while crediting him with inadvertently teaching her valuable lessons. His intimidating nature prepared her to face powerful men like Larry Summers; his wrongheadedness sharpened her critical thinking skills; his cruelty taught her to be a kinder parent. The piece is a masterclass in honoring the dead without dishonoring the truth.

Key Insight: Testing ideas with an engaged, patient community can validate a book's narrative and build author confidence in ways solitary writing cannot.

Cathy O'Neil recounts her week as a 'Theme Speaker' at Star Island, a Unitarian-Universalist retreat off the coast of New Hampshire. She spent six days presenting her forthcoming book on shame to an enthusiastic audience, using Occupy-style progressive stack discussion. The experience, which she humorously describes as 'cult-like' with its communal singing and rituals, proved invaluable for testing her book's narrative. By the end, both she and her son were completely won over by the community's warmth and traditions.

Key Insight: China's deployment of facial recognition in health insurance decisions previews a dystopian future that Americans should actively work to prevent.

This brief post announces a new Bloomberg Opinion column about facial recognition technology being used by Chinese insurance companies. O'Neil was motivated to write after becoming alarmed by developments in China's use of AI for health insurance decisions. The post serves primarily as a pointer to her longer Bloomberg piece, which examines how Chinese insurers are using facial recognition as a cautionary example for American readers. She frames the Chinese case as a preview of what could happen to health insurance in the U.S.

Key Insight: The proliferation of scoring systems and market logic into every domain of life has transformed shame from an informal social mechanism for collective benefit into a privatized, externalized system of quantified inadequacy.

Cathy O'Neil examines shame as a social mechanism that operates in tension with free market ideology. She argues that shame traditionally functioned to get individuals to act against narrow self-interest for collective benefit, but our market-obsessed culture has created new scoring systems (dating apps, college rankings, social media metrics) that generate novel forms of shame. The piece uses the PUA/incel phenomenon as an example of how market logic applied to human relationships produces toxic outcomes. O'Neil suggests we've externalized and privatized traditional shaming mechanisms through technology and scoring systems, creating an overwhelming proliferation of ways to feel inadequate.

Key Insight: Shame operates as invisible power precisely because it enforces silence and inaction, making its effects nearly impossible to measure while systematically controlling marginalized groups.

O'Neil argues that shame functions as an overlooked but powerful form of social control that explains many societal patterns we take for granted. She uses Japanese hikikomori as a case study, noting how shame keeps them isolated and their families silent, while also leading to their criminalization. She contrasts this with how we'd respond to a physical health crisis affecting the same number of people. The post also examines how shame silenced sexual assault victims in the military, with the system itself perpetuating trauma. O'Neil concludes by noting that shame is difficult to measure because it's informal, operates through silence rather than action, and comes with a built-in amnesia that makes us avoid examining it.

Key Insight: Shame operates as a group phenomenon with statistically predictable outcomes, and understanding its collective dynamics—rather than treating it as purely individual—is essential for dismantling exploitative systems that profit from or maintain power through shame.

Cathy O'Neil explores why we need a concept of 'group shame' rather than treating shame as purely an individual experience. She argues that shame is inherently social—it only exists in relation to group norms—yet our language and frameworks focus almost exclusively on individual narratives. Using the #MeToo movement as an example of shifting norms creating new forms of group shame, she begins developing a taxonomy of shame types ('punching down, punching up, punching nowhere') and argues that understanding group shame dynamics is essential for predicting social outcomes and changing exploitative systems.

Key Insight: Shame functions as invisible social infrastructure—systematically deployed by those in power, amplified by big data, and worthy of the same rigorous analysis we apply to algorithms.

Cathy O'Neil announces her second book 'Shame Machine: an owner's manual,' a follow-up to Weapons of Math Destruction with the same editor and publisher. The book will explore shame as a social mechanism—examining who profits from it, who gains power through it, and how it operates in the age of big data. She traces her interest to diverse experiences including insurance battles over bariatric surgery, Twitter interactions, and the value-added model used to score teachers. The book represents a more personal direction than her previous work, which is why she hasn't been testing ideas publicly on her blog. She's now transitioning from planning to active writing, with a tentative January 2021 release date.

Key Insight: This announcement post points to O'Neil's speculative fiction work on human-robot intimacy, showing her range beyond technical algorithm critique.

This is a brief announcement post where Cathy O'Neil shares her essay 'A History of Cyborg Sex, 2018-73' published in Boston Review. The post itself contains no substantive argument, serving only as a pointer to the external publication. O'Neil expresses pride in the piece, which appears to explore the future history of human-robot sexual relationships.

Key Insight: Corporate self-regulation of facial recognition technology will fail because companies have no financial incentive to limit a profitable but invasive technology.

This is a brief announcement post directing readers to Cathy O'Neil's Bloomberg Opinion piece about facial recognition technology. The linked article argues that Amazon cannot adequately fix the problems with facial recognition because companies lack proper incentives to address the technology's 'creepiness' and potential for misuse. O'Neil suggests that market forces alone are insufficient to curb the harms of facial recognition systems.

Key Insight: While the MOOC threat to math education has faded, academic mathematics still faces structural problems in how it treats faculty and credentials contributions.

O'Neil reports from the 2019 Joint Mathematics Meeting where she received the MAA Euler Book Prize and reconnected with the math community. She reflects on the state of academic mathematics, noting she's less worried about MOOCs disrupting teaching than she once was, but remains concerned about the adjunctification of faculty positions. She also critiques the slow pace of change in mathematical publishing and credentialing systems, wondering when collaborative projects like the Stacks Project will become the norm.

Key Insight: The race for AI dominance between nations carries dystopian risks, with China's surveillance state offering a preview of what unchecked AI deployment looks like.

This is a brief announcement post directing readers to O'Neil's Bloomberg Opinion column about international AI competition and surveillance. The column argues that China represents a cautionary tale for AI-enabled dystopia and that the U.S. should oppose surveillance state practices. The post itself contains minimal content beyond the link and headline.

Key Insight: Sometimes a blog post is just a delightful personal observation shared with readers, no deeper analysis required.

This is an extremely brief personal post where the author shares her discovery that karaoke is more enjoyable when performed in French, based on her experience during a trip to Paris. The post is playful and conversational rather than analytical, ending with a humorous challenge to readers to disagree with her assessment.

Key Insight: Voting decisions should prioritize anti-corruption credentials and policy substance over party loyalty, even when that means supporting third-party candidates unlikely to win.

This guest post by Cathy O'Neil's daughter Aise details her voting choices in the 2018 New York elections with extensive reasoning. She voted for NYC ballot proposal #1 (campaign finance reform) but against #2 and #3, which she saw as creating unnecessary bureaucracy and mandating demographic discrimination. For state offices, she supported Green Party candidates for Governor, Attorney General, and Comptroller, prioritizing anti-corruption stances and progressive policies over Democratic establishment candidates she views as compromised by their ties to Governor Cuomo.

Key Insight: The real challenge of constant travel isn't logistics but maintaining genuine human connection and personal worthiness amid endless shallow interactions and displacement from home.

Cathy O'Neil offers a humorous yet philosophical guide to surviving excessive work travel. She argues that business class and airport lounges, while relatively better than coach, are still far worse than being home—a perspective that should temper class resentment between travelers. Beyond practical advice (pack peanut butter crackers, bring less stuff), she explores the psychological toll of constant travel: the shallow interactions, existential angst, and righteous anger that accumulate. Her core prescription is to remain 'open and connected' with strangers, treating each interaction as potentially meaningful, while also working on one's own sense of worthiness.

Key Insight: This post contains no substantive content to analyze—it appears to be a test or placeholder post unrelated to its title.

This is an extremely brief post consisting of only two words: 'I'm animated!!' The post appears to be a test or announcement about some animation feature, possibly related to the blog's technical functionality. There is no substantive argument or analysis present. The title 'The Truth About Algorithms' is entirely disconnected from the actual content, which contains no discussion of algorithms whatsoever.

Key Insight: Tech companies' professed surprise at discovering biased algorithms is a deliberate performance designed to avoid accountability for discrimination they could have anticipated and prevented.

O'Neil announces her Bloomberg Opinion piece arguing that Amazon's sexist recruiting algorithm should surprise no one in the tech industry. She contends that tech companies deliberately maintain 'plausible deniability' about algorithmic bias, feigning surprise when discriminatory systems are exposed. The post frames this as an ongoing pattern where biased algorithms are everywhere but companies actively avoid acknowledging them.

Key Insight: This is a personal announcement about the author's musical side project rather than her typical analytical content about algorithms and society.

This is a brief promotional post announcing an upcoming performance by the TomTown Ramblers band at The Rockwood Music Hall. Cathy O'Neil shares her excitement about playing with her bandmates and promotes ticket sales. She highlights the original songwriting talent of band members Blair Bodine and Jamie Kingston, expressing humble appreciation for being part of such a talented group. The post also mentions merchandise will be available at the show.

Key Insight: China's social credit system isn't simply dystopian surveillance—it's a complex fusion of commercial credit scoring and social control that could either entrench inequality or, if designed thoughtfully, potentially increase social mobility.

This guest post by Jianyin Roachell examines China's emerging National Social Credit System, which will quantify citizens' social worth based on purchasing behavior and conduct by 2020. The author distinguishes between two components: a commercial credit system (similar to FICO) and a social management system that blacklists offenders. Rather than simply condemning the system as Orwellian, Roachell argues that cultural context matters—Confucian values have long emphasized trusting authority. The post raises concerns about how integrating credit scores into dating apps could solidify class divisions and limit social mobility, while also suggesting big data could potentially be leveraged to reduce inequality.

Key Insight: Incremental reforms to financial risk management may be futile if the system itself is fundamentally flawed and requires a complete paradigm shift.

This is a brief announcement post directing readers to Cathy O'Neil's Bloomberg Opinion column arguing that the financial system may be fundamentally unfixable. The column's subtitle suggests that incremental improvements to risk management are insufficient and that a paradigm shift is needed. The post serves primarily as a pointer to her external writing rather than containing substantive analysis itself.

Key Insight: Tech companies' algorithmic bias is a real problem worth addressing, even when the complaints come from bad-faith actors focused on the wrong issues.

This is a brief announcement post linking to Cathy O'Neil's Bloomberg Opinion column about Google and algorithmic bias. The post suggests that Trump's complaints about Google bias contain a kernel of truth, though not in the way he frames it. O'Neil argues that tech companies have a real problem with bias that they need to acknowledge, even if the specific political complaints are misguided.

Key Insight: Reputation scoring systems and government purchasing decisions both represent institutional power that can be wielded for either surveillance or consumer protection.

This is a brief promotional post where Cathy O'Neil shares her new Bloomberg Opinion column arguing that Facebook's reputation scores, while creepy, are probably beneficial. She also promotes a piece by her friend Laura Strausfeld about making tampons free and safer through better testing requirements when government entities become bulk purchasers.

Key Insight: The titans of tech have built platforms with enormous societal impact but lack the wisdom and preparation to responsibly manage them.

This is a brief announcement post directing readers to O'Neil's Bloomberg Opinion column arguing that Mark Zuckerberg and other major tech leaders are unprepared for the societal responsibilities their platforms carry. The post serves primarily as a teaser linking to the full argument published elsewhere. O'Neil characterizes her take as 'snarky,' signaling her critical stance toward Silicon Valley leadership.

Key Insight: America needs homegrown solutions for regulating technology and data rather than importing European frameworks wholesale.

This is a brief announcement post directing readers to Cathy O'Neil's Bloomberg View column about regulating technology and algorithms in the United States. The post indicates that America cannot simply follow Europe's regulatory approach (likely referring to GDPR) but suggests Senator Mark Warner has proposed viable alternatives. The substantive analysis appears in the linked Bloomberg article rather than this blog post itself.

Key Insight: Algorithms increasingly control important aspects of our lives and need systematic auditing for accountability.

This is a brief announcement post where Cathy O'Neil shares that she co-authored an op-ed in the Financial Times with Bart Schermer. The op-ed, titled 'Audit the algorithms that are ruling our lives,' argues for algorithmic accountability and transparency. The post itself contains minimal content beyond the link and acknowledgment that the article is behind a paywall.

Key Insight: Mathematical proofs can serve as empowering experiences that inspire people to pursue careers in mathematics.

This is a brief announcement post directing readers to a Bloomberg View column about the author's path to becoming a mathematician. The post mentions that dominoes and mathematical proofs played a formative role in her decision. The actual content and arguments are contained in the linked Bloomberg article rather than the blog post itself.

Key Insight: Science For The People, a publication examining the intersection of science and politics, has been revived after nearly three decades.

Cathy O'Neil announces an event celebrating the revival of Science For The People, a publication returning after nearly 30 years. The event features a mini-documentary about the organization's revitalization and readings from their new essay collection on geoengineering. This is a brief promotional post directing readers to the NYC event.

Key Insight: Data-driven education reform, even when backed by enormous resources and good intentions, can cause real harm when it treats students and teachers as optimization problems.

This is a brief announcement post directing readers to O'Neil's Bloomberg View column criticizing the Gates Foundation's education reform efforts. The post frames the foundation's big-data approach to improving public schools as not merely unsuccessful but actively harmful. O'Neil signals her core argument through the column's subtitle: that well-intentioned data-driven reforms can cause real damage when implemented poorly.

Key Insight: A technology's usefulness in narrow applications doesn't redeem its broader negative implications for society.

This is a brief promotional post directing readers to O'Neil's Bloomberg View essay criticizing Google Glass. The post expresses clear disdain for both Google Glass and airplane flights, teasing her argument that the device's utility in specific contexts doesn't justify its broader existence. The actual substantive arguments are contained in the linked Bloomberg article rather than this blog post itself.

Key Insight: Acknowledging the inherent unfairness of elite college admissions is a prerequisite for improving the system.

This is a brief announcement post directing readers to O'Neil's Bloomberg View column about the Harvard admissions controversy. The post itself contains no substantive argument, only linking to the external essay with the provocative claim that college admissions will never be fair, but suggests we can build a better system if we acknowledge this reality.

Key Insight: Social media platforms like Facebook have become so influential in elections that they should be subject to the same kind of external observation and scrutiny we apply to traditional electoral systems.

This is a brief post linking to a Bloomberg View essay arguing that Facebook's influence on elections should be studied by scientists. O'Neil proposes treating Facebook like a jurisdiction that needs election observers, suggesting external scrutiny of how the platform affects democratic processes in the 2018 midterms.

Key Insight: Making technical content visually fun and compact removes the intimidation barrier that keeps people from learning, proving that accessibility and depth aren't mutually exclusive.

Julia Evans explains her unconventional approach to teaching programming concepts through self-published zines—short, illustrated booklets traditionally associated with punk and art movements. She argues that zines are an efficient teaching tool because they make intimidating technical content accessible and fun, allowing busy professionals to learn on their commute. The format's low barrier to entry—both for readers who find illustrated content less intimidating and for creators who don't need drawing skills—makes it an effective way to spread programming knowledge.

Key Insight: Algorithms that affect people's lives should be subject to independent quality certification, similar to how we certify food safety.

Cathy O'Neil announces that her algorithmic auditing company ORCAA was featured in Fast Company. The piece highlights ORCAA's 'seal of approval' concept for algorithms, comparing it to organic certification for food. The post is brief, serving mainly to share the coverage and credit the designers who created the certification seal.

Key Insight: Marijuana arrest data serves as a rare window into policing bias precisely because usage rates are equal across races, and losing this metric through legalization means losing one of our few tools for measuring systemic discrimination.

This is a brief announcement post directing readers to O'Neil's Bloomberg Opinion column about the problematic nature of crime data. The column argues that marijuana arrests serve as one of the few reliable measures of racial bias in policing because usage rates are similar across races but arrest rates differ dramatically. O'Neil warns that marijuana legalization, while positive, will eliminate this important barometer for measuring systemic bias in the criminal justice system.

Key Insight: Algorithmic auditing is gaining recognition as an important emerging field, with mainstream tech publications now covering companies that hold algorithms accountable for bias.

Cathy O'Neil announces that her algorithmic auditing company ORCAA was featured in MIT Technology Review's 'Download' section on emerging technology. The brief post celebrates recent media coverage, noting that both the MIT Technology Review piece and a Wired article from the same week have generated business inquiries for ORCAA. O'Neil expresses optimism about the growing interest in algorithmic accountability services.

Key Insight: Algorithmic auditing is emerging as a viable business service that companies can use to demonstrate the fairness of their automated systems.

Cathy O'Neil announces that her algorithmic auditing company ORCAA was featured in Wired magazine. The article by Jessi Hempel covers ORCAA's first customer engagement auditing Rentlogic. This brief post serves as an announcement directing readers to the full Wired piece about algorithmic accountability in business.

Key Insight: Mathematical rigor and probability theory can provide objective, statistically sound methods for detecting partisan gerrymandering in court cases.

This guest post announces a new speaker series on Mathematics and Democracy, covering topics like election theory, redistricting, algorithmic governance, and quantitative fairness. The inaugural talk features CMU professor Wesley Pegden, who served as an expert witness in the Pennsylvania gerrymandering case. Pegden's team developed a probability theorem that provides statistical rigor to methods for detecting gerrymandering. The Pennsylvania Supreme Court used this mathematical evidence to rule the state's congressional districting an unconstitutional partisan gerrymander.

Key Insight: The dominance of a single social network like Facebook raises questions about whether alternative models might better serve society.

This is a brief announcement post where Cathy O'Neil shares her NPR Morning Edition appearance discussing alternatives to Facebook. The post links to the NPR segment exploring what a better social network might look like in light of Facebook's flaws. No substantive argument is made in the post itself, as the content exists in the linked audio segment.

Key Insight: True mathematical understanding, rather than formula memorization, builds transferable thinking skills that benefit students across all areas of life.

Guest author Allison Pacelli, a mathematics professor at Williams College, reflects on her connection with Cathy O'Neil and promotes her PZ Math Calculus Prep Camp. She argues that math is genuinely difficult for everyone eventually, yet society wrongly accepts mathematical illiteracy. Pacelli contends that math education too often focuses on memorization rather than understanding, and that her camp addresses this by teaching the conceptual foundations of calculus. The post emphasizes that mathematical thinking builds transferable skills like logical reasoning, precision, and perseverance that benefit students across all fields.

Key Insight: Congressional oversight of tech companies fails without establishing fundamental data rights for citizens.

This is a brief post directing readers to O'Neil's Bloomberg View article about Congressional hearings on Facebook. The post argues that Congress is failing to address the fundamental issues with Facebook and tech platforms. O'Neil advocates for a 'data bill of rights' as the necessary policy response. The post serves primarily as a pointer to her more detailed argument in Bloomberg.

Key Insight: Ethical guidelines for data scientists are meaningless without enforcement mechanisms that hold practitioners accountable for the societal harm their work can cause.

This is a brief pointer post directing readers to an interview Cathy O'Neil gave to Wired UK about establishing a Hippocratic Oath for data science. The interview explores the ethics of AI and the need for accountability mechanisms in the field. O'Neil suggests that data scientists need ethical guidelines with real enforcement power, not just aspirational principles.

Key Insight: The real problem isn't self-driving car ethics—it's that other algorithmic systems cause harm without the visible accountability that fatal car accidents demand.

This brief post links to O'Neil's Bloomberg View column arguing that the ethics of self-driving cars shouldn't be a primary concern. She suggests that road deaths create strong incentives for accountability that are missing in other algorithmic domains. The post serves as an announcement directing readers to her fuller argument in Bloomberg.

Key Insight: As algorithms increasingly govern our lives, we must first solve the philosophical problem of defining fairness before we can encode it into machines.

This is a brief announcement post directing readers to Cathy O'Neil's Bloomberg View column about algorithmic fairness. The post itself contains no substantive argument, serving purely as a promotional link to her external writing. The referenced piece addresses the challenge of defining and encoding fairness into algorithmic systems. O'Neil suggests that increased reliance on algorithms necessitates grappling with fundamental questions about what fairness means.

Key Insight: There's value in showing up and finishing at your own pace rather than competing, and gratitude for health matters more than performance metrics.

Cathy O'Neil shares a personal update about her running progress, having completed her second 5K race. She improved her time by 5 seconds (from 39:00 to 38:55) and found the race easier than her first attempt in February. She's now training for a sprint triathlon in May, combining running, biking, and swimming. The post is deliberately lighthearted and self-deprecating, celebrating small victories and the joy of simply participating rather than competing.

Key Insight: Technological solutions like risk-assessment algorithms cannot substitute for addressing the deeper structural problems in the bail system.

This is a brief announcement post directing readers to Cathy O'Neil's Bloomberg View column about algorithmic bail reform. The piece argues that Philadelphia should be cautious about implementing risk-assessment algorithms in its bail system. O'Neil suggests that big data solutions alone cannot fix the fundamental problems with bail, continuing her broader critique of algorithmic decision-making in criminal justice.

Key Insight: Secret deployment of predictive policing technology, hidden from democratic oversight, represents a dangerous union of corporate arrogance and government power that threatens civil liberties.

This post shares a satirical poem by Ernie Davis responding to the revelation that Palantir secretly deployed predictive policing technology in New Orleans for six years without public knowledge. The poem draws parallels between Tolkien's palantíri (seeing stones that corrupted their users) and Peter Thiel's Palantir company, suggesting that wielding such powerful surveillance tools without wisdom leads to downfall. It traces Thiel's career from PayPal billionaire to Trump supporter, criticizing his arrogance and anti-intellectual stance. The poem recounts how political consultant James Carville brokered a secret deal to bring Palantir's predictive policing to New Orleans, bypassing city council oversight entirely. Davis argues this secrecy represents a betrayal of public trust and warns of the civil rights implications of algorithmic policing.

