Source

Why doesn’t network difficulty fall?

Bitcoin’s proof-of-work algorithm has a difficulty ratchet to keep production steady. Network difficulty adjusts every two weeks, approximately, based on whether miners have been finding new blocks for the blockchain too quickly or too slowly.

Difficulty tightens when a higher bitcoin price is encouraging miners to join the market, that much makes sense. But when a lower bitcoin price is squeezing margins and pushing the least efficient miners out, shouldn’t difficulty fall?

Source

Maybe difficulty lags behind the price? As said above, the ratchet only adjusts approximately every two weeks. Total hashrate offers a more dynamic measure of power deployed to the network, albeit an estimated one based on how long it’s taking to mine a block. Any change in market structure might show up there first.

And has it? Not really:

Do miners with sunk costs keep running on negative margins in the hope of getting lucky? Are a handful of big miners, maybe advantaged by free power or whatever, keeping difficulty high to drive out competitors, either inadvertently or as part of some devious plan to centralise production and control the network? Or has mining become 55 per cent more efficient since last November?

Source

Related, is it a worry that just three mining pools accounted for more than 45 per cent of this week’s block production? Given nearly all the hardware used across the network is Chinese-made, with Beijing-based Bitmain Technologies alone having an estimated market share of 82 per cent, when do concentration levels among a few organisations become a security concern?

Bitcoin mining is quietly staging a comeback in China despite being banned four years ago, as individual and corporate miners exploit cheap electricity and a data center boom in some energy-rich provinces, according to miners and industry data.

China had been the world's biggest crypto mining country until Beijing banned all cryptocurrency trading and mining in 2021, citing threats to the country's financial stability and energy conservation.

After having seen its global bitcoin mining market share slump to zero as a result of the ban, China crept back to third place with a 14% share at the end of October, according to Hashrate Index, which tracks bitcoin mining activities.

What’s the deal with Tether?

Stablecoins have long been pitched as crypto’s on-ramp. Swapping fiat money for a fiat-pegged stablecoin like Tether’s USDT or Circle’s USDC allows a trader to switch in and out of positions without having to touch tradfi.

Shouldn’t an on-ramp also work as an off-ramp? There’s not much evidence it does. In the six weeks or so when $1.2tn in value was drawn down from the cryptoverse, the market cap of USDT has increased by approximately $20bn:

Source

USDC hasn’t been quite as resilient but over the period is still basically flat:

Source

Even if we assume a large percentage of stablecoins are used for non-crypto things (sports betting, remittances, crimes), the recent issuance still looks at odds with the trend.

Maybe demand is high because crypto traders have been parking money rather than seeking to withdraw it?

Source

More idle money in the system ought to be good news for the likes of Coinbase, which uses the promise of higher yields on USDC deposits to sell monthly subscription schemes. And how have Coinbase shares been doing?

What explains trash crash PTSD?

A popular argument among crypto commentators is that token prices are down because traders are still digesting one bad day in early October.

Reasons for the October drawdown go from banal (maybe the high-beta cryptosphere just amplified an equities pullback on US-China trade tensions?) to wonky (maybe it all cascaded from a weird synthetic stablecoin depegging on one marketplace?) to the darkly cynical (maybe the big sharp drop was to let bucket-shop crypto brokers close out customer positions they’d never actually bought?).

Source

If a market’s not deep, efficient or clean enough to digest a bit of one-day volatility, why get involved?

Where’s the volume?

Crypto exchange-traded products have been haemorrhaging money all week. Spot ETP net redemptions yesterday were $1.14bn, including $901mn just from bitcoin ETPs, according to JPMorgan estimates. That’s the worst single day for net outflows since February.

With so much selling, you might expect to see an increase in bitcoin velocity, which measures the rate at which tokens move on the chain.

Source

90% of transaction volume on the Bitcoin blockchain is not tied to economically meaningful activities but is the byproduct of the Bitcoin protocol design as well as the preference of many participants for anonymity ... exchanges play a central role in the Bitcoin system. They explain 75% of real Bitcoin volume.

Source

Bitcoin velocity has been plummeting for years, for reasonable reasons. “Digital gold” overtook “internet money” as the preferred reason to hold, while derivatives like perpetual futures removed any need to faff around with the underlying asset.

Nevertheless, is it odd to have a sudden wave of selling that’s almost invisible in the underlying asset? Bitcoin velocity has barely changed over the past month, having bounced meekly off a record low in early October. Why?

Is past performance indicative of future results?

Source

The halving cycle is a reason that long-time Bitcoin holders are nervous. We show the price performance in the years after halvings in Figure 3 with the second year showing weakness. These crypto winters have been associated with 80%+drawdowns in the past as shown in Figure 4.

This post was written by Emily Woehrle, who attended the 2025 DLF Forum as an Emerging Professionals Fellow. The views and opinions expressed in this blog post are solely those of the author and do not necessarily reflect the official policy or position of the Digital Library Federation or CLIR. 2025 Emerging Professionals Fellowships were supported by a grant from MetaArchive.

Emily Woehrle is a Digital Content Librarian at the University of Toronto Libraries, where she supports a large-scale website renewal project and manages the library’s LibGuides service. She also works as a part-time librarian at the Toronto Public Library. With a background in non-profit communications and content management, Emily brings an interdisciplinary perspective to her work and looks forward to sharing experiences and learning from peers at the DLF Forum to advance sustainable, user-centered digital library practices.

Documentation as responsible digital stewardship

Attending the DLF Forum as one of the Emerging Professionals fellows was an incredibly positive experience that left me with new connections, new ideas, and a validating sense of solidarity. The Forum was only my second library conference, yet I felt immediately comfortable among people who get it and were ready to share and listen to each other’s experiences. It was also a joy to navigate the conference with my fellow Fellows.

Two presentations stood out to me over the course of the Forum, and neither focused on trendy hot topics. Instead, they highlighted the importance of documentation; the practical, behind-the-scenes work that keeps digital libraries and archives running by codifying tacit knowledge and establishing the workflows, structures, and guidelines that sustain digital library work.

In my current role, I’m coordinating a large-scale website consolidation project that requires my colleagues and me to build processes and governance structures impacting 20+libraries and departments. Documentation has become essential to scaling the project and keeping everyone aligned. Over the past year, I’ve spent a lot of time thinking about the most effective ways to develop knowledge-management systems, why some workplace cultures prioritize them more than others, and what that means for long-term success.

The first session, “Agile Documentation Development for Digital Preservation Systems,” offered strategies to make documentation immediately useful, iterative, and collaborative. The presenters emphasized creating “minimum viable documents” that favor progress over perfection – start with something usable, then refine it over time. They also underscored how role clarity and interdepartmental culture shape the success of documentation efforts. This session helped me reframe documentation as a living tool whose maintenance must be built into our work rather than treated as an afterthought.

The second session, “Renaming Failure as “readme.files”: Lessons Learned from Early and Mid-career Archives Perspectives,” reminded me that unexpected challenges are inevitable and that they can serve as learning opportunities instead of being perceived as failures. The speaker spoke about the importance of recording “detours” as they happen and how documentation can play a key part in reflection on lessons learned. She also discussed the value of using documentation to close “open loops” when offboarding from a project, ensuring that future staff can build on past work rather than unknowingly duplicating it. It was a practical reminder that documenting failure isn’t about dwelling on mistakes; it’s about giving the next person a clearer path forward.

Taken together, these sessions reinforced that documentation is more than a checklist. It can be a form of care—not only for colleagues and users who will later take on or inherit the work, but also for the library systems that depend on it. Creating collaborative documentation is an often overlooked and undervalued core competency, yet it is fundamental to both project and organizational success. I left the Forum with a renewed commitment to integrating these approaches into my own knowledge-management practices and to advocating for clearer, more collaborative documentation across the teams I work with.

The post Fellow Reflection: Emily Woehrle appeared first on DLF.

This post was written by Dorian McIntush, who attended the 2025 DLF Forum as an Emerging Professionals Fellow. The views and opinions expressed in this blog post are solely those of the author and do not necessarily reflect the official policy or position of the Digital Library Federation or CLIR. 2025 Emerging Professionals Fellowships were supported by a grant from MetaArchive.

Dorian McIntush is the Open Scholarship and Data Resident Librarian at Washington and Lee University, where he supports faculty and students with digital research and open knowledge initiatives. He has a commitment to creating equitable access to knowledge and is particularly interested in exploring the environmental impact of digital technologies and open scholarship models that prioritize accessibility and long-term sustainability. Beyond his professional work, he enjoys tromping through Virginia’s hiking trails with his dog, taking on new knitting projects, and cooking interesting recipes for dinner.

When I stepped into my current role of Data and Open Scholarship Resident Librarian at Washington and Lee University in July of this year, I was also stepping for the first time into the world of academic libraries. I focused on public libraries during my MLIS and after graduation I worked at the DC Public Library. My new academic librarian position was also brand new at my institution and was constructed to be a librarian residency. This meant that I would have a lot of freedom to grow into and shape the role, but also no real history to use as a support and learn from. This was simultaneously a gift and also a little daunting.

Coming from public libraries, I had grown used to thinking about access in very practical, immediate terms. Who has a library card? Who can physically get to our building? What barriers keep people from the resources they need? But the sessions at DLF pushed me to think about access in ways that felt more expansive.

Amber Dierking’s presentation on the Queer Liberation Library was a particular highlight for me. I’d already been a user and huge fan of QLL, but hearing Dierking talk about the work behind it reinforced everything I loved about their approach. QLL didn’t reinvent the wheel. They focused on using existing tools, keeping it simple, making it free. As someone building a role from scratch, the creative pragmatism of QLL felt like a blueprint I could make use of.

I was also drawn to Mariam Ismail’s presentation on the 23/54 Project. The work of preserving a community quilt through 3D scanning and building an interactive digital exhibit felt like a perfect example of what digital humanities could be at its best: deeply rooted in community, respectful of material culture, and genuinely expanding access rather than just digitizing for digitization’s sake. It made me think about the special collections and archives at my own institution and how we might engage descendant communities and students in similar ways.

The Data Advocacy for All Toolkit presentation tied these threads together for me in a way I didn’t expect. The team was talking about who gets to tell stories with data, who gets left out of those stories, and how we can teach people to use data ethically for social change. This toolkit offered a framework that felt aligned with my public library values, one that’s accessible, focused on equity, and designed to empower data users and learners.

DLF gave me permission to think big while starting small. I’m returning to W&L with a clearer sense of what this residency could become, not a replica of someone else’s role, but something shaped by the communities I serve and the values I bring from public libraries into this new academic space.

The post Fellow Reflection: Dorian McIntush appeared first on DLF.

Hopefully this isn’t perceived as me caving, but I’m trying to redirect my ire at the spread of genAI into learning how to evaluate genAI, especially when comparing it to existing non-genAI systems, but also between different genAI solutions (models, etc).

I don’t consider myself a doomer or a boomer, and see genAI as normal technology that needs to be evaluated as a technology. I know this is a broad area, that overlaps somewhat with how the models are themselves built (benchmarking), but if you have recommendations please let me know?

https://xkcd.com/329/

Eight courses from the Web Science and Digital Libraries (WS-DL) Group will be offered for Spring 2026.  The classes will be a mixture of on-line synchronous, asynchronous, and f2f. 