Key Insight: Facebook has the power to answer crucial questions about election interference but lacks the motivation to pursue uncomfortable truths.

This brief post announces a Bloomberg View article about Facebook's potential to measure the impact of Russian ads on the 2016 election. O'Neil argues that Facebook has the data and capability to determine whether election meddling made a difference, but questions whether the company has the willingness to conduct such analysis. The post serves primarily as a pointer to the fuller argument in the Bloomberg piece.

Key Insight: Health advice that acknowledges its own ineffectiveness while still prescribing the same solutions reveals more about cultural attitudes toward bodies than about actual health.

Cathy O'Neil critiques a Jane Brody New York Times article about exercise and diets, pointing out its logical contradictions. She argues that Brody acknowledges diets don't work, yet still recommends them, and titles her piece to imply exercise causes weight loss despite evidence to the contrary. O'Neil sees this as a frustrating example of incoherent health advice that shames people for accepting their bodies.

Key Insight: Virtual reality technology could become a dystopian tool for making extreme inequality tolerable rather than a force for human flourishing.

This is a brief announcement post directing readers to Cathy O'Neil's Bloomberg View column about virtual reality and inequality. The linked piece appears to argue that VR technology could be used to pacify poor people by giving them virtual experiences while real-world inequality grows unchecked. The dark framing ('virtual hamster cage') suggests technology as a tool of social control rather than liberation. O'Neil explicitly notes her pessimistic mood when writing the piece.

Key Insight: Even simple crowdsourced systems like Amazon Q&A demonstrate how algorithmic platforms can produce absurd outcomes when human behavior doesn't match the system's assumptions.

This post appears to be a brief, humorous observation about Amazon's customer Q&A feature, likely showcasing an absurd or ironic exchange between customers. The post relies on visual content or embedded examples that aren't captured in the text excerpt. O'Neil seems to be highlighting the comedic dysfunction of crowdsourced product information, where questions and answers often miss each other entirely or produce nonsensical results.

Key Insight: The stock market no longer represents broad economic health but rather tracks the extraction of value from workers to shareholders, making market declines potentially favorable for ordinary workers.

O'Neil argues that most Americans have little reason to care about stock market performance because they lack meaningful market exposure through savings or retirement accounts. She contends that the stock market has decoupled from worker prosperity, with recent evidence showing an inverse relationship where wage increases actually trigger market declines. The wealthy are insulated through diversification while ordinary workers have almost nothing at stake. Recent policy changes like the tax bill have further entrenched this anti-correlation between market gains and worker welfare.

Key Insight: VR's greatest danger isn't that it introduces new problems, but that it dramatically amplifies existing technology harms - surveillance, addiction, and manipulation - that society has already proven unable to confront.

This guest post by Eugene Stern argues that virtual reality poses an amplified threat because it scales up the most troublesome aspects of existing technology: personalization, addictiveness, and marketing manipulation. VR will collect unprecedented amounts of biometric and behavioral data through eye tracking, heart rate monitoring, and brain activity sensors, enabling hyper-targeted manipulation. The immersive nature of VR will make virtual experiences far more addictive than smartphones or social media. Most concerning, VR advertising will achieve unprecedented precision in measuring and predicting consumer behavior. The post concludes that VR's greatest danger is that it represents 'more of the same, but on steroids' - amplifying problems we've already failed to address.

Key Insight: This is a simple announcement post directing readers to an external book review, with no substantive content to analyze.

This is a brief announcement post where Cathy O'Neil shares that she has written her first book review for the New York Times. The review covers two books about Virtual Reality by Jeremy Bailenson and Jaron Lanier. The post itself contains no substantive analysis, serving only as a pointer to the external review.

Key Insight: The expansion of insurance into trivial domains like pizza delivery represents a distortion of insurance's legitimate purpose.

This is a brief announcement post directing readers to O'Neil's Bloomberg View column about pizza insurance. The column argues against the proliferation of trivial insurance schemes that distort the concept of insurance. O'Neil suggests that while insurance serves useful purposes for significant risks, extending it to minor purchases like pizza represents a problematic trend. The post itself serves primarily as a link to the full argument rather than containing substantive analysis.

Key Insight: Personality tests in hiring processes serve as gatekeepers that exclude qualified workers based on flawed criteria.

This is a brief announcement post directing readers to Cathy O'Neil's Bloomberg View article about personality tests in hiring. The full argument about how personality tests fail American workers by filtering out candidates for wrong reasons is contained in the linked Bloomberg piece, not in this blog post itself. The post serves as a pointer to her external journalism work.

Key Insight: The simple faith of a Haitian child serves as a quiet but powerful rebuke to political leaders who demean entire nations.

O'Neil shares a photograph from Haiti of a young boy pointing to the sky, using it as a pointed response to Trump's reported 'shithole countries' comments about Haiti and African nations. The brief post serves as both a tribute to Haiti's dignity and a rebuke of presidential disrespect. She frames the child's gesture as a reminder that moral authority transcends political power.

Key Insight: A medical diagnosis can be strangely liberating when it finally explains inexplicable feelings of anxiety and unrest.

The author shares her recent diagnosis of Graves' Disease, an autoimmune disorder causing thyroid hyperactivity, anxiety, and compulsive exercise. She explains the condition is treatable with medication (Methimazole) or potentially radioactive iodine therapy. The diagnosis actually brought relief by explaining her deep anxiety, sleep troubles, and rapidly changing eyesight. She notes that quitting Twitter has helped her anxiety, and she's channeling her hyperactive energy into training for races. The post ends with a plug for her band's upcoming gig.

Key Insight: True poverty is spiritual, not material, and refusing to amplify cruelty is itself a form of resistance.

O'Neil responds to Trump's reported vulgar comments about Haiti by refusing to repeat them and offering a pointed contrast. She draws on her personal experience traveling in Haiti to make a moral comparison between Haitian people and the president. The brief post uses restraint and dignity to make its critique more powerful than outrage would. Her refusal to perpetuate the comments while still addressing them is itself a rhetorical choice.

Key Insight: The post teases a contrarian take on sex robot discourse, suggesting the technology's social impact may disadvantage men rather than women.

This is a brief promotional post directing readers to O'Neil's Bloomberg View column about sex robots. The linked piece apparently argues that sex robots might create competition for men rather than replace women, flipping the typical narrative about this technology. The post is too short to contain substantive analysis itself, serving primarily as a teaser for her external writing.

Key Insight: Even amid heightened awareness of sexual harassment through movements like #MeToo, we still lack the rigorous data needed to truly understand the scope of the problem.

This is a brief announcement post where Cathy O'Neil shares her new Bloomberg View column about the lack of reliable data on sexual harassment prevalence. She also announces she's quitting Twitter because it was negatively affecting her wellbeing. The post itself contains minimal analysis, serving primarily as a pointer to her external column.

Key Insight: A quick logistical post announcing speaking engagements on data science education.

This is a brief announcement post where Cathy O'Neil promotes two upcoming speaking engagements. She will be giving a talk at Tufts University on Thursday and participating in a data science education roundtable in Washington D.C. on Friday morning. The D.C. event will be livestreamed and is open to in-person attendance.

Key Insight: Grassroots movements like the Women's March gain cultural traction that becomes visible even in unexpected places like yarn store inventory.

O'Neil announces the upcoming 2018 Women's March, noting she discovered it through the telltale sign of pink yarn shortages at craft stores. She urges readers to participate and marks the date while connecting the march's importance to the ongoing #MeToo movement. The post links to revelations about radio host John Hockenberry's systematic harassment and discrimination against women of color colleagues as evidence of why such activism remains necessary.

Key Insight: Algorithmic systems amplify our existing blind spots, making the identification of false negatives—what we fail to see or measure—a matter of life and death.

This brief post announces two items: a new Bloomberg View column about false negatives in algorithmic decision-making for life-and-death situations like autonomous weapons, and an upcoming Occupy Alt Banking book club meeting. The Bloomberg piece argues that algorithms will perpetuate human mistakes unless we understand what data we're missing. The book club will discuss contrasting perspectives on capitalism through Milton Friedman's 'Capitalism and Freedom' and Eric Williams' 'Capitalism and Slavery.'

Key Insight: Algorithms and platforms, whether in hedge fund management or social media, risk amplifying human biases rather than providing objective solutions.

This is a brief announcement post linking to two of O'Neil's Bloomberg View columns. The first examines Ray Dalio's algorithmic management system at Bridgewater, questioning whether it simply reinforces its creator's biases. The second addresses Facebook's fundamental design problems and the challenge of preventing users from sharing biased and misleading content. Both pieces continue O'Neil's ongoing critique of how algorithms and platforms can amplify human biases rather than correct them.

Key Insight: Academia has a responsibility to critically engage with the tech industry and its algorithms rather than remaining detached in the ivory tower.

This is a brief announcement post where Cathy O'Neil shares that she published her first op-ed in the New York Times. The op-ed, titled 'The Ivory Tower Can't Keep Ignoring Tech,' argues that academia needs to engage more seriously with technology and algorithms. The post itself contains minimal content beyond the announcement and link to the full piece.

Key Insight: Weight loss surgery doesn't just change your body—it fundamentally alters your relationship with hunger, comfort, and coping mechanisms, forcing you to find new ways to deal with life's anxieties.

Three months after gastric sleeve surgery, Cathy O'Neil reflects on the physical, dietary, and psychological changes she's experienced. She's dropped several clothing sizes and adopted an all-black wardrobe for simplicity. Her relationship with food has fundamentally changed—hunger no longer feels like a panic alarm, and her taste preferences have shifted toward simpler foods. She's started running again for the first time in twelve years, partly to manage anxiety about parenting, politics, and the state of the world. She candidly discusses the temptation of alcohol and the loss of food as a comfort mechanism, while noting that thin people use food for comfort without stigma.

Key Insight: This is a brief media appearance announcement rather than a substantive post, pointing readers to a radio discussion about AI futures.

This brief post announces Cathy O'Neil's appearance on WBUR's Open Source radio program with Christopher Lydon. The discussion focused on Max Tegmark's book 'Life 3.0' about artificial intelligence. O'Neil notes her personal connection to the show from when she lived in Somerville. The episode featured fellow guests Erik Brynjolfsson and Yarden Katz.

Key Insight: The Equifax data breach sparked public debate, but the conversation focused on the wrong aspects of how personal data is collected and used.

This is a brief announcement post where Cathy O'Neil shares two new essays she published. The first, in Boston Review, examines futurism and how to critically evaluate futurist claims. The second, in Bloomberg View, argues that the Equifax data breach prompted the wrong public conversation about personal data. The post serves primarily as a pointer to her work published elsewhere rather than containing substantive analysis itself.

Key Insight: Algorithms that can infer sensitive personal information like sexual orientation raise urgent questions about data scientists' ethical obligations.

This is a brief post directing readers to O'Neil's Bloomberg View column about the ethical implications of 'gaydar' algorithms. The piece features an interview with Michal Kosinski, the researcher behind a controversial algorithm that claimed to detect sexual orientation from facial images. O'Neil frames this as an example of how algorithms can become 'creepy' and raises questions about data scientists' ethical responsibilities.

Key Insight: The framing of internet users as 'consumers' rather than 'citizens' fundamentally shapes how platforms treat and potentially exploit them.

This is a brief post linking to O'Neil's Bloomberg View column about transparency in digital advertising. She frames the piece as imagining an internet designed to serve citizens rather than consumers, suggesting a fundamental tension between commercial interests and public good in how platforms like Facebook and Google operate. The post itself serves primarily as a pointer to the full argument published elsewhere.

Key Insight: Weight loss is not a triumph of beauty but a complicated negotiation between health, identity, and a society that wrongly equates thinness with worth.

O'Neil reflects on her dramatic weight loss following bariatric surgery, using vivid metaphors (gallons of milk, sponges) to describe the physical transformations her body is undergoing. She shares her theory that fat cells lose weight uniformly, creating increasingly spongy tissue and excess skin. Drawing on a failed diet attempt seven years earlier, she speculates about gut biome changes and their role in sustainable weight loss. Most pointedly, she pushes back against congratulations that equate thinness with beauty, arguing she preferred her fuller body aesthetically and only pursued weight loss for health reasons.

Key Insight: This post contains no original content to analyze—it is merely a pointer to external writing published elsewhere.

This is a brief announcement post directing readers to Cathy O'Neil's Bloomberg View column about the opioid crisis. The post contains no substantive content beyond linking to the external article. It serves purely as a cross-posting notice for her blog audience.

Key Insight: Medical and nutritional guidelines often confuse correlation with causation and design for non-compliance, making them counterproductive for people actually trying to follow them.

O'Neil reflects on her experience relearning how to eat after weight loss surgery, discovering her body now rejects meat but accepts fish and tofu. She describes the difficulty of finding healthy food while traveling in areas dominated by fast food chains, highlighting systemic issues with American food availability. She critiques the overly strict nutritional guidelines from her surgeon's office, which she finds impractical and based on flawed causal reasoning. Ultimately, she chooses to listen to her body's new signals rather than forcing compliance with rules that require eating unappetizing combinations like meat with diet mayonnaise.

Key Insight: The post serves as a gateway to O'Neil's journalism on algorithmic accountability rather than offering standalone analysis.

This is a brief announcement post directing readers to Cathy O'Neil's Bloomberg View column about tools and efforts to combat problematic algorithms. The post itself contains no substantive analysis, serving purely as a pointer to external content. O'Neil uses her characteristic framing of algorithms as 'overlords' to signal the adversarial relationship between automated systems and the public interest.

Key Insight: Countering dangerous ideologies requires proactive education, and the titles of freedom literature themselves form a powerful collective statement when read together.

This guest post by Becky Jaffe presents a curated reading list of Black history and literature as a response to white supremacy, particularly in the context of the 2017 political climate. The author arranges book and documentary titles into a poem format, creating an artistic framework for education. The list spans African and African American literature, slave narratives, civil rights history, and contemporary works. Jaffe explicitly frames this educational approach as resistance, urging readers not to 'give an ideological inch to the white nationalists in the white house.' The post serves as both a practical resource and a call to action through learning.

Key Insight: Society's uncritical trust in big data and algorithms needs to be challenged and replaced with skepticism and accountability.

This is a brief announcement post where Cathy O'Neil shares that her TED talk has been published. The talk, titled 'The era of blind faith in big data must end,' addresses the dangers of uncritical acceptance of algorithmic decision-making. The post itself contains minimal content beyond the link and a personal note hoping readers will enjoy it.

Key Insight: Quantifying health through metrics like weight can undermine actual wellbeing, and recognizing when a measurement tool is harmful gives you the power to reject it entirely.

O'Neil recounts her experience recovering from bariatric surgery and her disastrous decision to start weighing herself weekly despite 23 years of evidence that scales were harmful to her mental health. After creating a predictive weight loss curve and then feeling devastated when she came in one pound over her made-up projection, she spiraled into toxic weight loss chat boards before a news event (Bannon's firing) snapped her out of it. She threw away her scale, recognizing that she had surgery to be healthy, not to lose weight, and that health cannot be measured on a scale. She criticizes the medical establishment's obsession with goal weights and concludes with a philosophy about discarding controllable things that cause harm.

Key Insight: The debate about women in tech becomes unproductive when biological arguments and bias discussions are conflated rather than examined separately.

This brief post promotes O'Neil's Bloomberg View article about the Google diversity memo controversy. She argues that the public debate conflates two separate issues that should be addressed independently: biological differences between genders and workplace bias against women. O'Neil suggests that by decoupling these discussions, we can have a more productive conversation about women in tech and other prestigious, well-paid fields.

Key Insight: Mathematics cannot be separated from its social context and power structures, so meaningful reform requires holding mathematical elites accountable through political and ethical frameworks rather than simply expanding mathematical education.

Michael J. Barany revisits his controversial argument that 'math is everywhere' is a misleading truism that obscures mathematics' historical role as exclusive knowledge wielded by elites. He examines responses to his original Scientific American essay, finding that historians often dismissed the political stakes while mathematicians engaged primarily with implications. Barany argues against the popular view that teaching more people math solves exclusion, instead advocating for political and ethical accountability frameworks that don't require mathematical expertise. He contends that separating math from its practitioners is precisely the error that prevents meaningful reform, since mathematics has always been guarded through intertwined technical and social barriers. The essay concludes that recognizing math's embedded inequality creates ethical imperatives rather than utopian goals of universal access.

Key Insight: This is a link post with no analyzable content; the actual argument exists in the referenced Bloomberg column.

This is a brief announcement post directing readers to Cathy O'Neil's Bloomberg View column titled 'Trump and Statistics Don't Mix.' The post itself contains no substantive argument, serving only as a pointer to external content. It follows a common blogging practice of cross-posting links to published work elsewhere.

Key Insight: Recognizing a destructive feedback loop in your health—where declining ability makes improvement harder—is the moment to intervene, not when you've hit bottom.

Cathy O'Neil provides a positive health update following bariatric surgery, explaining her decision-making process and recovery progress. She describes her changing relationship with food and her goal of returning to cycling. The post reveals that the turning point came when she could no longer bike challenging hills, triggering fears of a destructive feedback loop of declining fitness. She addresses her identity as a 'fit fat person' and her long-term thinking about staying active into old age. The surgery was motivated both by pre-diabetic concerns and a desire to maintain an active lifestyle for decades to come.

Key Insight: The post title alone signals O'Neil's concern that big data will be weaponized against consumers in the health insurance market.

This is a brief announcement post where Cathy O'Neil directs readers to her Bloomberg View column about the threat big data poses to health insurance. The post itself contains no substantive argument, serving only as a pointer to external content. She signals her return to regular blogging after a hiatus.

Key Insight: Medical information sources often sanitize or minimize the reality of procedures because they're written by people who haven't actually experienced them.

Cathy O'Neil shares her recovery experience one week after bariatric sleeve surgery, having just stopped pain medication. She critiques online resources that minimize the surgery's severity by focusing on incision pain rather than the intense internal pain from having most of her stomach removed. Despite moments of regret and panic during recovery, she reports rapid improvement and no lasting regret. The post concludes with her decision to reinvent her wardrobe entirely in goth style as she loses weight.

Key Insight: A quick personal health update sharing good news about a successful surgery and speedy recovery.

This is a brief personal update from Cathy O'Neil announcing that she successfully came through surgery. She reports being mobile already and expresses appreciation for her pain medication. She expects to be discharged the following day.

Key Insight: Even when you follow medical protocols diligently, the protocols themselves may be flawed, requiring patients to become their own advocates and problem-solvers.

Cathy O'Neil shares her experience preparing for bariatric surgery, describing her strict pre-operative liquid diet and an unexpected health scare from sodium deficiency. She discovered that the 'meal replacement' drinks prescribed didn't contain adequate salt, causing weakness and disorientation despite her careful compliance. After researching online forums, she found that pre-op diet protocols vary widely and many patients don't follow them strictly. The post blends personal medical narrative with her characteristic skepticism about supposedly authoritative guidelines.

Key Insight: This post serves as a brief redirect to O'Neil's Guardian article on algorithmic accountability rather than containing original analysis.

This is a brief announcement post directing readers to an external article O'Neil wrote for The Guardian/Observer. The piece was published to coincide with the UK softcover launch of her book 'Weapons of Math Destruction.' The linked article addresses how algorithms can deceive and what can be done about it. The post itself contains no substantive argument, serving purely as a pointer to her journalism work elsewhere.

Key Insight: After six months of navigating insurance requirements, O'Neil shares a personal health milestone, revealing a more vulnerable side outside her usual algorithmic critique.

Cathy O'Neil announces that on her 45th birthday, she received approval from her insurance company for bariatric gastric sleeve surgery after six months of working toward this goal. She expects to have the surgery within two weeks. In preparation, she must begin a strict pre-operative diet of protein shakes and celery only, which means giving up coffee temporarily and carbonated beverages permanently. She expresses excitement about proceeding with this major health decision.

Key Insight: This is a brief event announcement with no substantive analytical content.

This is a brief announcement post about Cathy O'Neil's upcoming trip to London for promotional events tied to the softcover release of her book 'Weapons of Math Destruction' with Penguin UK. She shares a promotional card and invites London-based readers to attend her events.

Key Insight: Medical decisions about one's body can be simultaneously an act of self-love and a pragmatic calculation about future quality of life, rejecting the false binary between body acceptance and health intervention.

Cathy O'Neil explains her decision to undergo bariatric surgery, framing it as a quality-of-life choice rather than an aesthetic one. She openly states she loves her body and dreads the physical changes surgery will bring, but argues that long-term health concerns—particularly diabetes risk and its effects on the brain—outweigh her reservations. She lists practical considerations including her desire to bike in summer heat, family history of arthritis and joint problems, and the reality that doctors treat overweight patients poorly. The post sets up a future series examining the shaming tactics, financial incentives, and bad medical information she's encountered around the decision.

Key Insight: The promise that makes drastic surgical intervention worthwhile is freedom from constant, overwhelming hunger—a transformation that those who haven't experienced it might not fully appreciate.