• CS 418/518 Web Programming, Ms. Nasreen Muhammad Arif, Mondays & Wednesdays 3:00-4:15 
  Topics: MySQL, React, Express.js, NodeJS, MVC (Model-View-Control), Search Engines, GitHub  
• CS 431/531 Web Server Design, Mr. David Calano, Mondays, Wednesdays, & Fridays, 10:00-10:50 
  Topics: HTTP, REST (Representational State Transfer), HATEOAS   
• CS 432/532 Web Science, Ms. Nasreen Muhammad Arif, asynchronous  
  Topics: Python, R, D3, ML, and IR.  
• CS 620 Intro to Data Science & Analytics, Dr. Bhanuka Mahanama, Web, Tuesdays & Thursdays 6:00-7:15 and asynchronous (8 week session starting in October)
  Topics: Python, Pandas, NumPy, NoSQL, Data Wrangling, ML, Colab.
  
• CS 625 Data Visualization, Dr. Bhanuka Mahanama, Tuesdays & Thursdays 1:30-2:45
  Topics: Tableau, Python Seaborn, Matplotlib, Vega-Lite, Observable, Markdown, OpenRefine, data cleaning, visual perception, visualization design, exploratory data analysis, visual storytelling 
• CS 732/832 Human Computer Interaction, Dr. Bhanuka Mahanama, Tuesdays 4:30-7:10 
  Topics: Cognitive and social phenomena surrounding human use of computers.  
• CS 733/833 Natural Language Processing, Dr. Vikas Ashok, asynchronous 
  Topics: Language Models, Parsing, Word Embedding, Machine Translation, Text Simplification, Text Summarization, Question Answering
• CS 734/834 Information Retrieval, Dr. Santosh Nukavarapu, Tuesdays 4:30-7:10 
  Topics: crawling, ranking, query processing, retrieval models, evaluation, clustering, machine learning. 

--Michael

This text, part of the #ODDStories series, tells a story of Open Data Day‘s grassroots impact directly from the community’s voices. The event ‘Leveraging Open Data for Child Advocacy in a Polycrisis Context’ was successfully held on 6th March 2025 in Imo State, bringing together child rights, teachers, advocates, policymakers, data analysts, church leaders and...

The post 🇳🇬 Open Data Day 2025 in Owerri: Leveraging Open Data for Child Advocacy first appeared on Open Knowledge Blog.

This text, part of the #ODDStories series, tells a story of Open Data Day‘s grassroots impact directly from the community’s voices. On March 4, 2025, the YouthMappers Chapter at the Institute of Remote Sensing and GIS (IRS), Jahangirnagar University, proudly celebrated Open Data Day 2025 with an impactful event titled “Mapping Resilience: Harnessing Open Data...

The post 🇧🇩 Open Data Day 2025 in Sunamganj: Harnessing Open Data for Flood Preparedness first appeared on Open Knowledge Blog.

I learned about Marimo a few months ago from this podcast episode and have been meaning to try it out. As a long time user of Jupyter notebooks I was interested to hear how Marimo was built (in part) to help solve the problem of reproducibility in data science and research software (Pimentel, Murta, Braganholo, & Freire, 2019). This is a problem I have had a lot of experience with, especially when sharing Jupyter notebooks with others.

The key thing that Marimo brings to the Python notebook to improve reproducibility is reactive execution. Marimo uses the Python AST to remember what cells depend on other cells, and when a change in one requires the execution of another, it will go and update it for you. Because of how they are created and edited, it’s very common for Jupyter notebooks to get into an inconsistent state because of the order in which cells are executed. This problem goes away with Marimo.

But, this aside, I thought I’d mention some somewhat superficial things that I immediately liked about Marimo…bearing in mind I’ve only been using it for one day.

• Pandas dataframes appear as nicely formatted tables that can be easily scrolled horizontally, without truncation of values, or elided columns. I’ve gotten this to work in Jupyter notebooks in the past, but it always requires some fiddling it seems, and Marimo does it out of the box.
• Table columns are sortable, filterable and can be summarized.
• You can page through large tables.
• You can easily download tables
• You can commit your notebook to a git repository, and diffs in pull requests make sense.
• You can easily run the notebook from the command line.
• You can embed tests in your notebooks, and run them separately.
• Built in basic charts (pie, bar, line, etc) which you can view source for and hand craft if you want (seems to use Altair).

This post was written by Darcy Ruppert, who attended the 2025 DLF Forum as a Public Library Worker Fellow. The views and opinions expressed in this blog post are solely those of the author and do not necessarily reflect the official policy or position of the Digital Library Federation or CLIR. The 2025 Public Library Worker Fellowship was supported by Platinum sponsor AM Quartex.

Darcy Ruppert is an archivist and librarian living and working in the greater Seattle area. Since receiving their MLIS from the University of Washington in 2023, Darcy has worked in various museums, special collections, and community archives, specializing in the digital preservation of audiovisual collections. Their professional work has been defined by a commitment towards democratizing access to digital archives themselves and the tools of digital preservation. They are currently managing King County Library System’s Memory Lab, a Mellon Foundation-funded community oral history project with a mission to record, preserve, and share the stories of King County.

I am so grateful to have had the opportunity to attend my first DLF Forum as a 2025 Public Library Worker Fellow. In my current work as project manager for a new community oral history project based out of a large public library system, I often feel somewhat separate from the day-to-day of the rest of my organization. The work that I do connects to and is born out of the work of the library, but the unique nature of the program within my organization means that I am often working through problems and making decisions on my own. For this and many other reasons, it was refreshing to be surrounded by a community of my professional peers who are facing similar challenges and grappling with similar emergent questions, both practical and philosophical, within our field.

The first day’s opening plenary by Dr. Kay Coghill provided, for me, a grounding moment at the very beginning of the Forum. Dr. Coghill shared an entreaty for all of us, as digital library professionals and as human beings, to use our platforms, skills, and resources to mitigate harms to members of systemically marginalized groups in digital spaces. Their incisive talk made me think about my own positionality in digital spaces, and drove me to reflect on my own professional decisions and the cascading effects these (at times, seemingly small) decisions can have. In the work that we do, I think it’s easy to become complacent, to lose sight of the fact that we hold power to protect and, by the same token, harm, when we introduce new solutions without appropriate testing, community consultation, and evaluation. I think there’s a real danger in our field of focusing so much on integrating new technologies and providing “results” to our institutions that we may passively introduce real risks into the lives of our users and members of our greater community.

With Dr. Coghill’s talk framing the rest of the Forum for me, I think I appreciated the sessions I attended with a rejuvenated perspective towards care and stewardship. I thought about the race to adopt “industry standard” digital preservation tools, and what we lose when we fail to properly evaluate these tools for the unique needs of our organizations and user groups. At the University of Denver’s session on curating digital exhibits, I reflected on our ability as cultural heritage stewards to uplift and uncover marginalized voices rather than bend to dominate cultural narratives. At a session on AI-powered transcription, I considered the balance of risk and reward that is inherent to the world of LLMs. I’m thankful for all of the speakers who generously shared their work at the Forum, and for the opportunity to reflect on these important issues alongside my peers. 

The post Fellow Reflection: Darcy Ruppert appeared first on DLF.

LibraryThing is pleased to sit down this month with Kelly Scarborough, who makes her authorial debut this month with Butterfly Games, a historical novel set in the Swedish royal court during the early 19th century. After working for two decades as a law firm partner and white-collar prosecutor, Scarborough returned to her interest in historical fiction and her love of writing, determined to tell stories about fascinating women who lived through challenging times. Scarborough sat down with Abigail this month to discuss her new book, due out later this month from She Writes Press.

Butterfly Games is based on a true story, and its heroine, Jacquette Gyldenstolpe, on a real person. Tell us a little bit about that story and how you discovered it. What made you feel that it needed to be retold?

Like so many turning points in my life, Butterfly Games began with a book. As a teenager, I fell in love with Désirée, Annemarie Selinko’s novel about Désirée Clary—the silk merchant’s daughter who was once engaged to Napoleon and later became Queen of Sweden. I read it over and over, fascinated by how a woman could be swept into history by forces she never chose.

Years later, during a difficult period in my life, that novel came back to me. I began researching Désirée’s descendants—the Bernadotte dynasty, which still reigns in Sweden today—and uncovered a world of political upheaval, fragile alliances, and private heartbreak. That’s when I stumbled across Jacquette Gyldenstolpe.

Jacquette appears in the historical record mostly as a scandal: a young countess who fell in love with Prince Oscar, the heir to the throne. But the more I read—letters, memoirs, court gossip—the more I realized how much of her story had been left untold. She wasn’t just a footnote in someone else’s rise to power. She was a young woman navigating impossible choices in a world where love could threaten a dynasty.

Once I found her, I couldn’t look away. I knew her story needed to be retold.

What kind of research did you need to do, while writing the book, and what were some of the most interesting things you learned in that process?

Can you see me smiling? I don’t think I’m capable of separating the research I needed to do from the research that simply called to me and took over my brain.

Over the course of several years, I spent more than eighty nights in Sweden, translated hundreds of handwritten letters, and built a chronology with more than five thousand entries to track who was where, with whom, and why. Jacquette’s world became a place I loved to inhabit. One day stands out above all others. I was granted special access to Finspång Castle, Jacquette’s childhood home—now a corporate headquarters, a place closed to the public. No photographs were allowed, so I took frantic notes on my phone as we walked through the women’s wing. In a sitting room, I noticed a small mother-of-pearl nécessaire—a sewing and writing box with tiny compartments for her most personal objects. It stopped me cold. My guide, a retired corporate executive who knew the house intimately, leaned in and whispered, “Jacquette’s.”

The box had been a gift from Jacquette’s husband, Carl Löwenhielm. That moment—imagining her hands opening it, choosing a needle or a quill knife—changed the direction of the book.

Suddenly, Jacquette wasn’t a scandal or a symbol. She was real.

Your book has been described as a good fit for admirers of Philippa Gregory and Allison Pataki. Did the work of these authors, or others, influence you when writing your story?

Absolutely—though in different ways. Philippa Gregory is a master of taking a story with a known, often tragic ending and making it feel suspenseful and intimate. I admire how she builds emotional momentum even when readers think they know what’s coming. Two of my favorites are The Kingmaker’s Daughter and her most recent novel, Boleyn Traitor.

Allison Pataki has also been influential, particularly in how she blends rigorous research with accessible storytelling. I love the smart, resourceful heroines she creates from women who otherwise might be lost to history. Her work reminds me that historical fiction can be immersive without being intimidating—and romantic without losing its seriousness. Both my book clubs loved Finding Margaret Fuller, and I did, too.

You’ve had a full career as a lawyer and prosecutor, before turning to writing. How has that work informed your writing and storytelling?

Don’t get me wrong, I had a lot to learn before writing a novel, but some of the things I loved about law proved useful for writing historical fiction. Law trained me to think in terms of evidence, motive, and connections. When you’re preparing a case, you assemble fragments—documents, testimony, inconsistencies—and shape them into a coherent narrative that persuades a jury.