O'Neil provides a personal update on her progress toward bariatric surgery, having completed all paperwork and awaiting insurance approval. She explains the pre-surgery requirements including a two-week protein-shake-only diet designed to shrink her fatty liver for the laparoscopic procedure. The post details the four stages of post-surgery dietary recovery, from liquid-only intake to eventually eating normal but much smaller portions. She acknowledges the intimidating behavioral changes required but expresses hope based on others' experiences that reduced hunger will make the transition manageable.

Key Insight: Hotels' anonymous, indestructible blandness is actually a feature that frees guests from the anxiety of being a good houseguest.

O'Neil shares lighthearted observations about hotel stays, touching on the excessive mirrors (particularly near toilets), the inadequate coffee pod allotments, and paradoxically, her love of hotels' anonymous blandness. The post is an informal, humorous departure from her usual analytical content, written while traveling with her child. She offers self-deprecating solutions to each complaint, like walking around without glasses to avoid mirror reflections and stealing coffee pods from maid carts. The piece celebrates hotels' durability and lack of preciousness as features rather than bugs.

Key Insight: This post is a lighthearted family travel diary without connection to the author's usual themes of algorithms, data science, or social critique.

This is a personal travel diary documenting a mother-son trip to Dublin, Ireland. The post chronicles their experiences over four days, including dealing with travel frustrations, staying at a luxurious castle hotel, horseback riding lessons, visiting the Cliffs of Moher, and stumbling upon Dublin's Pride Parade. The narrative emphasizes leisure, bonding activities like swimming and chess, and embracing Irish culture through food, music, and spontaneous encounters with street musicians.

Key Insight: Sometimes the best adventures come from finally saying yes to a child's persistent question instead of endlessly deferring.

Cathy O'Neil shares the charming story of how a spontaneous St. Patrick's Day celebration with her son Wolfie led to an actual trip to Ireland. What began as dressing up for a parade evolved into a shared obsession through flag projects and YouTube castle videos. After her son repeatedly asked when they could visit, she finally bought tickets on impulse. The post is a lighthearted departure from her usual analytical content, focusing purely on family adventure.

Key Insight: A simple publishing error in a school rating algorithm drove families away from a quality school for years, demonstrating how algorithmic systems can inflict real damage on communities even when they're just plain wrong.

A data scientist father discovered that his children's elementary school had been incorrectly rated by GreatSchools.org for years due to a publishing error. Despite the low rating, his family's personal experience was positive, yet he watched neighbors avoid or leave the school based solely on the single-digit score. After persistent emails, GreatSchools acknowledged their error and corrected the rating from 5 to 8. The post illustrates how algorithmic ratings can create self-fulfilling prophecies that damage communities, even when the underlying data is simply wrong.

Key Insight: This announcement post contains no analyzable argument; the substantive content exists only in the linked Bloomberg column.

This is a brief announcement post directing readers to Cathy O'Neil's Bloomberg View column about the problems with technology controlling education. The post itself contains no substantive argument, serving only as a pointer to the full article published on Bloomberg. O'Neil uses her blog to cross-promote her mainstream media writing.

Key Insight: The myth that mathematicians must be geniuses is harmful and inaccurate, suggesting mathematical ability is more accessible than commonly believed.

This is a brief announcement post directing readers to Cathy O'Neil's Bloomberg View column about the myth of genius in mathematics. The post itself contains minimal content beyond the link and title. The referenced column appears to argue that mathematicians are not all geniuses, challenging common perceptions about mathematical talent. The post serves primarily as a pointer to her external writing rather than containing substantive analysis itself.

Key Insight: A lighthearted call to action for the mathematical community to show support for the Stacks Project through merchandise.

This is a brief promotional announcement for Stacks Project hoodies being sold to support a mathematical research initiative. The post directs readers to sign up by June 16th to receive hoodies in time for the Stacks Project Workshop in Michigan. It's a community-building effort for mathematicians interested in algebraic geometry.

Key Insight: Tech leaders would rather escape to Mars than be held accountable for the problems their platforms create on Earth.

This brief post serves as a pointer to O'Neil's Bloomberg View column criticizing Silicon Valley entrepreneurs. The core argument, captured in the subtitle, is that tech leaders prioritize exciting, ambitious projects like space travel over addressing accountability and societal problems their technologies create. O'Neil frames this as a fundamental misalignment between tech industry incentives and public interest.

Key Insight: This post is merely an announcement linking to external content and contains no analyzable argument or insight.

This is a brief announcement post directing readers to Cathy O'Neil's Bloomberg View column about algorithmic hiring at Fox News. The post itself contains no substantive analysis, serving purely as a link to external content. The actual argument and insights would be found in the linked Bloomberg article, not in this blog post.

Key Insight: Personal activism can take creative forms, from legal action to crafting protest gear for causes you believe in.

Cathy O'Neil shares a knitting project she created to support period equity activism - a tampon-themed hat designed for her friend Laura Strausfeld, who was speaking at a rally for safe feminine care products. The post briefly mentions O'Neil's personal involvement in the issue as a plaintiff in the successful New York State tampon tax case. She learned a new knitting technique called 'intarsia in the round' to create the hat and plans to share the pattern on Ravelry.

Key Insight: A bell curve distribution doesn't validate a measurement's meaning—the Central Limit Theorem guarantees that even complete nonsense will produce one.

This guest post by Eugene Stern eviscerates a New York Times profile that praised William Sanders' value-added models for teacher evaluation. The article's central criticism is that the Times treated the bell curve distribution of teacher ratings as evidence of the model's validity, when in fact the Central Limit Theorem guarantees that almost any measurement—including random noise—will produce a bell curve. Stern argues this is mathematical illiteracy dressed up as scientific credibility, pointing out that a completely meaningless metric would produce the same impressive-looking distribution.

Key Insight: The post merely promotes external content comparing Trump's mental processes to algorithmic A/B testing, with no substantive analysis in the post itself.

This is a brief blog post that links to O'Neil's Bloomberg View column comparing Trump's cognitive style to Google's ad testing methodology. The post itself contains no substantive analysis, serving only as a pointer to external content. The intriguing title suggests the column explores how Trump's thinking resembles path-independent algorithmic processes, where outcomes don't depend on the sequence of inputs.

Key Insight: When data becomes a tool for political manipulation rather than objective measurement, democratic decision-making is undermined.

This is a brief announcement post directing readers to Cathy O'Neil's Bloomberg column about how politically motivated manipulation of data poses a threat to democratic processes. The post itself contains no substantive argument, serving only as a pointer to the external article. O'Neil uses her blog to cross-promote her mainstream media writing on data integrity issues.

Key Insight: Creative projects with tangible, repetitive work can serve as powerful coping mechanisms during periods of political anxiety and emotional turmoil.

Cathy O'Neil shares her massive knitting project completed as a coping mechanism for post-election anxiety. She bought yarn the day after the 2016 election and finished the 144-square blanket exactly 100 days after the inauguration. The project served as a way to channel nervous energy and manage overwhelming emotions during a politically turbulent period. She also knitted pussy hats and other projects during the same timeframe.

Key Insight: Algorithmic teacher evaluation models are bad algorithms that shouldn't be used for high-stakes grading decisions.

This brief post announces O'Neil's Bloomberg View column arguing against using Value-Added Models (VAM) to evaluate teachers. She signals this is among her most important work on the topic, suggesting the algorithmic approach to teacher grading may finally be losing credibility. The post serves primarily as a pointer to her longer argument about why these statistical models are fundamentally flawed for high-stakes personnel decisions.

Key Insight: O'Neil dismisses transhumanist predictions about human-robot merger as hype that shouldn't be taken at face value.

This is a brief post linking to O'Neil's Bloomberg View column where she critiques Yuval Harari's ideas about human-robot merger. The post signals skepticism toward transhumanist narratives, suggesting Harari's predictions are not to be taken seriously. O'Neil frames herself as unconvinced by the techno-utopian vision of humans merging with machines.

Key Insight: Tech's meritocracy myth obscures a reality where success depends on networks that women cannot equally access due to the industry's informal culture that blurs professional and personal boundaries.

This anonymous guest post argues that tech's informal culture, which blurs professional and social boundaries, creates unique barriers for women seeking mentorship. The author recounts her experience with a VIP who ignored her prepared questions, talked at her for an hour, then pursued her with unwanted emails and texts. She notes that even well-intentioned senior men may avoid mentoring women to prevent appearances of impropriety, while senior women are too few to fill the gap. The post concludes that tech's emphasis on networking and connections, combined with these mentorship barriers, will continue to stall diversity in the industry.

Key Insight: This post contains no analyzable content beyond a link to an external column about applying Simpson's Paradox to Facebook's gender data.

This is a brief pointer post directing readers to the author's Bloomberg View column about female engineers at Facebook and Simpson's Paradox. The post itself contains no substantive argument, serving only as a cross-posting notice to drive traffic to the full article on Bloomberg. The actual analysis of gender bias at Facebook and the statistical phenomenon of Simpson's Paradox is contained in the linked external piece.

Key Insight: Data infrastructure and analytics can create police accountability independent of federal support, but only if technically skilled people choose to apply their abilities to justice work.

Phil Goff argues that despite the federal government's retreat from police accountability under the Trump administration, data science offers a path forward for civil rights protection. The Center for Policing Equity has built the National Justice Database, the first standardized database of police behavior, with voluntary participation from departments covering one-third of the US population. Google's $5 million commitment to automate data processing will dramatically accelerate analysis that once took months. Goff calls on data scientists to become 'JusticeNerds' and apply their skills to questions of policing equity, arguing that metrics-based accountability can proceed regardless of federal support.

Key Insight: This post is a brief redirect to O'Neil's Bloomberg column on AI risks rather than containing original analysis.

This is a brief placeholder post directing readers to Cathy O'Neil's Bloomberg View column about the robot apocalypse. She also mentions an upcoming recap of her TED experience. The actual substantive content lives in the linked Bloomberg article rather than this blog post.

Key Insight: This link post offers no original analysis but signals O'Neil's interest in examining how corporate behavior and technology intersect with human dignity.

This is a brief link post directing readers to two Bloomberg View articles by Cathy O'Neil. The first article discusses the United Airlines passenger removal incident and what it reveals about societal conceptions of dignity. The second is a philosophical essay exploring hypothetical brain-computer interfaces for instant knowledge transfer. The post itself contains no substantive analysis, serving purely as a pointer to external content.

Key Insight: Claims about capitalism's benefits deserve rigorous empirical scrutiny rather than ideological acceptance.

This is a brief meta-post announcing Cathy O'Neil's Bloomberg View debate with economist Noah Smith about whether capitalism reduces inequality. O'Neil directly challenges Smith's earlier column claiming capitalism shrinks inequality, stating she believes the evidence contradicts his position. The post serves primarily as a pointer to the full debate published on Bloomberg View rather than presenting the full argument itself.

Key Insight: Small acts of resistance, like postcard campaigns, can accumulate into meaningful political pressure.

O'Neil celebrates the news that Steve Bannon was demoted from the National Security Council, attributing it partly to the viral #PostcardsToBannon campaign she had previously promoted. She views this as a small but meaningful victory for resistance efforts. The post also briefly introduces her Bloomberg column arguing against using big data for jury selection.

Key Insight: Science advocates are finding creative ways to merge crafting culture with political resistance, creating accessible forms of protest.

This brief post celebrates the intersection of knitting, science activism, and political resistance ahead of the March for Science. O'Neil highlights 'resistor hats' - knitted hats featuring the electronic resistor symbol - as a creative form of protest. She showcases variations including chemistry-themed, DNA helix, water molecule, and combination pussy hat/resistor designs, all available as free patterns on Ravelry. The post is essentially a celebration of nerdy craftivism.

Key Insight: When you can't hide from surveillance, you can still resist it by drowning your signal in noise.

Following Congress's vote to allow ISPs to sell browsing data without consent, Angela Grammatas proposes a counterintuitive strategy: instead of hiding from data collection, flood it with meaningless noise. She illustrates how targeted advertising can manipulate voters through a hypothetical 'Jane Internet' who gets selectively shown political ads based on her cat website visits. The solution is Noiszy, a browser plugin that generates fake browsing data to confuse targeting algorithms. If adopted widely, this approach could weaken the correlations that make targeted advertising effective, potentially forcing companies to earn user engagement rather than exploit it.

Key Insight: This is a non-substantive personal announcement post with no analytical content to extract.

This is a brief announcement post about an upcoming performance by the TomTown Ramblers band on Sunday. The post contains no substantive content beyond the invitation to attend. It appears to be a personal announcement rather than analytical content typical of the blog.

Key Insight: This is a link-only post with no analyzable content beyond signaling the topic of algorithmic governance.

This is a brief announcement post directing readers to Cathy O'Neil's Bloomberg View column about managing algorithmic systems. The post itself contains no substantive content beyond the link and title. The actual argument and analysis exist in the external Bloomberg piece, not in this blog entry.

Key Insight: When a mathematically perfect solution exists (globes), framing an educational debate as a choice between two imperfect alternatives (map projections) obscures objective truth and risks teaching relativism where none is warranted.

Ernie Davis critiques the media coverage of Boston schools' switch from Mercator to Gall-Peters maps, arguing that framing either projection as 'more accurate' is misleading. He explains that all flat maps involve tradeoffs: Gall-Peters preserves area but distorts shapes, while Mercator preserves shapes but distorts area. The mathematically correct solution is simple: use globes, which are now inexpensive and provide perfect scale models with no distortion. Davis worries that presenting map projections as competing theories based on political preference obscures the actual geographical truth that globes perfectly represent.

Key Insight: Algorithms cannot reliably automate decisions that humans themselves cannot clearly articulate, and treating human welfare as an optimization externality has profound ethical costs that organizations must explicitly account for.

Arthur Doskow offers a systematic critique of algorithmic decision-making systems, arguing that organizations often adopt these tools without adequately questioning whether quantitative methods can reliably make the human judgments they're meant to replace. He emphasizes that algorithmic systems merely substitute individual human bias for consistent algorithmic bias, and whether this represents improvement must be assessed case by case. The post advocates for rigorous testing, meaningful accountability (including CEO sign-off on system claims), reformed software contracts that hold developers liable, and recognition that 'being humane may cost money.' Doskow concludes by identifying the core tension: optimization techniques designed for inanimate problems are now being applied to life-altering human decisions with little consideration for human consequences.

Key Insight: This link post highlights O'Neil's ongoing Bloomberg coverage of algorithmic exploitation and corporate data practices.

This is a brief link post directing readers to two of O'Neil's Bloomberg columns. The first examines the dystopian implications of algorithmic price discrimination, while the second analyzes Snap's business model and its need to deeply understand user psychology. The post itself contains no substantive analysis, serving purely as a pointer to external content.

Key Insight: Blaming individuals for obesity is intellectually dishonest when the population-wide trend clearly points to systemic causes we haven't yet identified.

Cathy O'Neil argues that the obesity epidemic has no single known cause, despite the many explanations commonly offered. She systematically debunks the 'blame the individual' approach by noting that exercise has increased yet obesity has risen, and that genetics can't explain rapid population-wide changes. While listing eleven potential contributing factors—from sugar consumption to gut bacteria to glyphosates—she emphasizes that none alone explains the trend. Her core argument is that intellectual honesty requires admitting uncertainty rather than scapegoating individuals, especially children.

Key Insight: Individual success stories about weight loss are statistically meaningless noise that obscure the population-level reality: we have no intervention that reliably helps most people lose weight.

O'Neil uses dieting as a case study to demonstrate proper statistical thinking, showing a hierarchy from useless anecdotal evidence to meaningful population-level studies. She argues that personal success stories about dieting are statistically meaningless when 97% of people in rigorous studies fail to lose weight. The post also dismantles the 'willpower' theory of weight loss by using herself as a counterexample—someone with extraordinary discipline who remains heavy. She suggests that her early dieting attempts, enabled by willpower, may have actually caused metabolic damage that made her heavier long-term.

Key Insight: When nearly everyone fails at something despite genuine effort, we should question whether 'choice' is the right framework for understanding that failure.

O'Neil argues that while diets statistically fail for the vast majority of people, we persist in treating diet failure as a personal choice rather than a near-universal outcome. She distinguishes between the technical freedom to eat (having access to food) and meaningful choice, noting that people who invest significant effort and money into dieting still fail at rates around 99%. The post challenges the common response of successful dieters offering advice, pointing out this ignores the statistical reality. She compares the 'choice' to go off a diet to the 'choice' to breathe—technically possible to resist, but only briefly.

Key Insight: Personal experiences like medical decisions can become powerful lenses for examining systemic problems in how science, data, and evidence function in society.

Cathy O'Neil announces two significant personal developments: she will give a TED talk despite previously criticizing the format, and she plans to undergo bariatric surgery. She acknowledges her change of heart on TED, joking that things can become 'so douchey they come out the other side and are super cool.' More substantially, she commits to documenting her surgery journey on her blog, using it as a lens to explore broader themes about science, evidence, bias, and personal identity.

Key Insight: This linking post suggests O'Neil's Bloomberg column explores how data systems can be weaponized against immigrant populations, but the blog post itself contains no extractable analysis.

This is a brief linking post directing readers to the author's Bloomberg View column about how data systems can criminalize immigrants. The post itself contains no substantive content beyond the headline and link. The actual analysis and arguments are contained in the external Bloomberg article, which is not included here.

Key Insight: This post is a link announcement only; the actual content resides on Bloomberg's website.

This is a brief announcement post linking to the author's Bloomberg View column titled 'Bigger Data Isn't Always Better Data'. The post contains no substantive content beyond the link itself.

Key Insight: When insurance companies gain access to granular health data from wearable devices, the power asymmetry allows them to use that information against consumers rather than for their benefit.

This is a brief blog post linking to a Bloomberg View column about the risks of health tracking devices like FitBits when combined with insurance company data practices. The post's title argues that insurance and big data are fundamentally incompatible, suggesting that the data collected by wearable devices could be used against consumers rather than to help them. O'Neil frames the 'free' health tracker as potentially costly, implying that the trade-off between convenience and privacy/fair treatment may not favor consumers.

Key Insight: Combating big data's harmful effects requires both intellectual critique and practical tools that empower individuals to reclaim their data from surveillance companies.

This brief link roundup post shares three pieces of O'Neil's recent work and collaborations on big data's societal impacts. She highlights her Bloomberg column examining how big data and search algorithms are distorting historical understanding, particularly around Holocaust information. She also points to a Vice collaboration on making technology serve human interests, and promotes a practical guide for requesting personal data from Cambridge Analytica.

Key Insight: The title implies a critique of how tech platforms like Facebook are acquiring quasi-regulatory power over financial systems without democratic accountability.

This post links to a Bloomberg View column by Cathy O'Neil. The title suggests the article examines how Facebook is taking on a regulatory role in financial services, likely critiquing the platform's expanding influence over financial decisions without traditional oversight. The post itself contains no substantive content beyond the link.

Key Insight: An invitation to a public conversation exploring how algorithms intersect with democracy and journalism.

This is a brief event announcement for a panel discussion at NYU Journalism School about algorithms, data, and democracy. Cathy O'Neil will be speaking alongside ProPublica journalist Julia Angwin, moderated by Robert Lee Hotz. The event offers both in-person attendance and livestream options.

Key Insight: Exploiting Trump's ego and need for attention may be more effective than direct confrontation in undermining his advisors' influence.

O'Neil promotes a viral Twitter campaign encouraging people to send postcards addressed to 'President Bannon' as a strategy to undermine Steve Bannon's influence over Trump. The tactic exploits Trump's ego by making him jealous of attention given to his advisor. The campaign gained significant traction on Twitter and was covered by USA Today. O'Neil frames it as a small but emotionally satisfying act of resistance.

Key Insight: This is a redirect post with no analyzable content beyond its provocative title linking Trump to the concept of singularity.

This is a brief announcement post directing readers to Cathy O'Neil's Bloomberg column where she argues that Donald Trump represents a kind of political singularity. The post itself contains no substantive argument, serving only as a pointer to the external piece. The actual analysis and arguments are contained in the linked Bloomberg article rather than this blog entry.

Key Insight: In times of political crisis, concrete action through supporting established advocacy organizations is more valuable than abstract discussion.

This guest post by Becky Jaffe provides a curated list of progressive organizations to support in the wake of political changes. The list covers immigrant and refugee rights, civil rights and legal defense, LGBTQ rights, disability rights, democracy-building, and environmental causes. It's presented as a starting point for people wanting to connect with community organizations that advocate for marginalized groups. The author promises a Bay Area-specific follow-up and encourages readers to create similar regional lists.

Key Insight: O'Neil appears to view Cambridge Analytica's much-hyped data operation as more ordinary than revolutionary, suggesting the 'secret sauce' is overhyped marketing rather than breakthrough technology.

This is a brief link post directing readers to Cathy O'Neil's Bloomberg column about Cambridge Analytica. The post references a Vice Motherboard article about Cambridge Analytica's role in Brexit and Trump's election, and links to her Bloomberg response titled 'Trump's Secret Sauce Is Just More Ketchup.' The dismissive title suggests O'Neil is skeptical of claims about Cambridge Analytica's revolutionary data capabilities.

Key Insight: When government databases designed to help vulnerable populations become potential surveillance tools, one form of resistance is for non-vulnerable people to flood those databases with their own information, creating protective noise.

This guest post urges New York City residents to apply for the ID NYC municipal identification card as a form of resistance against potential federal immigration enforcement. The program, established in 2014, allows residents to obtain photo ID without standard government documents, making it particularly useful for undocumented immigrants. A recent court ruling required the city to retain application records indefinitely, creating a potential database that could be subpoenaed by federal authorities. The author argues that if millions of additional New Yorkers apply, it would 'gum up the works' and make any fishing expeditions through the records far more difficult. The post provides practical instructions for applying and frames this bureaucratic act as accessible civic resistance.

Key Insight: Legal victories against unjust policies mean little if enforcement agencies refuse to comply with court orders.

This brief post documents the author's participation in the JFK Airport protest against Trump's immigrant ban in January 2017. O'Neil shares footage and photos from the protest while highlighting how information about additional demonstrations spread through crowd chants. She also raises concerns that despite a judicial stay on the Executive Order, there are reports of border patrol agents not complying with court orders.

Key Insight: Tech insider perspectives on data often fail to address fundamental questions of access and equality.

This brief post announces two new publications by Cathy O'Neil: a Bloomberg column titled 'When Algorithms Come for Our Children' and a book review for Science Magazine of Andreas Weigend's 'Data for the People.' The post serves primarily as a pointer to these external pieces rather than containing substantive analysis itself. The book review title suggests O'Neil found Weigend's perspective too narrowly focused on tech-insider viewpoints while overlooking equity concerns.

Key Insight: AI systems are not immune to fake news—they inherit the same vulnerabilities as the humans whose data trains them.

This is a brief announcement post where Cathy O'Neil shares that she has become a Bloomberg View columnist. She links to her first column about how fake news that deceives humans will also deceive AI systems. The post serves primarily as a notification to her blog readers about this new platform for her writing.

Key Insight: Craft communities like knitters are natural activists because their culture already values generosity, community, and creating things with purpose.

O'Neil documents her experience knitting a pussyhat during Trump's inauguration and discovering that the knitting community had organically mobilized for political activism. She describes seeing pink yarn and pussyhats everywhere—at Vogue Knitting Live, in yarn stores, on planes heading to D.C., and proliferating on Ravelry. The post celebrates how craft communities can become unexpected sites of political resistance, channeling creative energy into collective action.

Key Insight: This post contains no content to analyze - it appears to be an empty placeholder or failed archive.

This appears to be an empty or placeholder post with only a title 'This might be interesting' and no actual content. Without any text to analyze, it's impossible to determine the author's argument, thesis, or conclusions. The post may have been a draft that was accidentally published, or the content may have been removed or failed to load during archival.

Key Insight: Algorithmic fairness requires choosing which type of unfairness to accept, and in criminal justice, we should prioritize not wrongly punishing people over not wrongly releasing them.

O'Neil continues her exploration of fairness in algorithmic scoring systems, demonstrating that while you cannot simultaneously achieve calibration, equal false positive rates, and equal false negative rates across groups with different base rates, you can achieve any two of these three criteria. Using a visual example with simplified groups, she shows how adjusting for equal false positive rates necessarily creates unequal false negative rates when the model remains well-calibrated. She argues that in recidivism scoring, prioritizing equal false positive rates may be the better tradeoff since the harm of incorrectly imprisoning someone longer is more severe than letting someone off earlier.

Key Insight: Even mathematically 'well-calibrated' algorithms can perpetuate discrimination if the data they're trained on reflects biased systems like over-policing of minority communities.

O'Neil revisits two previous posts with important clarifications. On algorithmic pricing, she concedes that automated price changes aren't necessarily harmful to consumers beyond occasional 'flash crash' disruptions. On recidivism algorithms, she acknowledges a more nuanced research paper that distinguishes between crude discrimination definitions and 'well-calibrated' models that account for different base rates. However, she maintains that even well-calibrated models are problematic if the underlying base rate differences stem from biased policing practices rather than actual behavioral differences.

Key Insight: Algorithms don't need to conspire to collude - market transparency itself can destroy price competition when machines react faster than consumers can benefit from discounts.

The post examines how algorithms can facilitate anti-competitive price-fixing, both intentionally and unintentionally. It highlights a case of deliberate algorithmic collusion among poster sellers, then introduces a more troubling scenario: a German petrol price-tracking app that inadvertently discourages price cuts by enabling instant competitor matching. The post warns that current legal frameworks are inadequate to address algorithmic collusion, especially when AI systems 'self-learn' to coordinate prices without explicit human instruction.

Key Insight: Rather than abandoning flawed recidivism algorithms, we should fix them using bias-reduction techniques while recognizing that anti-discrimination laws ironically limit our ability to correct for the very discrimination they aim to prevent.

O'Neil argues that despite their discriminatory nature, recidivism risk models should be fixed rather than abandoned, since the current justice system is already fundamentally unfair. She outlines three technical approaches from academic research—massaging, reweighing, and sampling training data—that can reduce algorithmic bias while complying with anti-discrimination laws. She notes the irony that these laws, which forbid considering race in decisions, actually constrain our ability to correct for discrimination. The post emphasizes that fixing these models requires broader societal conversations about justice and sentencing goals.

Key Insight: Optimizing recidivism algorithms for accuracy while ignoring fairness guarantees discrimination because the historical data itself encodes systemic bias.

Cathy O'Neil argues that recidivism risk algorithms are inherently discriminatory due to a fundamental mathematical trade-off between accuracy and fairness. Citing a 2011 academic paper by Kamiran and Calders, she explains that when training data reflects historical bias against certain groups, optimizing for accuracy necessarily perpetuates that discrimination. The paper demonstrates this trade-off is linear or sub-linear—reducing discrimination costs only a fraction of accuracy. O'Neil traces the data bias to broken windows policing and systemic racism in the justice system, referencing Michelle Alexander's work on mass incarceration.

Key Insight: Stop wasting energy on individualistic self-improvement guilt and redirect it toward collective action and political resistance.

Cathy O'Neil rejects traditional New Year's resolutions as individualistic, self-defeating exercises that distract from more important work. She argues that conventional resolutions make us focus on personal imperfections when we should be thinking bigger about societal issues. Instead, she offers her own unconventional list that prioritizes meaningful work, family, political engagement, and resistance over self-improvement. The post is a call to stop obsessing over personal flaws and start doing 'heavy lifting' on what matters. Her resolutions range from serious (categorizing algorithmic fairness research for activists, forming a plan to revive the Left) to deliberately provocative (gain 10 pounds, let the kitchen stay messy).

Key Insight: The convergence of speech and facial manipulation technologies will soon enable synthetic video content so convincing that it will fundamentally undermine our ability to trust video evidence.

O'Neil warns about two emerging technologies that will dramatically worsen the fake news problem: software that can edit recorded speech and technology that can manipulate facial expressions in video. She argues that combining these technologies will enable the creation of entirely fictitious but believable video conversations between public figures like Trump and Putin, making our current fake news challenges seem minor by comparison.

Key Insight: Platforms are not neutral conduits but profit-driven machines whose designs actively shape content spread, making their legal immunity from that content increasingly untenable.

Cathy O'Neil argues that Section 230 of the Communications Decency Act, which protects internet platforms from liability for user-generated content, is outdated and inadequate for addressing modern problems like fake news, online abuse, and radicalization. She challenges the premise that platforms are merely innocent bystanders, noting that Facebook's and Google's business models directly benefit from the spread of false and misleading information. While acknowledging she lacks a specific legal alternative, O'Neil contends that holding platforms at least partly responsible for harmful content is necessary from a public interest perspective.

Key Insight: Algorithmic ethics isn't about whether to make moral tradeoffs—it's about making them explicitly rather than hiding them in code.

O'Neil argues that building ethical algorithms requires explicitly quantifying and weighing the relative harms of different types of errors. She illustrates this through three examples: child abuse prediction (removing children unnecessarily vs. missing abuse), recidivism scoring (releasing someone who reoffends vs. jailing someone who wouldn't), and unemployment fraud detection (false accusations vs. missed fraud). She contends that without deciding on these moral tradeoffs explicitly and building them into penalty functions, we cannot claim to have created moral algorithms. The difficult conversation about relative harms must happen before optimization can proceed ethically.

Key Insight: As algorithmic accountability becomes mainstream, critics like O'Neil are shifting from blogs to major media platforms where their arguments about Facebook and FBI surveillance reach broader audiences.

This brief meta-post announces several recent publications by the author across different platforms. O'Neil explains her reduced blogging frequency is due to writing commitments elsewhere, then provides links to four pieces: an article arguing Facebook's algorithm threatens democracy, a co-authored letter critiquing FBI machine learning proposals, responses to her book from Data & Society, and a Quartz interview. The post functions primarily as a directory to her work published in more prominent venues.

Key Insight: Disconnecting from social media can free up time and mental space for simple pleasures.

This is a brief personal post where Cathy O'Neil shares her morning music playlist after quitting Facebook. She notes that leaving the platform has freed up time for activities like listening to music. The post features three songs: a folk ballad from the Llewyn Davis soundtrack, a Bruno Mars song her son enjoys, and a Bright Eyes track she loves.

Key Insight: The most effective way to break a compulsive digital habit may not be willpower but rather embracing any friction that interrupts the unconscious loop.

O'Neil recounts how she accidentally broke her Facebook addiction when the platform required a password reset and she simply refused to comply. She describes her previous behavior as unconscious scrolling through a 'fetid swamp' of partisan content and petitions, unable to control her usage. The password reset requirement created an unexpected barrier that gradually extinguished her habit. She reports her life is better without the platform, though she occasionally wonders about friends still trapped in the same cycle.

Key Insight: Being called 'insightful and disturbing' is exactly the reaction a book critiquing algorithmic harm should provoke.

Cathy O'Neil announces that her book 'Weapons of Math Destruction' was reviewed in the New York Review of Books alongside 'Virtual Competition' by Ezrachi and Stucke. The review, written by Sue Halpern and titled 'They Have, Right Now, Another You,' describes O'Neil's book as 'insightful and disturbing.' O'Neil expresses satisfaction with this characterization of her work.

Key Insight: Data scientists have both the unique capability and potentially the moral obligation to apply their skills toward social justice causes, especially in times of rising bigotry and threats to vulnerable communities.

This guest post announces a series of community discussions about using data science for social good. Written by Roger Stanev and Chris French, it calls on data scientists and IT professionals to consider how their skills can support non-profit organizations working on social justice issues. The post frames the conversation in the context of rising fear, hatred, and bigotry following the 2016 election. It poses several questions about how technical expertise can address problems facing minority and vulnerable communities, and invites readers to attend meetings in the Seattle/Tacoma area.

Key Insight: Arrest statistics don't measure crime rates—they measure the intersection of crime and biased enforcement, and we need methods to disentangle the two.

O'Neil questions who actually gets arrested for possessing box cutters in New York, suspecting it's not a random sample but disproportionately affects minority men through stop-and-frisk practices. She uses the case of a Latino construction worker jailed for 10 days for carrying a job-required tool to illustrate how seemingly neutral laws can be selectively enforced. The post calls for comprehensive data on racial discrepancies in arrest rates across different crimes compared to actual offending populations. O'Neil suggests statisticians could develop methods to account for racial bias in criminal records if such data existed, though she acknowledges this ground truth is extremely difficult to collect.

Key Insight: Political grief must transform into organized resistance rather than passive acceptance or futile bargaining over institutional remedies that won't materialize.

O'Neil maps her emotional response to Trump's election onto the Kübler-Ross grief model, moving from denial through West Wing binge-watching to anger on New York streets, past bargaining over recounts, through depression, and finally to acceptance. She distinguishes her version of 'depression' as clear-eyed realism, noting she's been in 'war mode' since Occupy Wall Street and never believed Hillary was a strong candidate. The piece culminates in a call to action, emphasizing that acceptance doesn't mean normalization but rather embracing activist roles to protect vulnerable communities in what she calls 'post-fact America.'

Key Insight: Algorithms alone cannot solve the fake news problem; human editorial oversight is essential for social media platforms.

This is a brief announcement post where Cathy O'Neil shares her contribution to a New York Times 'Room for Debate' feature on combating fake news. Her position argues that social media companies like Facebook need to hire human editors rather than relying solely on algorithms to curate content. The post itself contains minimal analysis, serving primarily as a pointer to her NYT opinion piece.

Key Insight: A lighthearted personal travel update with no analytical content about algorithms, math, or society.

This is a brief personal travel post about O'Neil's attendance at the Miami Book Festival as a National Book Award longlister. She describes reading from her work, exploring Miami's Wynwood Walls graffiti art district with fellow author Karan Mahajan, and visiting South Beach with Jonathan Rabb. The post is light and humorous, ending with her falling asleep at 8pm after mojitos.

Key Insight: Facebook's engagement metrics don't measure what users actually want—they measure human weaknesses, and optimizing for these proxies floods our information environment with the intellectual equivalent of junk food.

O'Neil argues that Facebook's algorithms contributed to election manipulation through fake news, targeted political ads, and filter bubbles that created hyper-partisan echo chambers. She proposes that the core problem is 'engagement proxy madness'—Facebook uses clicks and likes as proxies for interest, but these actually expose human weaknesses rather than genuine preferences. This creates feedback loops where both users and news organizations are incentivized toward junk content. She calls for Facebook to hire external auditors like her company ORCAA to measure the algorithm's effects, since the data is private and cannot be independently analyzed.

Key Insight: Moral injury from political catastrophe is real and should be acknowledged, but the cure is ambitious, creative action at the local and global level rather than despair.

Written in the aftermath of the 2016 election, this anonymous guest post describes the author's visceral emotional breakdown at a black-tie event honoring Hank Greenberg, former AIG head. The jarring contrast between elite normalcy and the political catastrophe triggered what the author identifies as 'moral injury.' After crying uncontrollably in their car, the author found clarity: the pain is real and must be acknowledged, but the antidote is action. The post calls for creative, ambitious thinking—focusing on local and global change rather than waiting for national politics to improve.

Key Insight: This post documents the international reach of the 'Weapons of Math Destruction' conversation, though it contains no substantive analysis itself.

This is a brief announcement post where Cathy O'Neil shares links to international press coverage of her book 'Weapons of Math Destruction' from publications in the Netherlands, Germany, and Israel. The post contains minimal commentary, serving primarily as a curated list of foreign media coverage.

Key Insight: The unresolved pain of the foreclosure crisis created a reservoir of voter resentment in battleground states that Clinton failed to address, potentially costing her the election.

Tom Adams argues that the foreclosure crisis played a significant role in Clinton's 2016 electoral defeat in key battleground states. He notes the striking overlap between states with the highest foreclosure inventories and the swing states that flipped from Obama to Trump. The devastating economic and emotional impact of foreclosures on middle-class families and communities was never adequately addressed by the Obama administration, and Clinton failed to make it a campaign issue. Adams suggests that a compassionate outreach to affected communities, similar to Bernie Sanders' advocacy, might have secured the few thousand votes needed to change the outcome.

Key Insight: When genuine uncertainty exists in your model's key parameters, stress testing specific scenarios is more honest and useful than generating precise-looking probabilities that create false confidence.

This guest post by Eugene Stern argues that Nate Silver's election modeling was superior to Sam Wang's because Silver acknowledged genuine uncertainty rather than generating false precision. Wang's 98-99% Clinton win probability was based on arbitrarily choosing a tiny standard error parameter despite historical evidence showing polls can easily be off by 2-3%. Silver's approach used stress testing—examining specific scenarios like Brexit where polls missed by 4%—which revealed that a Brexit-sized polling error would hand Trump an easy victory. The post concludes that communicating uncertainty honestly, rather than false confidence, is the responsible way to model inherently unpredictable events.

Key Insight: When political promises inevitably fail, we must both document the failure and protect those who will be scapegoated for it.

Written immediately after the 2016 election, this post argues that Trump voters trusted his unrealistic promises while either embracing or ignoring his hatred toward marginalized groups. O'Neil calls for a two-pronged response: documenting how Trump fails to deliver on his promises so voters can see the truth, and actively protecting vulnerable populations from emboldened hateful actors. She emphasizes that passive bystanders who ignored the hate cannot be relied upon, and urges citizen journalism, civil engagement across divides, and support for civil rights organizations. The post frames this as a long-term strategic effort requiring local organizing and reimagining progressive politics.

Key Insight: In moments of political crisis, the luxury of passive citizenship disappears - distinguishing truth from propaganda becomes everyone's personal responsibility.

Written immediately after the 2016 election, this brief call to action urges readers to move past shock and analysis paralysis. O'Neil dismisses post-mortem discussions about polling failures and demographics as distractions. Instead, she advocates for supporting independent journalism and preparing for active civic engagement. The post frames the moment as a turning point where passive citizenship is no longer an option.

Key Insight: Political engagement can start before voting age, with young people capable of researching candidates and articulating policy-based reasoning for electoral choices.

This post shares a voting guide created by the author's son Aise, who is too young to vote but wanted to influence his mother's ballot. The guide covers federal, state senate, and state assembly races in New York, providing brief justifications for each candidate recommendation. The selections lean progressive and libertarian, with emphasis on drug legalization, criminal justice reform, and social welfare policies. The author presents it as an example of her son distilling real political information into actionable recommendations.

Key Insight: When platforms optimize for engagement while ignoring truth, misinformation isn't a bug—it's the predictable consequence of the system working exactly as designed.

Cathy O'Neil argues that Facebook's engagement-optimizing algorithm has become a major threat to democracy by spreading fake news and creating ideological echo chambers. She notes that because Facebook optimizes for time-on-site rather than truth, misinformation has proliferated unchecked. The post also criticizes media obsession with polling data at the expense of substantive policy discussion. O'Neil calls for Facebook to hire editors to curate content, redesign its algorithm to burst filter bubbles, and for media to reduce poll coverage in favor of real issues.

Key Insight: Librarians represent an ethical model for data work that modern data scientists should aspire to emulate.

Cathy O'Neil announces receiving the PrivCo Spotlight Award from the Special Libraries Association of New York. The award recognizes bloggers, website founders, and those working outside traditional library roles. She was chosen for her work as a data scientist, her MathBabe blog, and her National Book Award-nominated book Weapons of Math Destruction. O'Neil expresses genuine enthusiasm about the honor, particularly because of her admiration for librarians.

Key Insight: Algorithms are neither inherently good nor evil—the same predictive model can either help struggling students or punish them, depending entirely on how institutions choose to act on its outputs.

Cathy O'Neil examines Georgia State University's algorithm that identifies at-risk students and connects them with advisers, arguing it represents a positive use of big data. Unlike the 'Weapons of Math Destruction' she typically critiques, this system helps rather than punishes vulnerable populations. The algorithm triggered 51,000 advising meetings and improved graduation rates, particularly for low-income and minority students. O'Neil emphasizes that the real value lies in the human intervention the algorithm enables, not the technology itself. She notes that the same algorithm could be used destructively, as nearly happened at Mount St. Mary's College, illustrating that algorithms are tools whose impact depends on their application.

Key Insight: The TSA's behavioral detection program exemplifies how pseudoscience dressed up in official scoring systems can waste hundreds of millions of dollars while providing security theater instead of actual safety.

This guest post by epidemiologist Matt Freeman critiques the TSA's SPOT (Screening Passengers by Observational Techniques) program, which claims to identify potential threats through behavioral detection. Freeman recounts his own absurd encounter with a SPOT agent asking nonsensical questions about 'provinces' and 'whereabouts,' causing him to miss his flight. He traces the program's pseudoscientific origins to Paul Ekman's facial emotion research, which has been thoroughly debunked by subsequent scientists who could not replicate his findings. Despite costing over $900 million and flagging 232,000 passengers with only 1,710 unrelated arrests, the TSA continues the program while ignoring more sensible alternatives like the PreCheck vetting system.

Key Insight: Voluntary data sharing programs are Trojan horses - they begin as opt-in benefits but inevitably become the baseline against which non-participants are penalized.

Insurance companies are exploring using Twitter data to predict health outcomes and adjust premiums accordingly, with research showing tweets can predict heart disease better than traditional health measures. The article's framing as 'lower rates for optimistic tweeters' masks the reality that pessimistic tweeters would face higher costs. While no insurer currently uses this openly, the trajectory is clear: voluntary data sharing to 'prove' healthiness will likely become compulsory. O'Neil connects this to personality prediction research showing employers could secretly score job applicants based on social media, bypassing disclosure requirements entirely.

Key Insight: This is a promotional travel post with no analytical content to extract.

This is a brief announcement post about Cathy O'Neil's visit to Madison, Wisconsin for the Wisconsin Book Festival. She mentions enjoying a meal at a local restaurant, being introduced by mathematician Jordan Ellenberg, and doing radio interviews at WORT and WPR. The post is purely logistical and personal, containing no substantive discussion of algorithms, data science, or the themes from her book.

Key Insight: When tech companies offer 'free' educational tools, students become unpaid labor training algorithms while surrendering their privacy rights and legal recourse.

O'Neil critiques Facebook's Summit Basecamp 'personalized learning' initiative as exploitative of students. She argues that the 20,000 students involved are essentially unpaid workers training Facebook's algorithm during classroom time, with no guarantee of actual learning. Parents are pressured to sign away privacy rights and agree to arbitration clauses that prevent lawsuits. O'Neil emphasizes that these students are disproportionately poor and minority children who are already the most surveilled population in history.

Key Insight: The personal data we casually entrust to tech giants creates surveillance capabilities that demand the same regulatory protections we grant to credit reporting agencies.