Writing historical fiction isn’t so different. The facts matter deeply, but facts alone don’t tell a story. You have to decide what belongs at the center, what remains in the background, and where the emotional truth lives. My legal background also made me comfortable sitting with ambiguity. History is full of unanswered questions, and I don’t feel the need to resolve every one neatly. Sometimes what’s most compelling is what can’t be proven.

Tell us a little bit about your writing process. Do you have a particular routine—a time and place you like to write, a particular method? Do you plot your stories out ahead of time, or discover how they will unfold as you go along?

When the stars align, I retreat early in the day to the attic office of my nineteenth-century house in Connecticut, take my Shih Tzu upstairs with me, and leave the modern world behind. I wrote Butterfly Games in nine drafts. There was an outline, but I changed the plot in significant ways as I went along. For the sequel, I’m trying to be a little more disciplined. I started with an outline—but found myself getting too granular—so I switched to ninety old-fashioned index cards. Each card holds one scene: chapter number, date, setting, point-of-view character, and the scene’s pivot point. There’s barely room left for anything else, which forces clarity. I transcribed those cards into Scrivener, and now I’m writing. We’ll see how closely I stick to the plan.

What comes next? Are you working on any additional books?

Yes. Butterfly Games is the first novel in a planned series. The second book picks up after the events of the first and follows Jacquette and Oscar into a far more dangerous phase of their lives—when love has consequences, secrets carry weight, and survival requires choices that can’t be undone.

Tell us about your library. What’s on your own shelves?

My physical library is filled mostly with historical fiction, especially novels with complex, non- linear structures. I return again and again to Hamnet and The Marriage Portrait, as well as The Time Traveler’s Wife and Pure.

On a special shelf, I keep books connected to Jacquette’s world—like Désirée and The Queen’s Fortune—alongside more than a hundred antique Swedish memoirs and histories, many written by people who actually knew Jacquette.

And for bedtime? A Kindle packed with historical romance by Sarah MacLean, Julia Quinn, Tessa Dare, and Lisa Kleypas.

What have you been reading lately, and what would you recommend to other readers?

For lovers of royal historical fiction, Boleyn Traitor is a must-read. I was also lucky enough to read an advance copy of It Girl, which I loved.

My favorite read last year was Broken Country—a deeply emotional novel with one of those intricate narrative structures that stays with you. In fact, I want to read it again.

These are some things I’ve wandered across on the web this week.

Once upon a time, I taught a course on AI in the Library in an iSchool. (Check out the reading list and syllabus, which are relevant context for this post.) I haven’t taught it for a while — I got busy with other things, and then ChatGPT dropped and immediately rendered my syllabus obsolete and I wasn’t up for overhauling it.

At least, that was why I wasn’t teaching it in 2022-2024. In 2025 and 2026, there are some pretty different reasons not to be adjunct faculty, particularly in a public university, if you don’t have to be. Like an enormous Project-2025-based, federally directed but decentralized system to dictate syllabi, destroy academic freedom, and fire professors who don’t toe the line. (Examples unfortunately abound, such as University of Oklahoma, Texas A&M, San José State University — the last of these being the university where I taught.)

And in the midst of all this, here’s the (excellent) Texas Tribune reporting on ChatGPT prompts used at Texas A&M to flag course content that might take a perspective on race and gender. And so, with a rising sense of nausea, I may as well see how my syllabus would fare in this brave new world.

First (and startlingly), the prompts only cover very high-level course metadata: course title, description, and learning objectives.¹ The week-by-week descriptions and reading list, or indeed the syllabus as a whole, aren’t covered, so these prompts don’t reach week 2’s case study of an AI metadata generation prototype in a Black photographer’s archive, or week 3 on the broad and heavily researched field of (ironically) algorithmic bias.

And so, the results:

Course Title. Unsurprisingly, “AI in the Library” raises no issues for the state of Texas (though of course chatgpt takes a great many words to be unconcerned about it).

Course Description.

This course covers what artificial intelligence is and critical analysis of AI systems, in the context of real-world library and archival AI applications. Programming skills not required.

To my surprise, chatGPT is pretty okay with this:

Review of value-neutrality

• Advocacy or ideological language:
  None identified. The description does not include prescriptive claims, affective outcomes, or ideological commitments.
• Potential neutrality concerns:
  The phrase “critical analysis of AI systems” is standard academic language and does not, on its own, imply advocacy. However, “critical” can be interpreted broadly; specifying analytical methods or perspectives can improve clarity without reducing neutrality.
• Overall assessment:
  The description is value-neutral, concise, and appropriate, though it is shorter than the requested 75–125 words and could benefit from greater methodological detail.

I was sure that “critical analysis” would get more pushback (how could we possibly want that in higher education in 2026??), but I guess mostly I’m supposed to be more unnnecessarily wordy, and I could keep waving this IYKYK flag.

Learning Outcomes.

Upon successful completion of the course, students will be able to:

1. Understand and explain the basics of AI: both its underlying principles and common machine learning techniques.
2. Discuss realistic ways that AI can be a part of library services.
3. Critically analyze potential pitfalls of AI systems, including the role of their underlying data sets and their ramifications in society.

Y’all. Despite some honestly fair ways chatgpt pointed out that these verbs are squishy and might be replaced with ones that are sharper and/or lend themselves more obviously to assessment, I absolutely slipped “critically analyze…their ramifications in society” right past it:

“Critically analyze potential pitfalls…” — This is largely value-neutral. The phrase “ramifications in society” is acceptable, though specifying analytical contexts (e.g., professional or institutional settings) would improve precision.

Again, let me reiterate that I have a case study on a Black archive. I have a reading list with, inter alia, Meredith Broussard and Catherine D’Ignazio and Safiya Noble and Virginia Eubanks. I have a reading list with Tay (aka “how to make a chatbot unbelievably racist and anti-Semitic in less than 24 hours”) and predictive policing and de-biasing word vectors and feminism and Soylent-Green-based limitations of a major AI ethical framework.

Pretty sure any antagonistic human reading learning objective number 3 there would know what I was getting at, but I guess chatgpt still can’t replace all the jobs.

1. There’s also a prompt for “course justificiation”, but I was not asked to write one of these for my syllabus. The closest it gets is a “core competencies” section aligning the course with learning objectives for the overall iSchool program. However, these objectives are govened by the iSchool, not the individual professors. Therefore I’ll be leaving the course justification prompt aside. ︎

Photo by Mateusz Zatorski on Unsplash

“Ghost kitchens” are pop-up restaurants geared entirely toward food delivery. They typically rent space in traditional restaurants to prepare food, take orders online, and deliver them to the doorstep via delivery apps like DoorDash or Uber Eats. Ghost kitchens proliferated during the COVID pandemic, which for a time practically extinguished dine-in food service. Restaurants of all descriptions needed to restructure their operations to scale up food delivery as their main service model; ghost kitchens were the extreme example, with the entire service model built around delivery.

The story of ghost kitchens is one of a specific business sector—restaurants—retooling traditional operational structures and service models to meet changing conditions in the marketplace. Gary Pearce, Director, Academic Services in the Monash University Library, touched on a similar theme in a recent OCLC Research Library Partnership webinar, describing how the Library reimagined its operational and service models to scale up research support capacities and better address institutional needs and priorities. As with ghost kitchens, Monash sought to reimagine its services in response to changing imperatives—specifically, the need to deliver research support at scale, within the confines of prevailing budgetary limitations. This situation will surely resonate with other research libraries, and there is much to be learned from Monash’s experiences and innovative solutions.

Retooling operational structures and service models

Academic Services is one of three portfolios at Monash University Library. To address the need to scale research support services and align more closely with stakeholder needs, Academic Services shifted from a traditional liaison librarian model organized on disciplinary lines to a functional specialization approach based on library expertise. This change moved away from multiple teams providing duplicate services to specific disciplines, in favor of agile, project-based service teams that work across disciplines.

A key aspect of Monash’s approach is the creation of a new Library Business Partner role, whose chief responsibility is strategic relationship management with senior leadership in a specific academic area. The Library Business Partner serves as a conduit for two-way communication between the Library and its academic stakeholders: on the one hand, communicating library messaging to the academic unit, and on the other, gathering intelligence and feedback on the unit’s needs and mobilizing capacity within the Library’s service teams to address them.

Pearce provided a rich description of how this retooling of operational structures and service models was conceived and implemented. Here are a few of the themes that emerged from his discussion:

• Acknowledging relationship management as a dedicated role: A key innovation was the creation of the Library Business Partner role to manage outreach and engagement with academic units. The Library Business Partner represents the entire Library and therefore can provide a comprehensive view of Library capacities, as well as expedite responses to stakeholder needs. Separating relationship management from service delivery facilitated a shift from a reactive, transactional model to a more proactive, two-way partnership.
• Emphasizing a culture of agility: Building a service model that was both scalable and responsive led Monash to adopt an agile approach. Academic Services implemented a matrix organizational structure in which staff have fixed reporting lines with flexible membership across multiple service teams—including research support. Staff have the option to rotate across teams to deepen expertise and experience. Work is divided between “business as usual” work and project work, the latter of which can be scaled up or down as needs and resource availability dictate. While this new operational structure could pose challenges to long-standing professional identities tied to traditional service models, it also opens up new pathways to leverage existing areas of expertise and develop new ones.
• Close attention to change management: The new operational structures were a significant departure from previous models. In recognition of this, the development and implementation processes were characterized by consultation, transparency, and communication, including a series of consultative visits to peer institutions facing similar challenges in adapting service development and delivery; presentations to stakeholder groups; regular updates to staff, along with clear milestones and timelines; open channels for questions and feedback; and planning for professional development needs.     

These are just some of the key themes that provide the foundation for Monash Library’s story of transformation, scalability, and responsiveness.

Managing implementation is crucial

Pearce’s presentation elicited many questions from the audience in attendance. Collectively, the questions reflected a keen interest in the implementation aspect of the shift to new operational structures and service models, touching on issues like:

• Stakeholder response and buy-in: how researchers and staff reacted to service model changes; channels for communication and feedback
• Staffing implications: impact of restructuring on staffing counts; work allocations between “business as usual work” and project work; cross-training opportunities
• Strategic relationship management: interest in the details of how the new Library Business Partner role works in practice

The audience’s interest in these topics highlight that a shift from a traditional/subject-focused service model to a functional/specialization model requires attention to both structural innovation and the staffing and stakeholder reactions to significant organizational change.

Additional reading

Pearce’s webinar intersects with several OCLC Research studies that complement some of the themes from the emerging from the service model transformation experience. First, check out our work on social interoperability, which we define as the creation and maintenance of working relationships between individuals and organizational units within an institution. Our report describes strategies and tactics that can help strengthen social interoperability skills—an essential element of roles like Monash’s Library Business Partner. In addition, OCLC Research’s forthcoming work on the Library Beyond the Library—an operational principle that emphasizes the importance of the library engaging with the broader institutional environment through strategic alignment, collaboration, and storytelling—connects with Monash’s ambitions to retool its service model to better align with institutional research needs and priorities.

Ready to dive deeper? Listen to the full recording

The webinar and subsequent Q&A offered a richly informative look behind the scenes of a major shift in operational structures and service models to better address the needs of stakeholders. If you didn’t have a chance to join us for the live webinar, please take some time to view the recording. Many thanks to Gary Pearce for sharing his perspective with all of us!