After losing all phone contacts due to a cloud sync error between Windows phones, the author reflects on how much personal data the major tech companies (Apple, Google, Microsoft) accumulate through our devices. He argues that these companies could easily create comprehensive social network maps and real-time location tracking of users worldwide. Drawing on examples of email scanning by Google and Yahoo's cooperation with government surveillance, he calls for consumer data protection rights similar to credit reporting laws, including the right to know what data is collected, correct it, and learn who it's shared with.

Key Insight: Building public discourse around algorithmic accountability requires bringing together technologists, legal experts, civil liberties advocates, and policy scholars.

Cathy O'Neil announces an upcoming panel discussion at the New America Foundation about her book 'Weapons of Math Destruction.' The event will feature experts from various fields including criminal justice technology, surveillance and civil rights, data innovation policy, and democratic governance. She also mentions a companion essay she wrote for New America called 'Alien Algorithms.' The post serves primarily as an event announcement rather than substantive analysis.

Key Insight: This is an event announcement post with minimal analytical content, primarily serving to publicize O'Neil's book tour appearances in New York.

This is a brief promotional post announcing three upcoming speaking events in New York City where Cathy O'Neil will discuss her book 'Weapons of Math Destruction.' The events span from mid-October 2016 and include talks at Thoughtworks (co-hosted with Data-Pop Alliance), Betaworks, and Data & Society. The post provides practical details like dates, times, locations, and registration links for each event. It's primarily logistical rather than substantive in nature.

Key Insight: When tech founders dismiss ethics with fatalism about progress, look at who's buying their products and who's already been harmed.

Facial recognition technology has reached alarming accuracy levels, with a Russian company achieving 73% match rates against million-image databases. The company's founders dismiss privacy concerns with cavalier arguments about progress being inevitable and privacy already being dead. O'Neil contrasts their dismissive attitude with concrete harms: the technology is being sold to authoritarian regimes, deployed for mass surveillance in Moscow, and has already been weaponized to harass women. The post highlights the dangerous gap between rapidly advancing surveillance capabilities and any meaningful ethical oversight.

Key Insight: Sometimes even critical voices deserve to celebrate their moments of mainstream recognition.

This is a brief celebratory post where O'Neil shares news about positive media coverage for her book 'Weapons of Math Destruction.' She highlights Clay Shirky's review in the New York Times Book Review, her appearance on the EconTalk podcast with Russ Roberts, and a Guardian Science podcast featuring Douglas Rushkoff. The post is essentially a promotional update during her book tour.

Key Insight: This is primarily an event announcement and media roundup post with little substantive analytical content.

This is a brief promotional post announcing Cathy O'Neil's book talk at Harvard Bookstore and celebrating media coverage of her book 'Weapons of Math Destruction.' She mentions being featured in The New Yorker's Talk of the Town section, which also covered her bluegrass band. The post lists several recent media mentions including the Boston Globe, Wall Street Journal, and another New Yorker article about online discrimination.

Key Insight: When a flawed algorithmic score is the only component with meaningful variance, it dominates outcomes regardless of its official weighting—making the 'one of many factors' defense mathematically hollow.

O'Neil dismantles the common defense that problematic algorithmic scores like teacher VAMs are acceptable because they're 'one of many' factors in decisions. She explains that when other evaluation components have little variance (most teachers rated 'acceptable' or 'excellent'), the VAM score was specifically introduced to create spread between teachers. This means even if VAM counts for only 50% of a score, it dominates the variance and therefore determines rankings. The mathematical reality is that mixing a noisy, unreliable measure with low-variance measures doesn't dilute its effect—it amplifies its importance in determining outcomes.

Key Insight: A weekly reading list that connects dots between corporate malfeasance, algorithmic discrimination, and systemic racial injustice.

This is a curated link roundup post sharing articles Cathy O'Neil found noteworthy during the week. The topics span data privacy breaches, algorithmic pricing discrimination, corporate whistleblower retaliation, insurance industry disruption from big data, racial inequity in gun rights and criminal justice, and how data-driven marketing reinforces class divisions. The collection reflects her ongoing interests in how technology and algorithms intersect with social justice issues. She also references discussing these topics on her Slate Money podcast appearance.

Key Insight: This is a logistical announcement post without substantive analysis or argumentation.

This is a brief announcement post where Cathy O'Neil shares her upcoming UK book tour schedule for 'Weapons of Math Destruction.' She lists speaking engagements at Cambridge University, London's How To Academy, and King's College London. The post also mentions media appearances including BBC's Today Programme, interviews with Significance magazine, the Guardian Science podcast, and BBC World Service's Business Daily.

Key Insight: This is a logistical update post with minimal analytical content, primarily serving to aggregate O'Neil's media appearances during her book tour.

This is a brief promotional post where O'Neil shares media appearances and publications related to her book tour for 'Weapons of Math Destruction.' She links to a Q&A in Jacobin about algorithmic black boxes, a Slate piece on big data's impact on college admissions, and a Times Higher Education review. The post closes with news that her book became a bestseller in Canada.

Key Insight: When statisticians deflect questions about systemic discrimination by treating them as individual cases, they're abandoning their expertise precisely when it's most needed.

O'Neil critiques statistician Richard Berk's dismissive response to a Black man's question about whether risk assessment tools discriminate against Black men. Berk deflected by treating the question as individual rather than statistical, asking about personal history instead of addressing systemic bias. O'Neil argues this is a fundamental failure for a statistician to interpret an obviously statistical question about group discrimination as a question about one person. She notes that while Berk later acknowledges fairness-accuracy tradeoffs, the constitutional guarantee of fairness should make such tradeoffs irrelevant.

Key Insight: Sometimes progress is real and worth celebrating, even in unglamorous forms like functional public bathrooms.

O'Neil reflects on the dramatic improvement of Penn Station bathrooms since the 1980s, using them as a symbol of broader urban progress in New York City. She recalls the grim conditions of public facilities during that era—dirty needles, lack of basic supplies, threatening atmospheres—and contrasts them with today's relatively clean, functional spaces. She extends this observation to JFK airport, noting similar improvements. The post is a brief meditation on 'anti-nostalgia,' acknowledging that while some things like rent were better in the past, basic public infrastructure has genuinely improved.

Key Insight: The validation that matters most for a book about algorithmic harm isn't awards but readers recognizing their own experiences in its pages.

Cathy O'Neil celebrates her book 'Weapons of Math Destruction' making the National Book Award Nonfiction longlist just 10 days after launch. She reflects on the significance of this recognition for spreading her ideas about algorithmic harm, and shares her experiences meeting readers on her book tour who validate her work with their own stories of encountering problematic algorithms. She expresses gratitude for the timing, marketing support, and encouragement from friends who helped her through the difficult writing process, particularly when she thought the book was unwritable.

Key Insight: The automation of surveillance doesn't make it less invasive—it makes the overbearing boss omnipresent and inescapable.

O'Neil highlights two new examples of algorithmic surveillance systems that emerged after her book's publication. Joberate scores employees on their likelihood of job-hunting, essentially automating and scaling the behavior of a paranoid boss. Litify aims to use AI to predict litigation outcomes and transform legal finance. O'Neil presents both as troubling extensions of the 'weapons of math destruction' she warns about, noting the irony of companies claiming they're not creepy while engaging in surveillance capitalism.

Key Insight: This post serves purely as an event announcement and contains no substantive content to analyze.

This is a brief announcement post where Cathy O'Neil promotes an upcoming talk about her book at an Alt Banking meeting associated with the Occupy movement. The post provides logistical details including time, date, and location at Columbia University's International Affairs Building. No substantive arguments or analysis are presented in the post itself.

Key Insight: This is a promotional announcement post with no substantive content to analyze beyond directing readers to a podcast discussion of her book.

This is a brief announcement post where Cathy O'Neil shares that her book 'Weapons of Math Destruction' was discussed on her Slate Money podcast. The post serves purely as a promotional notice directing readers to listen to the episode. No substantive arguments or analysis are presented in the post itself.

Key Insight: When systemic fraud stems from management's unrealistic quotas, punishing only low-level workers while executives escape accountability reveals a fundamental power imbalance that unionization could help correct.

Wells Fargo was fined $185 million for creating unauthorized accounts and charging customers fees without consent. The bank fired 5,300 low-level workers over five years, but the real culprit is management's unrealistic sales quotas that incentivized fraud. O'Neil argues that executives either knew about the widespread fraud and didn't care because it was profitable, or had insufficient controls—either way, they should be held accountable, not just the workers. She concludes that low-paid, pressured bank workers need union protection to shield them from being scapegoated for systemic problems created by management.

Key Insight: A simple event announcement with no substantive content to analyze.

This is a brief promotional post announcing two media appearances to discuss 'Weapons of Math Destruction.' O'Neil will appear on the Brian Lehrer Show on WNYC radio and later that evening will have a conversation with journalist Felix Salmon at a Barnes & Noble location in Manhattan. The post provides practical details for listeners and attendees.

Key Insight: The real danger isn't that math is biased, but that mathematical complexity provides cover for discriminatory systems to operate without accountability.

This post compiles early reviews of O'Neil's book 'Weapons of Math Destruction' from various publications including Boing Boing, Scientific American, CNN Money, and academic blogs. O'Neil pushes back against a CNN headline calling 'math racist,' clarifying her actual argument is that discriminatory practices hide behind mathematical legitimacy. She expresses excitement about the growing public discussion of algorithmic bias and asks readers to review the book on Amazon, acknowledging its controversial nature makes it vulnerable to negative review campaigns.

Key Insight: The release of 'Weapons of Math Destruction' marks a pivotal moment in bringing algorithmic accountability critiques to mainstream attention.

Cathy O'Neil announces the release day of her book 'Weapons of Math Destruction,' expressing nervous excitement about the milestone. She also shares a few recommendations: Carrie Fisher's new advice column at the Guardian, and a comprehensive report by Team Upturn on predictive policing. The post is brief and personal, mixing self-deprecating humor about her anxiety with genuine enthusiasm for work by others in adjacent spaces. She closes by promoting an upcoming speaking engagement in Berkeley.

Key Insight: Allowing powerful tech companies to secretly self-regulate AI ethics, without public transparency about how they define harm and benefit, is fundamentally unacceptable.

O'Neil reacts with outrage to news that major tech companies are forming a secret partnership to self-regulate AI ethics. She highlights the absurdity of unnamed insiders from Google, Amazon, Facebook, IBM, and Microsoft making decisions about AI's benefits and harms without public input or transparency. The post argues that such consequential decisions about who gets helped or hurt by AI cannot be left to closed-door industry discussions.

Key Insight: Small infrastructure expansions like bike-sharing can meaningfully improve daily urban life by providing alternatives to overcrowded public transit.

Cathy O'Neil expresses enthusiasm about Citi Bike's expansion to the Columbia University area. The new bike stations will allow her to drop her kid off at school and bike downtown instead of taking the subway. She sees this as a major quality-of-life improvement, particularly avoiding the crowded 1 train during rush hour. She acknowledges with humor that the benefit depends on not everyone else having the same idea.

Key Insight: This post is purely promotional, announcing media coverage and events for the Weapons of Math Destruction book launch rather than offering substantive analysis.

This is a brief promotional post announcing that an excerpt from 'Weapons of Math Destruction' has been published in The Guardian. O'Neil shares her excitement about the UK publicity push and promotes an upcoming London book tour event at the How To: Academy. She also links to a Scientific American review of her book by Evelyn Lamb.

Key Insight: A quick announcement sharing mainstream media coverage of O'Neil's work on algorithmic accountability as her book launch approached.

This is a brief promotional post where Cathy O'Neil shares media coverage of her upcoming book 'Weapons of Math Destruction.' She highlights a Time Magazine piece written by Rana Foroohar, whom she had previously interviewed on Slate Money about Foroohar's book on finance. She also mentions being profiled as a 'math nerd' by Bloomberg.

Key Insight: Technical criticisms of algorithmic fairness tools often mask political opposition to addressing discrimination, as the methodology's limitations are fixable but the will to fix them requires actually wanting to find bias.

O'Neil explains the BISG (Bayesian Improved Surname Geocoding) methodology, which uses last names and zip codes to estimate racial probabilities via Bayes' rule, relying on America's persistent residential segregation. She addresses Republican criticism that called it 'junk science' when the CFPB used it to detect racial discrimination in auto lending. O'Neil argues the methodology could be improved by updating priors with auto lending population distributions, which would address the technical complaints. She suggests the real opposition isn't statistical but political—similar to voter fraud concerns where rare errors are used to justify inaction on systemic problems. The post includes Python code implementing BISG as a practical resource for researchers.

Key Insight: This is a logistical book tour announcement rather than a substantive blog post, though the book title itself signals O'Neil's thesis that algorithmic systems can amplify societal harms.

This is a promotional announcement post where Cathy O'Neil shares the upcoming book tour events for her new book 'Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy.' The post lists approximately ten public events across major US cities including New York, Seattle, San Francisco, Washington D.C., Cambridge, and Madison between September and October 2016. Several events feature conversations with notable figures like Felix Salmon and Jen Golbeck. The post serves purely as an informational calendar for readers who want to attend.

Key Insight: Predictive policing systems like Chicago's Heat List don't actually prevent crime or protect people—they simply predict and facilitate increased police surveillance and arrests of targeted individuals.

O'Neil examines a peer-reviewed study evaluating Chicago's predictive policing 'Heat List' system, which claims to identify individuals likely to be involved in shootings. The study found that being on the list did not protect people from becoming victims of violence, nor did it reduce crime rates beyond pre-existing downward trends. However, individuals on the list experienced dramatically increased police surveillance and were more likely to be arrested. The key finding is that the list predicts who police will arrest, not who will commit crimes—essentially forecasting police behavior rather than criminal behavior. O'Neil argues this exposes the circular logic of predictive policing systems that claim success based on arrest rates.

Key Insight: The financial industry's flexible definition of alpha allows portfolio managers to hide substantial risk while claiming superior performance, putting investors' money at risk.

O'Neil challenges the common definition of alpha as risk-adjusted returns above market performance, arguing that alpha actually represents returns uncorrelated to the market. She explains that the standard regression formula only captures market-correlated risk, leaving other risks invisible in the error terms. Using CAPM theory embeds assumptions that alpha should be zero, but her quant definition makes no such assumptions. The post warns that this distinction matters because hedge funds and portfolio managers exploit the ambiguity to hide risk while claiming impressive performance metrics.

Key Insight: What we call 'clean' pool water is actually chemically-treated water so toxic that nothing can live in it—a reminder that common descriptive language often obscures reality.

O'Neil uses her personal experience with a backyard pool to challenge our perception of 'clean' swimming pool water. After her pool turned green from two weeks without maintenance, she observed how massive amounts of liquid chlorine transformed it back to blue. She argues that what we call 'clean' pool water is actually so chemically toxic that algae cannot survive in it. The post is a lighthearted observation about semantics rather than a health warning, clarifying in an update that she's not anti-chemical, just questioning the word 'clean.'

Key Insight: Algorithms, like Trump, can produce harmful outcomes not through malicious intent but through bad success metrics and biased training data—there is no escaping the need for human judgment about values.

Cathy O'Neil argues that Donald Trump operates like a biased machine learning algorithm, lacking moral principles but following a consistent method: optimizing for crowd engagement at rallies. He iteratively tests rhetoric, keeping what gets reactions and discarding what bores. This process, trained on biased data (rally attendees), produces outputs that work for his audience but alienate everyone else. O'Neil uses this analogy to illustrate broader dangers of algorithmic decision-making and critiques utopian visions of robot politicians.

Key Insight: Even authors who write serious critiques of algorithmic power must navigate the awkward transition from intellectual commentator to book promoter.

Cathy O'Neil announces that the final physical copy of her book 'Weapons of Math Destruction' has arrived, sharing photos of her editor and son holding it. She notes the book has received a Reuters review and was listed in Scientific American's new releases. She apologizes in advance for the upcoming promotional blitz, acknowledging that self-promotion isn't her usual style but is necessary for book launches.

Key Insight: The question that defines any futurism isn't what technologies it imagines, but whose liberation those technologies are meant to serve.

This guest post by mathematician Matilde Marcolli challenges the Silicon Valley monopoly on futurist thinking by tracing its themes back to early 20th century Russian Futurism and Cosmism. She argues that the crucial difference between capitalist and anarchist futurism isn't the topics they explore—immortality, space, AI—but whose future they envision. The post introduces Anarcho-Transhumanism as an anti-capitalist alternative that embraces technology for liberation rather than elite preservation. Marcolli highlights 'morphological freedom' as a key concept, emphasizing transgender rights and disability rights over botox treatments for billionaires. She concludes that science and technology have enormous potential when used inclusively rather than for profit and exploitation.

Key Insight: Tech companies deflect moral responsibility for surveillance tools by invoking 'futurists' as ethical authorities, despite futurists being ill-equipped to address civil liberties and representing the same power structures that threaten them.

Cathy O'Neil criticizes Microsoft for deflecting responsibility for a facial recognition tool designed to read emotions at political rallies by suggesting ethical questions should be directed to 'futurists.' She argues this response both absolves technologists of accountability for their creations and absurdly positions futurists—a group she characterizes as homogeneous Silicon Valley insiders obsessed with the singularity—as moral authorities. O'Neil contends that futurists lack the historical and political knowledge needed to address civil liberties concerns, comparing them to the Empire in Star Wars while protesters represent the Rebel Alliance.

Key Insight: Individual retirement savings are not just failing in practice—they're mathematically inferior to pooled systems like Social Security because risk-pooling eliminates the need for everyone to save for worst-case scenarios.

O'Neil argues that the American 401(k) system is fundamentally flawed both in practice and theory. Most Americans aren't saving adequately for retirement, and even if they were, individual savings accounts are mathematically inefficient compared to pooled systems. She contends that expecting everyone to save for worst-case scenarios (living to 105 with expensive end-of-life care) represents massive societal inefficiency. The solution is to expand Social Security, which pools risk effectively and remains financially viable despite popular misconceptions.

Key Insight: Even when algorithms avoid obvious proxies for race and class, claims of 'objectivity' ring hollow without transparent auditing processes.

O'Neil shares a roundup of recent developments she finds encouraging, from the elimination of New York's tampon tax to new efforts at algorithmic accountability. She highlights a new bail algorithm from the Arnold Foundation, praising its avoidance of race and class proxies while criticizing its marketing claims of 'objectivity' without transparent auditing. The post also points to Julia Angwin's call for algorithmic accountability, new research on debiasing sexist word corpora, and concerns about underpaying Mechanical Turk workers in academic research.

Key Insight: By criminalizing the methods researchers need to audit algorithms for discrimination, the CFAA inadvertently protects algorithmic bias from scrutiny.

The Computer Fraud and Abuse Act criminalizes violations of website terms of service, even when researchers are investigating algorithmic discrimination for the public good. O'Neil highlights how this law impedes important work like ProPublica's investigation into racist recidivism models. She cites research arguing that discrimination audits, while technically violating ethical guidelines, serve a clear public good and should be permitted. The authors argue that moving discrimination into algorithms has actually protected the wicked by making it harder to detect while simultaneously restricting the audit methods that could expose it. The post advocates for CFAA reform to include auditing exceptions in the public interest.

Key Insight: When companies claim alternative credit scoring 'expands access,' they're often masking systems that weaponize people's most vulnerable moments against them while evading accountability.

Cathy O'Neil warns about a dangerous new partnership between ZestFinance and Baidu that will use Chinese citizens' browsing history to determine credit scores. She argues that the phrase 'giving consumers more access to credit' is propaganda that masks discriminatory practices. The partnership combines ZestFinance's philosophy that 'all data is credit data' with Baidu's massive trove of personal search data, including sensitive health and employment queries. Unlike the U.S., China lacks anti-discrimination laws for credit scoring, meaning anything goes. O'Neil sees this as a nightmare scenario where people's most vulnerable moments become weapons against them, with no transparency or appeals process.

Key Insight: Algorithmic auditing must examine both the model's coefficients and its choice of inputs to detect discrimination, though detecting problems is easier than knowing how to fix them.

Cathy O'Neil announces the launch of ORCAA (O'Neil Risk Consulting and Algorithmic Auditing), a company dedicated to auditing algorithms and assessing their risks. She discusses her research into algorithmic auditing methods, highlighting Oaxaca's decomposition as a framework for separating discriminatory outcomes into 'worse attributes' and 'worse coefficients.' Using credit scoring as an example, she explains why both the model's coefficients and the choice of input variables matter for detecting discrimination. She acknowledges a key limitation: current methods can detect problems but don't necessarily tell us how to fix them.

Key Insight: A study's sampling methodology can mathematically mask the very bias it claims to measure, making headlines about 'no racial bias in police shootings' dangerously misleading when the underlying data and methods don't support such conclusions.

This guest post by Brian D'Alessandro critically examines Roland Fryer's Harvard study claiming zero racial bias in police shootings. The analysis identifies two fundamental flaws: the data comes from a single precinct (Houston) using self-reported data, creating potential selection bias, and the sampling methodology conflates different populations in ways that can mathematically mask real bias. D'Alessandro demonstrates through Bayes' rule how a biased sampling scheme can neutralize actual population-level bias, making the study's conclusions unreliable. He argues the real question isn't whether cops shoot equally at armed suspects of different races, but why black men are more likely to face police force in the first place.

Key Insight: Wellness culture serves as a depoliticizing force that transforms systemic failures into personal inadequacies, making people blame themselves rather than question the systems that constrain them.