The post Scaling research support at Monash University Library appeared first on Hanging Together.

This post was written by Cláudia De Souza, who attended the 2025 DLF Forum as the Grassroots Archives and Cultural Heritage Workers Fellow. The views and opinions expressed in this blog post are solely those of the author and do not necessarily reflect the official policy or position of the Digital Library Federation or CLIR.

Cláudia De Souza is an associate professor at the College of Communication and Information at the University of Puerto Rico. She teaches in the Graduate Program in Information Science, focusing on information organization and retrieval, with particular emphasis on the analysis, evaluation, and design of digital libraries and archives. She is also the academic coordinator of the UPR Caribe Digital project, an initiative dedicated to advancing Digital Humanities research and scholarship in and for the Caribbean. She advocates for and fosters open access to knowledge, supports digital preservation, and promotes the dissemination of documentary heritage. Her work is driven by a commitment to enhancing the visibility and accessibility of information resources across the Insular Caribbean.

Attending the 2025 DLF Forum for the first time was an exceptional experience, especially as the sole fellow selected for the Grassroots Archives and Cultural Heritage Workers program. I arrived with high expectations, and the event exceeded them, offering a rich environment for learning, collaboration, and professional growth. I had the chance to meet and engage with attendees from a wide range of institutions and backgrounds, including students, emerging professionals, librarians, and archivists. I am especially grateful to the selection committee for choosing such a diverse group of fellows, which I truly appreciate. This experience of connection and dialogue with new voices gave me the opportunity to interact with a diversity of approaches in professional and academic practice, as well as the value of building collaborative networks.

One of the challenges was choosing among the multiple sessions happening simultaneously over the three days. I decided to focus on the topics most closely related to my work at the University of Puerto Rico and that I am most passionate about in library and information science: open access, metadata, information organization and retrieval, community archives, and the visibility of digital collections.

Among the new tools and approaches I explored, the Marriott Reparative Metadata Assessment Tool (MaRMAT) stood out, an open-source application designed to support reparative metadata evaluations and processes of repair and justice in information. It allows librarians to identify harmful, outdated, or otherwise problematic language in tabular metadata using pre-curated or custom lexicons. I definitely plan to explore its use in the work we are developing with community groups in Puerto Rico, as part of the UPR Caribe Digital project. I was also inspired by the presentation of Krystyna Matusiak, which demonstrated how digital collections can be expanded through exhibits that highlight the stories of underrepresented communities. Digital curation not only extends the reach of archives but also provides new opportunities to tell stories in an inclusive and meaningful way, showing how information organization can impact representation and access to collective memory. I leave the Forum inspired and motivated to apply these insights as part of the Digital Humanities initiatives we are planning at my institution, and it will serve as a model to follow.

I am grateful for the opportunity to participate and look forward to continuing these conversations and collaborations. I also hope that next year I will have the chance to submit a presentation to share with others all that I have put into practice. See you at DLF 2026!

The post Fellow Reflection: Cláudia De Souza appeared first on DLF.

This post was written by Amaobi Otiji, who attended the 2025 DLF Forum as a Student Fellow. The views and opinions expressed in this blog post are solely those of the author and do not necessarily reflect the official policy or position of the Digital Library Federation or CLIR. 2025 Student Fellowships were supported by a grant from MetaArchive.

Amaobi Otiji is pursuing his Master of Information at Rutgers University concentrating in the Technology, Information, and Management pathway. Prior to entering this program, Amaobi earned a bachelor’s degree in history from Howard University and has worked in roles involving federal collections, both digitized and born-digital. His professional interests center on digital curation, metadata development, and exploring new approaches to preserving and sharing underrepresented histories. He is focused on increasing equitable access to information and helping to shape how emerging technologies influence our cultural memory. In his spare time, Amaobi enjoys playing baritone ukulele, attending live theater, and playing video games.

Digital Memory Work Across Regions and Histories: Reflections on Community-Driven Projects

As I attended this year’s DLF Forum, I kept returning to two themes that resonated with me across multiple different presentations: community engagement and the quiet connective work that builds the infrastructure for it. Two of the sessions I attended during the conference stood out to me in particular because they approached these ideas from different perspectives but utilized similar underlying practices for catering their projects to their communities’ needs. These sessions were about the HBCU Digital Library Trust and the Borderlands storytelling initiative. Both of these projects work with different kinds of communities, rooted in distinct histories spread across North America. Yet all of them demonstrated how effective digital stewardship can be when community engagement is built into the planning of a project rather than treated as a final step at the end.

The Historically Black Colleges and Universities (HBCU) Digital Library trust session outlined a fantastic model that was centered on providing long term support for HBCUs and their unique archival histories. Their emphasis on shared ownership reflected a great understanding of these institutions and their historical challenges trying to navigate limited resources, public scrutiny, and a society that too often worked against them. What stood out to me in particular was how intentional their approach felt. Rather than expecting HBCUs to adapt to them, they adapted to the HBCUs by meeting them where they were. The Trust, hosted by the Atlanta University Center’s Woodruff Library, focused on building capacity in ways that supported institutional autonomy and reflected the needs of the communities they were trying to serve. Their model felt refreshing, informed and grounded in culturally informed practices that support long term institutional resilience.