O'Neil reviews 'The Wellness Syndrome,' a book arguing that modern wellness culture has become a tool for shifting blame from systemic failures onto individuals. The authors trace how politicians from Thatcher to Clinton promoted individual responsibility as a shield against governmental accountability. The result is a depoliticized populace that internalizes shame for failing to be sufficiently optimized, rather than questioning inadequate job markets or discriminatory systems. O'Neil connects this to her own work on how employers increasingly control workers' bodies and lives through wellness programs and data collection.

Key Insight: Broken Windows policing creates a vicious cycle where biased enforcement generates biased data that then feeds algorithmic systems, compounding discrimination at every stage from patrol routes to prison sentences.

Cathy O'Neil endorses Campaign Zero's call to end Broken Windows policing, arguing that this approach has criminalized communities of color and created feedback loops that perpetuate bias. She critiques a New York Times study on police shootings for ignoring how uneven policing increases the likelihood of police interactions for Black Americans. Using NYC's Stop and Frisk as a case study, she shows how the policy disproportionately targeted minorities while being ineffective at addressing violent crime. O'Neil connects this to predictive policing algorithms, explaining how biased data from these encounters gets fed into systems that compound the original discrimination in sentencing and parole decisions.

Key Insight: The appropriateness of AI depends not on the technology itself but on whether diverse, thoughtful teams interpret and apply its results in ways that help rather than harm vulnerable populations.

O'Neil argues that the question of when AI is appropriate depends less on the technology itself than on how results are interpreted and applied. Using recidivism risk models as a case study, she demonstrates how algorithms like COMPAS perpetuate racial bias by producing different error rates for Black and white defendants. Rather than calling for elimination of such models, she advocates for understanding what they actually measure, questioning why they're used punitively, and retooling them to help rather than harm disadvantaged populations. She concludes that AI development isn't inherently inappropriate, but the homogeneous teams building it and their narrow optimization-focused training make most current AI applications problematic.

Key Insight: Essential scientific infrastructure like arXiv should be publicly funded rather than left to survive on individual donations when federal science agencies have billion-dollar budgets.

O'Neil argues that the NSF should fund arXiv rather than leaving it to beg for small personal donations. She points out that arXiv plays a pivotal role in basic research across multiple scientific fields, yet operates on less than $1 million per year while the NSF has a $7.5 billion budget. Given the NSF's stated mission to 'promote the progress of science,' she sees the lack of support as inexplicable and outrageous.

Key Insight: Even authors who write confidently about weighty topics like algorithmic bias experience vulnerability and anxiety when their work faces public judgment.

Cathy O'Neil shares her experience recording the audiobook version of 'Weapons of Math Destruction' at Random House studios. She describes the challenges of voice recording, including retakes for flat tone and difficult phrases. Despite initial worries about finding typos or outdated content, she expresses satisfaction with the book while bracing for future negative reviews. She also announces the book's new website featuring early blurbs and a positive Publisher's Weekly review.

Key Insight: Even those who have fought to accept their own fatness can harbor internal fat-shaming toward people larger than themselves, revealing how deeply society's weight biases are internalized.

Cathy O'Neil reflects on a This American Life episode about fatness that challenged her own assumptions about weight and identity. She relates to the first story about accepting being fat and rejecting euphemisms like 'overweight,' but the third story forces her to confront her own internal fat-shaming toward people larger than her 'Lane Bryant Fat' category. She admits to harboring judgmental thoughts toward very fat people despite knowing how hard she's tried to change her own weight. The piece ends with her personal dilemma about gastric bypass surgery for diabetes prevention, which feels like betraying her hard-won fat identity.

Key Insight: When social pressure makes certain positions uncomfortable to admit publicly, traditional polling methods may systematically misread voter intent, revealing the gap between what people say and how they actually behave.

In the wake of Brexit's surprising outcome, the post examines why polls, betting markets, and financial markets all predicted a Remain victory when Leave actually won. The author notes that polls had consistently shown Remain ahead by 52-48%, but the actual result was the exact opposite. Rather than offering definitive answers, O'Neil presents five hypotheses for the polling failure, ranging from demographic voting patterns to social desirability bias to methodological limitations in exit polling. The piece serves as a cautionary tale about the limits of prediction when human behavior and social pressures complicate data collection.

Key Insight: Statistical analysis reveals that the foundational assumption of Broken Windows policing—that aggressive enforcement of minor violations reduces serious crime—has no empirical support, making the case for data-driven reform of police practices.

The Office of the Inspector General for the NYPD released a report using statistical analysis to demonstrate that 'Broken Windows' policing doesn't reduce felony crime. The six-year study found no correlation between quality-of-life summonses or misdemeanor arrests and drops in felony crime. In fact, as quality-of-life enforcement dramatically declined between 2010-2015, felony crime did not increase. The report recommends the NYPD take a more data-driven approach and analyze whether their enforcement disproportionately impacts minorities and young males. The NYPD has called the report flawed and promised a response within 90 days.

Key Insight: Even perfectly fair, algorithmically-drawn districts would still favor Republicans because Democrats' natural geographic clustering in cities creates an inherent structural disadvantage that no technical solution can fix.

O'Neil explores the mechanics of gerrymandering, explaining how parties manipulate district boundaries to maximize wins (by thin margins) while concentrating opponents into 'loss' districts. She discusses algorithmic solutions for creating compact, fair districts, then examines detection methods like compactness scores and mean-median statistical tests. However, she reveals a crucial complication: even fairly-drawn compact districts show statistical bias because Democrats naturally cluster in cities and college towns, effectively 'gerrymandering themselves.' This geographic reality means Republicans benefit from any districting system, giving them no incentive to reform it.

Key Insight: The 'smart' city movement isn't about solving problems—it's about monetizing public space and personal data while actual human needs remain unfunded because they lack profit potential.

Cathy O'Neil critiques the 'smart city' movement as a cynical exercise in surveillance capitalism disguised as innovation. She argues that most smart products exist not because they solve real problems, but because Silicon Valley has too much money chasing unnecessary ideas. The essay highlights the absurdity of having smart WiFi kiosks but no public bathrooms, smart tampons but no free tampons for homeless women. O'Neil exposes how 'smart' infrastructure like NYC's LinkNYC kiosks are really about harvesting data and selling targeted ads, not serving public needs. The core problem isn't technological enthusiasm—it's that profit motives determine what gets built while actual human needs go unaddressed.

Key Insight: Treating for-profit college debt as a sunk cost rather than a recoverable asset allows the government to stop wishful thinking and actually fix the broken accreditation system that enabled predatory lending.

O'Neil argues that student debt from predatory for-profit colleges should be forgiven as a 'sunk cost' that the government incurred through regulatory failure. She explains how for-profit colleges like Corinthian, ITT Tech, and University of Phoenix gamed the federal aid system through questionable accreditation and deceptive marketing to poor students. While the government hesitates due to expense and precedent concerns, O'Neil contends the debt will never be repaid anyway, so acknowledging it as a sunk cost allows for learning from the mistake and fixing the broken accreditation system.

Key Insight: The computer revolution has made statistical analysis trivially easy compared to when math curricula were designed, yet schools still treat calculus as the pinnacle while relegating statistics to an afterthought—a mismatch that fails to prepare students for a data-driven world.

This guest post by Kevin H. Wilson argues that the high school mathematics curriculum should be radically restructured around statistics, programming, and data analysis rather than the traditional calculus-focused track. Wilson identifies four biases preventing change: the historical difficulty of computation, the preference for easily measurable skills, the isolation of learning within classrooms, and the challenges of retraining teachers. He proposes merging algebra with programming, replacing geometry with discrete math and data structures, substituting Algebra II with statistics courses using open datasets, and making machine learning the capstone course instead of calculus. The ultimate goal would be community-based capstone projects where students use data to advocate for local change.

Key Insight: When platforms are acquired, years of carefully cultivated personal data get transferred to new corporate owners with different agendas and the power to change privacy rules unilaterally.

O'Neil explains why she deleted her LinkedIn account before Microsoft's $26 billion acquisition, citing spam, bogus requests, and 'hyped up crap about magic bullets using big data' as reasons. She argues the Microsoft acquisition represents a significant drop in alignment between the platform and its users' interests. The post warns that years of professional social data people have carefully built will now belong to a corporation that can change privacy policies at will. She suspects Microsoft will use the network for marketing experiments and urges readers to delete their accounts.

Key Insight: Institutional corruption thrives when union leadership prioritizes protecting abusers over accountability, turning what should be oversight into systematic obstruction of justice.

Cathy O'Neil celebrates the arrest of Norman Seabrook, head of the Correction Officers' Benevolent Association, who was charged with investing union members' pensions in a failing hedge fund for kickbacks. She details her long-standing disgust with Seabrook's fear-mongering radio ads that dehumanized Rikers prisoners while he systematically protected abusive guards. O'Neil documents his pattern of obstruction, including shutting down prison bus systems to prevent testimony against officers and training guards to abuse prisoners off-camera. She expresses hope that his removal will finally enable reform of Rikers Island.

Key Insight: Tech utopianism often dresses up vague promises in paradigm-shift language, while the real insight about algorithmic complicity comes from understanding how our clicks actively shape harmful systems.

O'Neil offers sardonic summaries of talks she attended, using humor to critique tech utopianism while highlighting genuinely valuable insights. She skewers Don Tapscott's blockchain evangelism and Andrew McAffee's techno-optimism about inequality and climate change. In contrast, she signals genuine appreciation for Carl Pomerance's math talk and Monica Lewinsky's nuanced take on cyber bullying. Lewinsky's point about how clicking on shame-based content trains algorithms receives the most substantive treatment, suggesting O'Neil found real value there.

Key Insight: Including a discriminatory algorithmic score as 'one of many factors' in decision-making doesn't cleanse the process of unfairness - it just obscures it.

O'Neil reflects on the irony of speaking about algorithmic injustice against the poor while staying at a luxury hotel overlooking the Louvre. She draws a parallel between historical injustices (reparations paid to British slave owners rather than slaves) and modern algorithmic discrimination, specifically Wisconsin's use of proprietary recidivism scores in sentencing. She questions whether including a flawed algorithmic score as 'one of many factors' in decision-making actually makes the overall process fair. The post closes with a celebration of middle-aged rebellion against social expectations.

Key Insight: Algorithmic harm can be prevented through rigorous validation practices: open datasets for scrutiny, expert curation for error detection, and causal models that withstand real-world interventions.

Guest author Kareem Carr, drawing from his experience validating cancer research algorithms, proposes three principles for preventing harmful algorithmic systems. He argues that open datasets enable scrutiny of proprietary algorithms, expert manual curation provides essential error detection, and causal models should be demanded over merely predictive ones. The post emphasizes that bad data science obscures real-world algorithm performance while good data science creates deeply understood, transparent systems. His central thesis is that rigorous validation and transparency are the antidotes to algorithmic harm.

Key Insight: The intimidation factor of combining math and genetics shields flawed algorithms from scrutiny, allowing unvalidated tools to make high-stakes decisions about people's freedom and health.

The post argues that the combination of mathematical and genetic complexity creates an intimidation effect that shields flawed algorithms from public scrutiny. O'Neil examines two cases: DNA forensics testing, which suffers from contamination issues and inconsistent results across labs yet heavily influences jury verdicts, and genetic tests claiming to predict opioid addiction risk, which lack transparent accuracy claims and scientific backing yet are being used to deny pain medication. She contends that this intimidation factor allows potentially inaccurate tools to be deployed in high-stakes decisions affecting people's freedom and health.

Key Insight: Transparent data analysis of criminal justice systems reveals that racial bias in sentencing dwarfs economic bias, and this kind of accountability-focused work offers hope for systemic reform.

Cathy O'Neil highlights a data analysis by lawyer-turned-data-scientist David Colarusso examining racial and economic bias in Virginia's criminal sentencing. Using recently opened criminal justice data, Colarusso found that race affects sentence length far more than income, though both matter. His analysis showed that a Black defendant would need to earn over half a million dollars more than a white peer to receive equivalent treatment. O'Neil praises this transparent, accessible approach to exposing systemic bias as a counterweight to harmful uses of big data.

Key Insight: When algorithmic evaluation systems produce wildly inconsistent scores that the state cannot explain, and mandate fixed failure rates regardless of actual performance, they fail the basic test of fairness that any consequential scoring system should meet.

A New York judge ruled that teacher Sheri Lederman's growth score evaluation was 'arbitrary and capricious,' validating long-standing criticisms of value-added models in education. Lederman's score dropped from 14 to 1 between years despite her students performing consistently well on state tests. The judge found multiple flaws in the statistical methodology, including unfair penalties for teachers with very high or low-performing students, unreliability with small class sizes, and the inability to explain year-to-year score swings. O'Neil sees this ruling as corroboration of her critiques of algorithmic evaluation systems.

Key Insight: Big data systems increasingly make consequential decisions using pseudoscientific methods and circular logic, reviving discriminatory practices like phrenology under a veneer of technological sophistication.

Cathy O'Neil highlights three alarming examples of big data misuse that emerged in a single day. Cable One uses FICO scores to determine service quality for customers. Chicago Police use social media data mining to identify gang members, then circularly validate their algorithm by arresting people on the list. Most egregiously, a startup called Faception claims to identify terrorists through facial recognition, essentially reviving phrenology with a tech veneer. O'Neil argues these examples vindicate her earlier warning that big data is becoming the new pseudoscience.

Key Insight: ProPublica's COMPAS investigation provides the first hard proof that recidivism risk algorithms systematically discriminate against black defendants, setting a new standard for algorithmic accountability.

O'Neil celebrates ProPublica's groundbreaking investigation into the COMPAS recidivism risk model, which revealed systematic racial bias in criminal sentencing algorithms. The analysis found that black defendants were nearly twice as likely to be falsely labeled high-risk, while white defendants were twice as likely to be incorrectly classified as low-risk. The study controlled for prior crimes, future recidivism, age, and gender, yet black defendants remained significantly more likely to receive higher risk scores. O'Neil argues this report validates concerns she and others have raised about algorithmic bias and sets a new standard for auditing such systems.

Key Insight: Opaque algorithmic systems, from DNA testing to social media, operate without accountability while institutions that should serve the public instead serve private interests.

This is a link roundup post where O'Neil curates articles on topics she finds noteworthy: universities functioning as financial institutions, flawed DNA testing algorithms in criminal justice, pharmaceutical company charity schemes, Facebook's algorithmic opacity, and debunking breakfast mythology. The post serves as a window into her intellectual concerns, highlighting systemic issues with algorithms, institutional corruption, and scientific misinformation. She briefly editorializes on each link, expressing particular concern about opaque algorithms in criminal justice and skepticism about Facebook's accountability.

Key Insight: Higher education has evolved from knowledge transmission into a complex system serving institutional revenue, employer convenience, and social sorting—with actual learning as perhaps its least important function.

O'Neil questions the true purpose of higher education after reading about failing for-profit colleges that enrich founders while saddling students with debt. She notes the proliferation of 'cash cow' MA programs at elite institutions like Columbia, observing that diversity concerns disappear when international students pay full tuition. Rather than accepting education's stated purpose of knowledge transmission, she offers eight cynical alternative explanations ranging from social networking to corporate training cost-shifting. The post suggests the education system has drifted far from its ostensible mission.

Key Insight: Government recognition that algorithms are not objective and require external accountability mechanisms represents a significant step toward addressing algorithmic discrimination.

The post reviews a White House report on big data and civil rights, authored by top Obama administration officials including CTO Megan Smith and Chief Data Scientist DJ Patil. O'Neil praises the report for debunking the myth of algorithmic objectivity and addressing problems of biased training data, opacity, and disparate impact. The report examines case studies in credit, employment, education, and criminal justice. O'Neil is particularly excited by the conclusion's call for algorithmic auditing and external testing of big data systems, which aligns with her own professional aspirations after her book tour.

Key Insight: When tech platforms make moral judgments about what ads to allow, they become de facto regulators with more power and speed than government agencies—a responsibility they're only beginning to acknowledge.

Google announced it will ban payday loan advertisements through AdWords, requiring lenders to show APR of 36% or less and loan terms of at least 60 days. O'Neil sees this as a positive step but raises questions about enforcement, given payday lenders' history of finding loopholes around regulations. She notes that Google is effectively becoming a quasi-regulator, leapfrogging agencies like the CFPB. The move represents a significant judgment call beyond banning obviously illegal products, suggesting Google is beginning to acknowledge responsibility for vulnerable users in its advertising ecosystem.

Key Insight: The bias in algorithms isn't eliminated by removing humans—it's inherited from them, since machine learning systems are trained on human decisions and perpetuate those biases indefinitely.

Cathy O'Neil argues that the Facebook trending news controversy reveals a fundamental misunderstanding about algorithms: people wrongly assume algorithms are neutral while human curators are biased. She points out that Facebook's human curators were explicitly training the algorithm, meaning any bias they introduced would be perpetuated by the machine learning system. The post challenges the naive belief that replacing humans with algorithms removes bias, asserting that algorithms are just as biased as their human creators.

Key Insight: The terminology we use to search for things matters—'riddles' and 'logic puzzles' return very different results despite seeming synonymous.

Cathy O'Neil makes a personal appeal to her readers for help finding logic puzzles suitable for her 7-year-old son, who has developed a nightly riddle habit after discovering the classic wolf-sheep-cabbage river crossing problem. She notes that searching for 'riddles' online yields wordplay-based puzzles rather than the logic-based challenges she's seeking. The post ends with her discovering that 'logic puzzles' is the better search term, sharing a useful resource she found.

Key Insight: When irrational people invoke terrorism to harass others, the story isn't about what they feared—it's about their irrationality.

O'Neil pushes back against the viral framing of a story where an economist was questioned on a plane for writing equations. She argues the real story isn't about math being under attack, but about an irrational passenger who reported a man for doing math while appearing 'swarthy.' The airline's response of removing and questioning the economist was standard procedure once terrorism was invoked, but the instigating complaint was absurd. O'Neil criticizes the media's framing while noting her New York-honed instinct for recognizing irrational behavior in public.

Key Insight: Forced arbitration clauses are a legal mechanism that corporations use to immunize themselves from accountability by making it economically irrational for individuals to seek justice.

Forced arbitration clauses in corporate contracts prevent consumers and employees from suing companies or joining class-action lawsuits, creating a system heavily stacked against individuals. The cost-benefit calculation of pursuing arbitration rarely makes sense for individuals, and the arbitration process itself favors corporations since businesses often choose the arbitrators. The CFPB has issued new rules to eliminate these clauses for consumer financial lending products, which is significant given how prevalent they are (up to 99% in payday loans). However, the rules don't cover nursing homes, for-profit colleges, or employment contracts, leaving significant gaps in consumer protection.

Key Insight: Opposing specific exploitative trade deals like the TPP shouldn't mean abandoning the goal of international cooperation—instead, we need to reimagine globalization in service of the many rather than the few.

This post promotes the Alternative Banking Group's essay arguing for a reimagined approach to globalization. The central thesis is that while the Trans Pacific Partnership represents harmful pro-corporate globalization, the concept of international cooperation itself shouldn't be rejected. O'Neil and her group advocate for trade agreements designed to benefit the 99% rather than corporate interests. The post announces both the Huffington Post essay and an upcoming Left Forum panel discussion on the topic.

Key Insight: The best popular mathematics writing dares to ask readers to engage with real formulas and technical concepts rather than dumbing things down, but must also consider whose stories get told.

O'Neil reviews 'Prime Numbers and the Riemann Hypothesis' by Barry Mazur and William Stein, praising it as an accessible yet rigorous introduction to deep mathematical concepts. The book combines elegant theoretical explanations with computational illustrations, covering topics from basic prime number theory to the Riemann zeta function. She particularly admires that the authors break the 'third rail' taboo of asking general audiences to engage with formulas and technical work. However, she critiques the book's heavy focus on white male mathematicians, noting the stereotype threat this poses for diverse readers.

Key Insight: The Eurozone's crisis stems not from individual countries' fiscal irresponsibility but from a fundamentally flawed institutional design that treated monetary union as self-evidently good rather than something requiring democratic structures and fiscal mechanisms to actually work.

Cathy O'Neil reviews Yanis Varoufakis's book on Europe's economic crisis, highlighting his argument that the Eurozone's fundamental design flaws make Greek debt essentially unpayable without restructuring or political union. She traces how the monetary union was built on romantic ideals rather than sound economics, creating an undemocratic system where technocrats override popular will during crises. The review emphasizes Varoufakis's critique of German dominance in European monetary policy and his colorful characterizations of bailout politics. O'Neil concludes with her own prediction that Greek default and Eurozone fragmentation are likely within three years.

Key Insight: The pre-publication period offers authors a brief, curated window of positive feedback before facing the full spectrum of public criticism that provocative work inevitably attracts.

Cathy O'Neil announces that her book 'Weapons of Math Destruction' is entering the pre-publication phase, with blurbs being solicited from carefully selected readers. She reflects on the psychology of this moment—enjoying positive early feedback while bracing for inevitable criticism when the book reaches a wider audience. She also announces a book tour starting in September 2016, primarily on the East Coast with possible stops in San Francisco and the Midwest, and solicits suggestions from readers for venues and communities to visit.

Key Insight: Even serious economic commentators can be fans of the public intellectuals they cover, and acknowledging that bias openly is more honest than pretending to pure objectivity.

Cathy O'Neil announces that economist Yanis Varoufakis will be a guest on the Slate Money podcast, where she'll have the opportunity to interview him about his new book on the European economic crisis and its implications for America. She mentions they'll likely discuss timely issues including Puerto Rico, Brexit, and the Greek economy. The post is notably informal and playful, with O'Neil openly admitting to having a 'celebrity crush' on Varoufakis and soliciting questions from readers who might be 'more level-headed' than she is.

Key Insight: Academic research credibility is being systematically corrupted by industry money, with lobbyists literally writing passages in supposedly independent papers to influence state-level regulation.

Cathy O'Neil exposes how academic researchers across multiple disciplines accepted money from the Consumer Credit Research Foundation, a payday lending lobbying group, in exchange for allowing industry lawyers to edit their research papers. The investigation, conducted by Freakonomics through FOIA requests, reveals that one economist allowed CCRF lawyers to write entire paragraphs of pro-payday lending propaganda that appeared verbatim in his published work, despite claiming the organization had no editorial control. This compromised research was then used to lobby state legislatures, where most payday lending regulation occurs. O'Neil argues this undermines public trust in academic objectivity and creates additional obstacles for regulators like the Consumer Financial Protection Bureau.

Key Insight: The sharing economy's reputation systems don't build genuine trust—they manufacture compliance through precarity, transforming workers into surveilled servants who must perform friendliness to avoid algorithmic punishment.

Cathy O'Neil reviews Tom Slee's book 'What's Yours Is Mine,' which critiques the sharing economy's gap between feel-good marketing and exploitative reality. Slee uses data analysis to expose how companies like Uber and AirBnB systematically mislead the public about their business models while skirting regulations and labor protections. The book examines how reputation systems function as surveillance and control mechanisms rather than trust-builders, and how these platforms erode various forms of commons. O'Neil particularly highlights Slee's argument that the sharing economy represents a 'pattern of betrayal' where ideals of community and sustainability have been captured by venture capital and Wall Street.

Key Insight: Data analysis could help identify and reduce racist policing patterns, but implementation at the systemic level is more promising than giving individual officers real-time algorithmic tools that could be gamed.

The post examines a controversy at Columbia where a machine learning assignment about Stop and Frisk data was poorly framed, sparking student protests. O'Neil argues this incident highlights the urgent need for CS departments to teach ethics alongside technical skills. She then explores a Stanford paper proposing algorithms to detect and reduce racist policing practices, while expressing skepticism about real-time algorithmic interventions. She suggests precinct-level analysis of hit rates might be more effective than equipping individual officers with algorithms, concluding that while data can help combat racist practices, the implementation details remain unclear.

Key Insight: Data reveals that Chicago's policing crisis isn't just about individual bad actors but about a systematically broken accountability structure where union contracts, oversight failures, and institutional bias work together to shield misconduct.

O'Neil analyzes the Chicago Police Department Task Force Report, highlighting data that reveals systematic racial bias in policing. The report documents how young black males are stopped at rates nearly equal to their total population, while whites are only stopped when contraband probability is much higher. Despite massive numbers of complaints against officers, 40% go uninvestigated and only 7% are sustained. The post excerpts key findings showing that community distrust of CPD is justified by substantial evidence, the police accountability system is fundamentally broken, and collective bargaining agreements have codified the 'code of silence' into official policy.

Key Insight: A lighthearted announcement marking a personal milestone in the publication of what would become an influential critique of algorithmic bias.

Cathy O'Neil announces that advance review copies (galleys) of her book 'Weapons of Math Destruction' are now available, ahead of the September 6th publication date. She shares a photo of the proof copies she received and jokes about immediately celebrating with drinks. The post serves as both a book promotion and a lighthearted personal update, with links to pre-order from Amazon and Barnes & Noble.

Key Insight: If the tech industry insists on marketing AI as intelligent and human-like, it forfeits the right to complain when people hold those systems to human standards of accountability for discriminatory outcomes.

O'Neil addresses the debate about whether algorithms can be 'racist,' acknowledging the technically valid point that algorithms merely reflect their training data. However, she argues that since the tech industry aggressively markets AI as human-like and intelligent, it's fair game to describe algorithmic outputs in human terms like 'racist.' She concludes that regardless of terminology, companies like Google bear responsibility for monitoring and correcting discriminatory outputs, even when there's no racist intent in the engineering.

Key Insight: The barrier to breaking up big banks isn't a lack of policy options—it's the absence of political will to pursue any of them.

O'Neil argues that breaking up big banks is not a technical problem but a political one. She outlines numerous policy mechanisms that could reduce bank size, from progressive capital requirements and trading restrictions to outright nationalization. After nearly five years attending Alt Banking meetings, she concludes that the real obstacle isn't a lack of viable plans but a lack of political will to implement them. She defends Bernie Sanders against criticism that his bank breakup plans lack specifics, arguing that political will is the crucial missing ingredient.

Key Insight: Free online tools like Brilliant, Desmos, and WolframAlpha can significantly enhance math learning by providing challenging problems and powerful computational aids that go beyond traditional classroom resources.

This guest post by Maxwell Feiner, a high school student, introduces three free online math tools he uses regularly. Brilliant.org provides challenging problems that require deeper thinking beyond standard formulas. Desmos offers an interactive graphing calculator with powerful features like adjustable variables and animations. WolframAlpha serves as a computational engine for graphing and solving various mathematical queries. The post includes practical examples and tutorials for each tool.

Key Insight: Technologies marketed as conveniences or preferences often mask surveillance mechanisms and ignore the economic coercion that drives adoption.

This post offers four quick observations spanning consumer marketing, surveillance, workplace gamification, and autonomous vehicles. O'Neil highlights an upcoming conference on racial bias in government surveillance while dismissing a conference that uses gamification rhetoric to extract data from workers. She critiques how 'millennial preferences' are really just adaptations to economic precarity. The post concludes with concerns about self-driving cars potentially enabling wasteful vehicle usage when cars can run errands without passengers.

Key Insight: Letting a private corporation make de facto encryption policy—even when we agree with the outcome—is anti-democratic and sidesteps the urgent public conversation Americans need to have about digital privacy rights.

O'Neil argues that while the FBI eventually accessed the San Bernardino shooter's phone through an expensive, unscalable technical process, the real issue remains unresolved. No legal precedent was set, and Apple continues to strengthen encryption by default. She contends that framing this as an 'Apple victory' misses the point: a private corporation is now making de facto privacy policy for Americans without democratic input. O'Neil calls for a genuine public conversation about what types of data the government should access, rather than leaving these crucial decisions to corporate discretion.

Key Insight: Well-intentioned metrics inevitably become punitive accountability tools that blame individuals for systemic problems, and grit scores for children will be no exception.

O'Neil traces the predictable lifecycle of metrics in education reform: from well-intentioned measurement to punitive accountability system that fails to address the original problem. She uses standardized testing as her primary example, showing how tests meant to identify struggling schools instead became tools to punish them. Now she sees the same pattern emerging with 'grit' metrics, where Angela Duckworth's research on resilience is being transformed into scorecards for children. O'Neil warns that despite Duckworth's own cautions about misuse, the metric will likely follow the same destructive trajectory. The post is a preemptive strike against yet another measurement system that will blame individuals for systemic failures.

Key Insight: Environmental injustice doesn't require racist intent—it results from systems that strip vulnerable communities of democratic participation while failing to protect their health.

O'Neil reviews the Flint Water Advisory Task Force Report, which she finds compelling in its assignment of blame to Michigan's state government and emergency management system. She highlights how the report frames the crisis as environmental injustice—not requiring malicious intent, but resulting from systems that failed to protect a predominantly Black, impoverished community. The report draws parallels to historical medical abuses of African Americans, including the Tuskegee syphilis study. O'Neil endorses the report's conclusion that emergency management prioritized fiscal concerns over public health, stripping residents of democratic participation in decisions affecting their lives.

Key Insight: This brief post serves as a personal interlude in O'Neil's typically analytical blog, showcasing the human side behind the data scientist persona.

This is a brief introductory note by Cathy O'Neil announcing a guest post written by her 7-year-old son Wolfie about his broken leg. The post itself contains minimal content beyond the introduction, serving primarily as a framing device for her son's contribution. O'Neil mentions Wolfie's recent accomplishment of learning to bike.

Key Insight: Tech companies frame new authentication features as conveniences, but they're actually mechanisms for extracting valuable biometric data that expands corporate and government surveillance capabilities.

O'Neil argues that data, not money, is the true currency of the digital age, explaining how Google, Facebook, and Amazon dominate their markets by controlling vast datasets about users. She examines Amazon's new patent for selfie-based payment authentication as a strategic move to acquire facial recognition data that Facebook already possesses. The post connects this corporate data grab to broader surveillance implications, warning that widespread facial recognition could make location tracking ubiquitous. O'Neil concludes with a rare call to action, urging readers to resist this particular form of data collection before it becomes normalized.

Key Insight: When teenagers shut down or refuse to engage, they're often responding rationally to parents who can't tolerate being challenged—the problem isn't difficult teenagers but difficult parents.

Cathy O'Neil pushes back against a parenting advice column that frames teenager communication styles as the problem in family conflicts. She argues that teenagers who 'refuse to engage' are often responding rationally to parents who have previously shut them down for expressing opinions. O'Neil contends that difficult teenagers are frequently the product of difficult parents who can't handle being challenged. She celebrates teenagers as brutally honest critics and advises parents to listen rather than dismiss.

Key Insight: The NYC luxury real estate bubble, inflated by easy credit and speculation rather than actual demand, is poised to burst in ways that could finally force a reckoning about housing priorities for ordinary New Yorkers.

Tom Adams presented to the Alt Banking group his analysis of the imminent collapse of NYC's luxury real estate market. Drawing parallels to the oil industry's oversupply crisis, he explains how easy credit fueled overdevelopment of luxury condos without regard for actual demand. The market is particularly vulnerable because ultra-luxury real estate isn't a necessity commodity but more like art, dependent on a finite pool of ultra-wealthy buyers who have largely already made their purchases. With thousands of units hitting the market and sales declining, developer defaults are beginning and the bubble appears ready to burst. The crash could ultimately benefit ordinary New Yorkers by forcing a conversation about housing priorities.

Key Insight: Free trade has been framed as benefiting everyone through lower prices, but this framing deliberately obscures that American workers—not just consumers—bear real costs that fuel dangerous political backlash.

Cathy O'Neil argues that free trade agreements have been sold to Americans by emphasizing cheaper consumer goods while ignoring the devastating impact on American workers, jobs, and wages. She endorses Jared Bernstein's critique that globalization treats Americans as consumers rather than workers. While acknowledging that free trade benefits workers in developing countries like China and Bangladesh, she frames it as a win-lose rather than win-win situation for American workers. She calls on economists to honestly acknowledge the damage done to American workers and warns that failure to address these grievances fuels support for Trump and similar politicians.

Key Insight: Unconscious biological drives can shape our social behavior and judgments in ways we don't recognize until those drives disappear.

O'Neil recounts refusing to switch airplane seats so an orthodox Jewish man wouldn't have to sit near women, connecting this incident to a broader personal transformation. She describes how her recent loss of desire to have more children fundamentally changed her relationship with men—she no longer unconsciously accommodates them or forgives offensive behavior from attractive people. The piece is a candid reflection on how biological drives she wasn't even aware of had been shaping her social interactions for years. She concludes that this hormonal shift has made her more clear-eyed about people, judging them by what they say rather than their charm or attractiveness.

Key Insight: Auditing algorithms for bias is technically possible but practically constrained by the frequent absence of protected attributes in the very datasets we need to examine.

O'Neil announces her upcoming talk at QCon London on algorithm auditing, using recidivism modeling as a timely case study. She notes the irony that the talk preceding hers will discuss the very type of algorithmic system that desperately needs auditing. The post surveys current tools for algorithmic auditing, including the FairTest toolkit, while highlighting a critical gap: these tools struggle when protected attributes like race and income aren't explicitly included in datasets. She suggests proxy methods using zip codes and names to infer race, but acknowledges the significant limitations.

Key Insight: Male discomfort with menstruation creates policy blind spots that harm women, and eliminating the tampon tax is just the first step toward treating menstrual products as the basic necessity they are.

O'Neil announces her role as lead plaintiff in a lawsuit challenging New York State's tampon tax, noting Governor Cuomo's quick response supporting repeal. She argues that male squeamishness about menstruation is precisely why such unfair policies persist. While celebrating the tax lawsuit as progress, she pushes further: tampons should be freely accessible like toilet paper, especially for poor and homeless women who can't even buy them with food stamps. The post uses humor and directness to highlight how absurd it is that a basic hygiene product is taxed as a luxury and unavailable in most public restrooms.

Key Insight: Sometimes challenging systemic unfairness requires moving beyond analysis and commentary to direct legal action.

Cathy O'Neil announces her role as lead plaintiff in a class-action lawsuit challenging New York State's sales tax on tampons and other menstrual products. She frames the tax as discriminatory since it treats essential hygiene products as non-necessities. The suit was developed by her friend Laura Strausfeld, a Columbia Law School graduate who crafted the legal theory. This brief announcement signals O'Neil's willingness to move from critique to direct legal action on issues of gender equity.

Key Insight: When we rely on corporations to protect our rights instead of demanding democratic accountability from government, we've already lost the more important battle.

O'Neil argues that the Apple vs. FBI encryption debate is fundamentally misdirected. Rather than celebrating Apple as a privacy hero standing up to government overreach, we should be asking why our government isn't protecting our privacy in the first place. Apple is a profit-seeking corporation accountable to shareholders, not citizens, and will comply with laws when forced. The real conversation should be about demanding democratic accountability from government institutions that are making unilateral decisions about privacy trade-offs without public input.

Key Insight: Mathematical truth and precision are not synonymous - a statement can be imprecise yet true, raising fundamental questions about what it means to be 'right' about anything.

Cathy O'Neil enthusiastically recommends her friend Ernie Davis's book 'Mathematics, Substance and Surmise,' a collection of essays exploring the ontology of mathematics from multiple perspectives. She admits she hasn't finished it yet but couldn't contain her excitement. The book covers topics ranging from mathematical collaboration to computer-assisted proofs to the fundamental question of whether mathematical objects exist. O'Neil connects it to her own philosophical musings about truth and precision, using her provocative claim that 'the sun goes around the earth' as an example of how statements can be imprecise yet true. She closes with a genuine invitation for the book's contributors to come discuss mathematics at her home.

Key Insight: Market-based solutions cannot solve problems rooted in inequality because private interests will always prioritize profitability over comprehensive social good.

The post argues that neoliberalism, the dominant ideology of the past 40 years, is finally being challenged on multiple fronts. O'Neil points to the failures in Flint and Ferguson as natural consequences of neoliberal policies that blame individuals for systemic failures. She finds hope in the failure of social impact bonds, voter rejection of corporatist politics, and the repeal of No Child Left Behind as signs that people are recognizing the limits of market-based solutions to social problems. The post concludes by contrasting two possible futures: continued decline or governments reclaiming their responsibilities from market forces.

Key Insight: The 'lovable bigot' archetype that was supposed to be satirical has always reflected a real and powerful political constituency that mainstream culture repeatedly underestimates.

O'Neil recounts a dream about Archie Bunker being elected president, drawing an implicit parallel to the 2016 election cycle. She explores the paradox that Norman Lear created Archie Bunker to be disliked, yet audiences embraced him as lovable and relatable. The post examines how views considered 'laughably old-fashioned' in the 1970s have resurfaced in mainstream politics. O'Neil questions whether the 'Archie Bunker for president' concept was ever truly a parody, suggesting it reflected genuine political sentiment all along.

Key Insight: Even someone who critiques algorithmic metrics for a living can't resist obsessing over their own Amazon ranking, revealing the universal human vulnerability to quantified validation.

Cathy O'Neil shares two pieces of personal news: the release of her book cover for 'Weapons of Math Destruction' and an upcoming live performance with her bluegrass band. She playfully obsesses over her book's Amazon ranking despite promising herself she wouldn't, finding humor in the absurdity of ranking metrics. The post serves as both a book pre-order announcement and an invitation to see her band play at a Brooklyn venue.

Key Insight: Sometimes the most compelling characters are the ones who simply exist in a space without demanding attention or advancing plots.

O'Neil shares her family's experience rewatching Star Trek: Deep Space Nine and their fascination with Morn, a background character who appears consistently but never speaks. She appreciates how DS9 differs from other Trek series by featuring non-Federation characters without military training, providing relief from the sometimes excessive Academy worship. The post celebrates the show's civilian characters and culminates in her discovery of the Morn Watchers fan club, which she playfully offers to restart.

Key Insight: Mathematical ability is equally distributed across all populations, but access to mathematical opportunity is not—making math education both a symptom of inequality and a uniquely powerful lever for addressing it.

Dr. Mark Tomforde describes CHAMP, a math outreach program at the University of Houston that serves predominantly Black and Hispanic students from Third Ward and Sunnyside—among America's most dangerous and impoverished neighborhoods. The program brings high school students to campus twice weekly for Math Circle-style lessons and SAT/ACT tutoring, while connecting them with college student mentors who share similar backgrounds. Tomforde argues that mathematics is uniquely positioned as an equalizer because math skills are critical gatekeepers for high school graduation, college admission, and STEM success. He emphasizes that mathematical ability is equally distributed across populations, but opportunity is not—meaning countless potential Einsteins are lost to poverty and under-resourced schools. The piece calls for universities to break out of their bubbles and engage surrounding communities, framing this not as charity but as central to their educational mission.

Key Insight: Making math enrichment accessible requires removing financial barriers entirely and meeting students in their own communities, not expecting them to come to elite institutions.

P.J. Karafiol describes Math Circles of Chicago, a free math enrichment program he founded in 2010 to address elitism in math education. The program now serves 500 students across five locations in Chicago, including some of the city's poorest neighborhoods, focusing on collaborative problem-solving rather than lectures or competitions. The organization operates on a donor-supported 'NPR model' with no financial barriers to entry—just a handwritten essay about why students want to participate. Karafiol argues that for many students, math circle provides their only exposure to 'real mathematics' as opposed to arithmetic and rote procedures taught in schools.

Key Insight: College ranking algorithms create perverse incentives that lead institutions to harm students in order to game metrics, and the same survey tools can be weaponized against the very people they claim to help.

Mount St. Mary's president Simon Newman, a former private equity executive, fired two professors for disloyalty after they exposed his plan to manipulate college rankings. Newman wanted to 'cull' 20-25 freshmen before the enrollment deadline by using a student survey disguised as a personal motivation assessment. The goal was to boost the retention rate by 4-5% to improve US News & World Reports rankings. The scandal reveals how algorithmic ranking systems create perverse incentives that harm students, and raises questions about how widespread such gaming practices might be at other institutions.

Key Insight: America's math enrichment boom is widening inequality rather than closing it, with the benefits flowing almost exclusively to already-privileged children.

O'Neil highlights the explosion of math enrichment programs for young people while pointing out their fundamental accessibility problem. She cites Atlantic data showing the U.S. has an 8-to-1 ratio of rich to poor math whizzes, far worse than other developed nations. The gains in advanced math achievement are almost entirely captured by children of the highly educated. She poses the challenge of how to democratize these programs when qualified teachers have lucrative alternatives. O'Neil suggests this may be a transitional problem that solves itself once enough 'fun math' speakers exist to train future generations.

Key Insight: Quality writing about aging can inspire deeper thinking about how technology intersects with elderly populations.

This is a brief recommendation post where the author shares a new blog called 'Time for One More' that she finds inspiring. She connects it to her own ongoing project examining the relationship between elderly people and technology. The post serves primarily as a pointer to external content rather than presenting original analysis.

Key Insight: Ethics education must become a standard component of data science training, not an afterthought.

O'Neil announces her Slate essay 'The Ethical Data Scientist' which argues for bringing better ethics to data science practice. She highlights that ethics education should be integrated into data science training programs. She points to Berkeley's School of Information as a model, noting their undergraduate and graduate courses on data science ethics. The post serves as both promotion for her longer piece and a call for readers to share other ethics education initiatives in the field.

Key Insight: Algorithms aren't inherently fair, but unlike human decision-making, their unfairness is at least measurable and auditable—an advantage we're failing to exploit.

O'Neil challenges the common misconception that algorithms are inherently fair simply because they're mathematical, using a 'resting bitch face' detection algorithm as an example of hidden bias. However, she argues that algorithms have one significant advantage over human decision-making: they're auditable. Using the example of wildly inconsistent bankruptcy court rulings, she suggests that algorithmic processes, while not automatically fair, could at least be systematically tested and improved. She concludes that we urgently need to start auditing high-impact algorithms rather than blindly trusting them.

Key Insight: Effective algorithmic accountability requires bridging the gap between what technologists can build and what regulators know to ask for.

O'Neil announces her upcoming talk at Harvard's Data Privacy Lab, where she'll present a technical version of her 'Weapons of Math Destruction' thesis and propose technological tools for auditing opaque algorithms. She emphasizes measuring algorithm outputs rather than scrutinizing source code as an approach to understanding black-box systems. The post highlights her interest in meeting Latanya Sweeney, who as FTC chief technologist can advise on what tools regulators might actually adopt. O'Neil expresses a practical concern about developing useful tools rather than academic exercises.

Key Insight: The Flint crisis wasn't a failure of infrastructure but a failure of democracy—when you remove elected accountability from governing decisions, the people who suffer are invariably the poor.

O'Neil argues that the Flint water crisis was not bad luck but a direct result of suspending democracy through emergency management, which disproportionately affects poor communities of color. She criticizes media coverage that frames corporate water bottle donations as solutions while ignoring systemic causes. The real issue is decades of neglected water infrastructure, with federal funding cut by 80% since 1977 and an estimated $1 trillion needed for updates. O'Neil contends that infrastructure neglect functions as a punitive tax on the poor, and without addressing these root causes, similar crises will occur nationwide.

Key Insight: Teaching children ethics and trusting them to find their own passions produces better outcomes than forcing achievement-oriented activities that confuse correlation with causation.

O'Neil endorses Adam Grant's NYT column arguing that forcing children into structured practice confuses correlation with causation—just because musicians practiced doesn't mean forced practice creates musicians. She advocates instead for teaching ethics and letting children find their own passions, noting that kids forced into music lessons rarely continue playing as adults. Drawing on her own experience growing up in competitive Lexington, Massachusetts, she observes that the most pressured children were often miserable and abandoned their instruments. She concludes by citing a Harvard study showing 80% of youth believe their parents prioritize achievement over caring, and argues that her approach of not forcing piano practice is precisely why her children understand what actually matters.

Key Insight: Algorithms that sort and classify vulnerable populations like the unemployed operate with opacity and immunity from transparency laws, making them particularly dangerous.

O'Neil reports from the CPDP conference in Brussels, where she's learning about privacy and data protection issues. She attended a panel on privacy and the world's poor, where she discovered an algorithm in Poland that sorts unemployed people—a system that is both opaque and high-impact, with information laws mysteriously not applying. She notes the stark contrast between European privacy protections and the weaker American framework, and expresses enthusiasm for learning how Europeans approach data protection enforcement.

Key Insight: Well-intentioned educational technology still creates dangerous surveillance infrastructure over vulnerable communities when data protection laws are inadequate.

O'Neil critiques Kinvolved, a new app that tracks Harlem schoolchildren's attendance and sends alerts to parents. While the app may have good intentions and could help some families, she argues the data it generates about poor Black and brown children is inadequately protected. She highlights that the user agreement can be changed at any time and that bankruptcy could lead to data being sold to brokers. FERPA's loopholes allow such apps to access education records without meaningful constraints on future use.

Key Insight: Sometimes data visualizations are so powerful they require no additional commentary to make their point about racial inequality.

This extremely brief post directs readers to a Bloomberg article featuring two graphs about wealth and race. O'Neil lets the data visualizations speak for themselves rather than providing her own analysis. The post implies the graphs are self-evidently powerful in illustrating racial disparities in wealth accumulation. By describing the article as 'amazing' and the graphs as 'fantastic,' she signals strong endorsement of the piece's findings.

Key Insight: Practical data analysis techniques can transform unwieldy datasets into accessible insights that challenge political narratives, as Schneider's 'medium data' approach revealed Uber may be displacing taxis rather than adding traffic.

O'Neil recaps a meetup by data analyst Todd Schneider, who analyzed 1.1 billion NYC taxi and Uber trips using what he calls 'medium data' - datasets that require creative solutions but can still run on personal computers. She highlights his practical techniques for handling large datasets, including using external storage and reducing geographic granularity. The post also surfaces an intriguing finding: Uber appears to be taking market share from taxis rather than increasing overall rides, which contradicts claims that Uber increases traffic congestion.

Key Insight: Demanding inclusion in corrupt institutions validates those institutions—sometimes the more radical position is to reject the premise that attendance equals progress.

O'Neil responds to a New York Times article lamenting the low percentage of women at Davos by arguing that women's absence from the World Economic Forum is actually something to be proud of, not ashamed of. She characterizes Davos as a 'dick-measuring contest' where elites gather in luxury to superficially discuss poverty while primarily advancing their own interests. Rather than pushing for gender parity at the event, she suggests the more beneficial action would be to cancel Davos entirely, as the forum represents and perpetuates the very inequality it claims to address.

Key Insight: Tax policy that makes holding empty luxury apartments cheap transforms housing from shelter into an asset class, and only higher taxes can restore housing's primary purpose of being lived in.

O'Neil argues that New York City's housing market has split into two competing segments: one for people who need homes and one for wealthy investors treating apartments as assets. The luxury investment market is winning, driven by foreign money laundering, global inequality, and perverse tax incentives that make holding empty apartments nearly frictionless. She proposes dramatically increasing real estate taxes on luxury properties, especially empty ones, to shift incentives back toward housing people rather than storing wealth. This would force wealthy owners to sell to actual residents, bringing prices down and making the city livable again.

Key Insight: Technical complexity is often blamed for bureaucratic delays that are actually caused by political or institutional considerations.

O'Neil debunks the myth that the NYC SHSAT school matching algorithm is computationally complex, arguing the real delay is political rather than technical. She and her husband devised a simple algorithm over lunch that could run in minutes once tests are scored. The actual complexity lies in the non-SHSAT public school matching system, not the exam-based placements. She suspects the delay exists to prevent the SHSAT results from creating unfair competitive pressure against the regular public school matching process.

Key Insight: Even data scientists immersed in algorithmic critique need moments of pure family joy and silly songs to balance the stress of serious work.

This is a light personal post where O'Neil shares a moment of stress relief during final book edits by enjoying music with her three sons. She highlights songs by Rob Cantor as family favorites and mentions wanting to see him perform live. The post culminates in sharing the 'O'Neil/de Jong family anthem' - a Songify This video. It's a rare purely personal glimpse into her family life, stepping away from her usual algorithmic critique.

Key Insight: When governments invest in both public amenities and surveillance infrastructure, expect the surveillance to be maintained while the amenities decay.

Cathy O'Neil comments on a Cuomo administration announcement about technological upgrades to NYC subways, including mobile tickets, wi-fi in stations, surveillance cameras on all trains and buses, and USB charging stations. She expresses skepticism about the priorities and longevity of these upgrades, suggesting that surveillance infrastructure will be maintained far better than amenities meant to serve the public.

Key Insight: Michigan's Emergency Manager law disproportionately strips democratic representation from Black communities, as demonstrated by mapping receivership cities against census racial demographics.

O'Neil investigates Michigan's 'Emergency Manager' law, which allows state-appointed officials to replace elected local governments in financially distressed municipalities. Prompted by Rachel Maddow's coverage of the Flint water crisis, she maps which Michigan cities are under receivership. By overlaying these locations with census data showing Black population distribution, she reveals a striking visual correlation. The cities stripped of their elected officials are predominantly Black communities, suggesting a racial dimension to this 'suspension of democracy.'

Key Insight: The weight loss industry profits not from delivering results but from exploiting the gap between statistical reality and individual hope, sustained by a culture that moralizes weight and rewards the appearance of trying.

The $64 billion weight loss industry thrives despite no evidence that diets work beyond 12 months. O'Neil argues that people continue paying for ineffective programs due to three psychological mechanisms: the belief they'll be the statistical exception, the entertainment value of fantasizing about a 'new thin self,' and the moral approval society bestows on dieters. The industry exploits these vulnerabilities, with the final phase being guilt when people inevitably fail—leading them to keep paying for nothing. She draws parallels to lottery ticket purchases and gym memberships as examples of 'quasi-moralistic self-help fads' that extract money through hope rather than results.

Key Insight: Unlike academic work where publication is the finish line, book publishing transforms the author's role from researcher to messenger, making the real work begin only after the writing ends.

Cathy O'Neil announces that her book 'Weapons of Math Destruction' is now available for pre-order and reflects on how book publishing differs from her previous work in mathematics and data science. She notes that while academic papers and data science projects end once published, a book's publication is just the beginning of spreading its message. She also announces hiring a booking agent to manage the growing volume of speaking requests, acknowledging this may surprise academics accustomed to free speaker availability.

Key Insight: The outrage over randomly assigned race in video games reveals that white identity is treated as the invisible default, while any other identity is seen as an imposition.

The post examines the backlash when the video game Rust changed from giving all players white avatars to randomly assigning race. White players complained vocally about potentially being assigned black avatars, despite minorities having played as white characters without complaint for years. O'Neil notes the irony that white players would accept being a green alien but not a black man, and suggests games could further challenge players by including experiences like menstruation or pregnancy.

Key Insight: Sometimes the most effective parenting intervention is removing pressure and trusting children to find their own path through difficult challenges.

O'Neil shares a personal story about her youngest son's recovery from a broken leg, where the physical healing was quick but the psychological barrier to walking again proved far more challenging. After weeks of frustration, conflicting medical advice, and a misdiagnosis of 'hysterical paralysis,' she finally gave up the intense pressure and took her son on a relaxing swimming trip. Once the stress was removed, he simply decided to walk again. The experience taught her that sometimes the best parenting means stepping back and trusting children to overcome obstacles on their own timeline.

Key Insight: If we're going to use algorithmic tools of social control against teachers and civilians, intellectual consistency demands we also build accountability systems that scrutinize police behavior—especially their use of violence.

O'Neil argues that while black box algorithms are increasingly used as tools of social control against teachers and civilians (particularly young men of color), there's a glaring absence of similar accountability mechanisms for police. She notes that existing police metrics like CompStat focus only on arrests and tickets, ignoring data on police violence. The post was prompted by the Tamir Rice case, where a prosecutor refused to 'second-guess' officers who killed a 12-year-old. O'Neil contends that our priorities are backwards when we scrutinize teachers more than police who kill children.

Key Insight: Good probability forecasts should be judged not just on accuracy but on their willingness to make confident predictions, since hedging at 50% is mathematically the worst possible strategy.

The post explores what precipitation forecasts actually mean and how to evaluate their accuracy. O'Neil proposes a rigorous method using 101 bins (0-100% probability) to collect forecast-outcome pairs over time, then measuring accuracy by comparing predicted probabilities against actual rain frequencies. She argues that squared error is the ideal penalty metric because it's minimized when forecasts match true probabilities and rewards models that commit to confident predictions near 0% or 100% rather than hedging at 50%. The analysis reveals that a 'dumb' forecast using historical averages would score poorly despite being technically accurate within its single bin, while the ideal forecast would only use bins 0 and 100.

Key Insight: The opacity of modern supply chains makes fraud nearly undetectable, leaving consumers dependent on accidental discoveries rather than systemic accountability.

O'Neil catalogs five notable consumer scams from 2015, arguing that global supply chain opacity enables widespread fraud. From VW's emissions cheating to fake artisanal chocolate to mislabeled yarn, she demonstrates how companies exploit consumers' inability to verify product claims. The common thread is that authentication came from external testing or allergic reactions, not internal accountability. She questions whether we'll develop better authentication systems or simply see scams proliferate.

Key Insight: The Star Wars Holiday Special is so legendarily bad that it can only be properly appreciated either while intoxicated or through the lens of teenagers finding joy in its awfulness.

O'Neil humorously reviews the infamous 1978 Star Wars Holiday Special, describing it as the perfect movie to watch while high due to its inexplicably long, subtitle-free scenes featuring Chewbacca's family. She admits to watching it despite not being in the ideal altered state, finding enjoyment mainly through her teenagers' amusement at how bad it is. The post notes George Lucas's reported desire to destroy every copy of the film.

Key Insight: Major creative projects often succeed precisely because we don't fully understand their scope when we begin—pure ignorance can be a gift that keeps us from being paralyzed by the enormity of the task.

Cathy O'Neil announces that her book 'Weapons of Math Destruction' is nearing its final draft and will be produced as an audiobook by Random House. She reflects on the long, uncertain journey of writing the book and the relief of finally seeing tangible progress. The post captures the emotional arc of completing a major project—from the disorienting uncertainty of research to the satisfaction of a finished product. She expresses enthusiasm about potentially narrating the audiobook herself, having developed an appreciation for how the book sounds read aloud.

Key Insight: Mathematical correctness requires not just logical validity but communicable insight—if experts cannot explain why a proof works to other experts, the burden of proof remains unmet regardless of time invested.

Brian Conrad's notes summarize the 2015 Oxford workshop on Mochizuki's claimed proof of the ABC Conjecture via Inter-universal Teichmüller Theory (IUT). Despite gathering top arithmetic geometers including Faltings and Wiles, the workshop failed to produce the hoped-for 'aha moment' that would clarify how IUT actually proves anything. The first three days provided useful background on pre-IUT material, but the final two days became incomprehensible to the audience, with speakers unable to provide satisfying examples or explain why elaborate constructions were necessary. Conrad concludes that the burden now falls on those who claim to understand IUT to communicate its key insights more effectively, noting that three years and three workshops have not produced the kind of clear exposition that accompanies other mathematical breakthroughs.

Key Insight: O'Neil signals her characteristic skepticism toward Big Data claims in policing, though this post is primarily a podcast announcement rather than substantive analysis.

Cathy O'Neil appeared on NPR's 'To The Point' to discuss predictive policing and whether Big Data can prevent crime. She notes she was the most skeptical among four guests, which included a journalist, a techno-futurist, and a public safety director. The brief post serves primarily to share the podcast link with her readers.

Key Insight: The best advice—whether about math careers or love—comes down to taking informed risks, appreciating what you have, and not overthinking what you can't control.

Cathy O'Neil announces the retirement of her advice column 'Aunt Pythia' after a long run. She provides a condensed summary of her most common advice topics: becoming a data scientist, understanding why math is difficult, dating advice, and encouraging people to act on their feelings. The column ends on a triumphant note with a reader writing back to report that following Aunt Pythia's advice to confess a long-term crush resulted in a successful relationship. The tone is warm, self-deprecating, and celebratory of the column's quirky mix of technical and personal guidance.

Key Insight: Academia's rigid structures and unspoken rules can hospitalize you, but reclaiming your work as something accessible and human—rather than performing for gatekeepers—transforms it from a weapon of escape into something that might actually help others.

Piper Harron reflects on her unconventional mathematics PhD thesis, which she describes as initially intended as a 'grenade' against academia but evolved into something more constructive. She candidly discusses spending 10 years on her dissertation while battling imposter syndrome, feeling behind her peers, and processing the systemic pressures of being a Black woman in mathematics. Her awakening to feminism and anti-racism, catalyzed by Trayvon Martin's death, led her to rewrite her thesis as something accessible to non-mathematicians. She hopes it will show that research math isn't alien and help graduate students feel less alone in their struggles.

Key Insight: Sustained grassroots financial reform efforts like the Alternative Banking group continue their work long after the initial protest movement fades from headlines.

O'Neil promotes her Occupy Wall Street offshoot group, the Alternative Banking group, which continues to meet weekly at Columbia University. The group maintains several ongoing projects including essay series at Huffington Post and Occupy Summer School. She highlights their curated reading list as a valuable resource for quality financial and economic articles. The post serves primarily as an invitation to engage with the group's work and resources.

Key Insight: Everyday objects like house keys encode surprisingly elegant mathematical systems, but understanding the math behind security mechanisms often reveals their vulnerabilities rather than reinforcing confidence in them.

O'Neil discovers that house keys aren't duplicated by physical copying but by reading embedded codes, leading her down a rabbit hole of lock mechanics. She learns her Medeco keys have 6 pins with both depth levels (6 options) and angles (3 options), creating 34 million possible combinations. Her curiosity then turns to master key systems, which work by having multiple slices in each pin. The investigation ends with a security concern: master key systems are potentially vulnerable, and she wonders if Medeco's angled cuts truly add protection or if master keys could still be derived.

Key Insight: The tech industry treats elderly people as subjects to be monitored or exploited rather than as users deserving thoughtfully designed tools that enhance their independence and connection.

O'Neil argues that the tech industry is failing elderly people by focusing on surveillance devices for caregivers rather than tools that enhance communication and quality of life. She uses her father-in-law's hearing difficulties as an example of how existing technology like voice recognition could easily help but isn't being deployed. The 'Silver Tech' industry largely produces monitoring tools that treat elderly people as problems to be managed rather than individuals to be empowered. Meanwhile, the data industry that does focus on the elderly often does so predatorily, selling lists of vulnerable seniors to scammers. O'Neil calls for a shift toward thoughtful, non-predatory technology designed with elderly end-users in mind, especially as Boomers retire with higher expectations for their quality of life.

Key Insight: Career satisfaction comes not from finding a perfect path but from honestly assessing what you enjoy, choosing work with feedback loops that match your temperament, and accepting that every job has frustrations.

This is an advice column where 'Aunt Pythia' (Cathy O'Neil's humorous alter ego) answers reader questions on topics ranging from linguistics to career satisfaction to workplace gender dynamics. A PhD student struggles with enjoying learning math but hating research, prompting advice about patience, choosing faster-moving fields, and finding projects that genuinely excite. Two questions address gender in professional settings: a male professor worried about perceptions of his friendship with a female student, and a manager concerned about socializing more with male than female reports. The column blends irreverent humor with practical wisdom.

Key Insight: Criminal intent requirements in white-collar crime would incentivize corporations to deliberately hide illegal decisions inside opaque algorithms, making accountability impossible.

Cathy O'Neil critiques H.R.4002, a bipartisan criminal justice reform bill backed by the Koch brothers that would require prosecutors to prove criminal intent in federal cases. While acknowledging the need for reform of drug sentencing laws, she argues this bill goes too far by extending protections to white-collar crime. She connects this to her research on algorithmic discrimination, warning that the bill would make it impossible to prosecute those who hide behind opaque algorithms. The post argues this would incentivize corporations to deliberately obscure criminal decisions within algorithms to avoid liability.

Key Insight: Transparency initiatives fail when they assume institutions collect inconvenient data—the real problem is that powerful actors systematically avoid measuring what might hold them accountable.

The post examines a whitepaper about the White House Police Data Initiative, which encourages police departments to share their data publicly. O'Neil argues that the fundamental problem isn't transparency but data collection itself—police departments simply don't gather information that might reflect poorly on them. She points out that even basic data like fatal police shootings lacks a reliable national registry, and information about public trust in police or jail conditions remains systematically uncollected. The piece concludes with skepticism about reform, noting that facilities like Rikers can't even be trusted to accurately count inmate fights.

Key Insight: The big data community is missing out on a growing pool of talent by failing to actively engage underrepresented minority students, and bridging this gap requires intentional outreach rather than passive inclusion.

This guest post by Professor Nii Attoh-Okine argues that underrepresented minority students are conspicuously absent from the big data revolution, despite opportunities across multiple industries. Using his experience at major conferences like Strata and Hadoop, he observes that minority participation as both attendees and presenters is virtually nonexistent. He frames this as both a moral imperative and a practical necessity, noting that underrepresented minorities will soon constitute the majority of school-age children in the US. The post concludes with concrete suggestions for how the big data community can actively engage these students through mentorship, institutional partnerships, and conference inclusion.

Key Insight: Evolutionary explanations for why math is difficult fail because they cannot account for why other survival-irrelevant cognitive tasks like reading fiction or understanding jokes are easy and enjoyable.

This guest post by computer scientist Ernie Davis challenges Brian Greene's evolutionary explanation for why humans find math difficult. Davis argues that claiming math is hard because our brains evolved for survival doesn't actually explain anything, since many other cognitively demanding tasks that were equally useless for cavemen (reading novels, telling jokes, understanding stories) are easy and pleasant for us. He points out that our ancestors encountered parabolic motion and spherical geometry daily yet never saw magic wands, yet we'd rather read fantasy novels than study parabolas. The post concludes that without understanding how the mind performs cognitive tasks, we cannot explain why some are harder than others.

Key Insight: Systemic solutions like blind review processes are more effective than individual guilt or vigilance at combating unconscious bias, because bias is universal and self-awareness alone cannot overcome it.

This is an advice column where Cathy O'Neil, writing as 'Aunt Pythia,' responds to reader questions on topics ranging from unconscious sexism in academic publishing to workplace sexual harassment intervention. The most substantive exchange involves a mathematician who realized years later that he had written a condescending rejection letter because he assumed the author was female. O'Neil validates that everyone harbors sexist biases and proposes concrete solutions: blind review processes for journal submissions, similar to blind auditions in orchestras. She also addresses a bystander who intervened in a sexual assault but used diversion rather than direct confrontation, arguing that while well-intentioned, this approach fails to create consequences or protect future victims.

Key Insight: Even when actively trying to find gratitude for financial systems, the exploitation of vulnerable people keeps undermining the case.

O'Neil struggles to find something in finance to be thankful for on a Thanksgiving podcast episode. She considers consumer loans, inflation, the financial system, and insurance, but finds significant drawbacks to each. Consumer loans raise prices and create crushing debt; inflation is unpredictable; the financial system exploits poor people. She settles tentatively on insurance as the most defensible option, acknowledging it mostly functions as intended to smooth out life's risks.

Key Insight: Charter school performance metrics are misleading unless we account for the fact that they systematically serve fewer students with disabilities and English language learners than traditional public schools.

O'Neil investigates whether charter schools systematically exclude difficult students by analyzing NYC Department of Education data. Using two proxies for challenging student populations—students with disabilities and English language learners—she finds charter schools have significantly fewer of both groups compared to traditional public schools. The statistical analysis, including t-tests with very small p-values, confirms this isn't random chance. While she notes this isn't an indictment of any specific charter school, she argues that performance comparisons between charter and non-charter schools must account for these demographic differences to be meaningful.

Key Insight: The unglamorous, painstaking work of careful editing is what elevates a book above a blog post.

O'Neil provides a behind-the-scenes update on the editing process for her upcoming book 'Weapons of Math Destruction.' She walks through the various stages of book production—from content edits to line editing to galleys to printing overseas. The post serves as both an explanation for reduced blog activity and an honest reflection on why books, despite being 'the opposite of sexy,' benefit from rigorous editing. She announces a tentative September 2016 launch date.