The Borderlands storytelling session approached their community engagement from another direction. Their work seemed shaped by the layered histories and cultural dynamics found in the U.S.-Mexico Borderlands and by the University of Arizona’s position as a Hispanic-Serving Institution (HSI) with ties to the region. Their presentation drew from movements, identities, and complex narratives that define the region as well as from the data intensive methods they were using to support the work. They spoke about their efforts mapping, visualizing, and other forms of data storytelling that were central to how researchers were interpreting that complexity. What especially stood out to me was how they treated their approach to “place” as more than a backdrop. It functioned as a structure that was allowed to shape the research itself and set the terms for how they partnered with their researchers. It felt rooted in the region in a way that kept the work responsive so that it could move across research, teaching, and student engagement while still staying grounded in the histories and contexts that give it direction.

Across both of the sessions, I found myself thinking about how our community and networks shape our digital memory work every day. The HBCU Digital Library Trust and the Borderlands initiative each operate within distinct historical and cultural environments, yet they are undeniably interconnected through their commitment to engagement shaped by the histories and needs of the communities they serve. Together, these sessions were a great illumination of how digital memory work is always anchored somewhere and shaped by places, relationships, and shared histories that give it meaning. For those hoping to steward this work, our role is to listen closely enough that those anchors guide the paths we build.

The post Fellow Reflection: Amaobi Otiji appeared first on DLF.

Win free books from the January 2026 batch of Early Reviewer titles! We’ve got 227 books this month, and a grand total of 2,976 copies to give out. Which books are you hoping to snag this month? Come tell us on Talk.

If you haven’t already, sign up for Early Reviewers. If you’ve already signed up, please check your mailing/email address and make sure they’re correct.

» Request books here!

The deadline to request a copy is Monday, January 26th at 6PM EST.

Eligibility: Publishers do things country-by-country. This month we have publishers who can send books to Canada, the US, the UK, Australia, Belgium, Netherlands, Poland, Latvia, Lithuania, Luxembourg and more. Make sure to check the message on each book to see if it can be sent to your country.

Thanks to all the publishers participating this month!

A monthly round-up of news, upcoming working group meetings and events, and CLIR program updates from the Digital Library Federation. See all past Digests here. 

Hi DLF Community! I’ve really enjoyed connecting with so many of you at the Forum and in Working Group meetings over the past few months. Every conversation, formal and informal, has left me inspired and grateful to be part of a community fueled by care, creativity, and collaboration. I’m excited to keep learning with you all and stay in touch. If you want to chat one-on-one, just send me an email anytime (swillis@clir.org). Wishing you an encouraged, grounded start to 2026! 

— Shaneé from Team DLF

This month’s news:

• New DLF Working Group: The new Open Source Capacity group kicks off at the end of this month with a meeting on January 28. Learn more about this group and join their inaugural meeting.
• Closing Today, Call for Proposals: Digital Library Pedagogy Group (DLF Teach) has extended its toolkit CFP through end of day today; submit here.
• Metadata Quality Benchmarks Announced: DLF-AIG Metadata Assessment Working Group (MWG) is pleased to announce the public release of benchmarks for metadata quality. Read about them on the DLF Blog.
• IIIF Call for Proposals: IIIF’s annual conference will be held June 1-4 in the Netherlands. Review and submit to the CFP here by January 23.
• CLIR Call For Proposals: CLIR is accepting applications for the thirteenth cycle of Recordings at Risk until February 24, 2025. See details here.
• Registration Open, Program Available: IIIF’s Online Meeting will take place January 27-29; see the full program on the IIIF website, and register on Eventbrite.
• Fellow Reflections: Starting tomorrow (January 6), look for reflections from the 2025 DLF Forum fellows on the DLF blog. Read past fellow reflections here.
• Office closure: CLIR’s offices are closed on Monday, January 19, for Martin Luther King Jr. Day.

This month’s open DLF group meetings:

For the most up-to-date schedule of DLF group meetings and events (plus NDSA meetings, conferences, and more), bookmark the DLF Community Calendar. Meeting dates are subject to change. Can’t find the meeting call-in information? Email us at info@diglib.org. Reminder: Team DLF working days are Monday through Thursday.

• DLF Digital Accessibility Working Group (DAWG): Tuesday, 1/6, 2pm ET / 11am PT
• DLF Born-Digital Access Working Group (BDAWG): Tuesday, 1/6, 2pm ET / 11am PT
• DLF AIG Cultural Assessment Working Group: Monday, 1/12, 1pm ET / 10am PT
• AIG User Experience Working Group: Friday, 1/16, 11am ET / 8am PT
• DLF Digitization Interest Group: Monday, 1/26, 2pm ET / 11am PT
• DLF Committee for Equity & Inclusion: Monday, 1/26, 3pm ET / 12pm PT
• DLF Climate Justice Working Group: Tuesday, 1/27, 1pm ET / 10am PT
• NEW GROUP! DLF Open Source Capacity Group: Wednesday, 1/28, 1pm ET / 10am PT
• DLF AIG Metadata Assessment: Thursday, 1/29, 1:15 pm ET / 10:15 am PT
• DLF Digital Accessibility Policy & Workflows subgroup: Friday, 1/30, 1pm ET / 10am PT

DLF groups are open to ALL, regardless of whether or not you’re affiliated with a DLF member organization. Learn more about our working groups on our website. Interested in scheduling an upcoming working group call or reviving a past group? Check out the DLF Organizer’s Toolkit. As always, feel free to get in touch at info@diglib.org. 

Get Involved / Connect with Us

Below are some ways to stay connected with us and the digital library community: 

• Subscribe to the DLF Forum newsletter.
• Join, start, or revive a working group and browse their work on the DLF Wiki.
• Subscribe to our community listserv, DLF-Announce.
• Bookmark our Community Calendar.
• Learn more about becoming a DLF member organization. 
• Follow us on LinkedIn and YouTube.
• Contact us at info@diglib.org.

The post DLF Digest: January 2026 appeared first on DLF.

Developing a long-term illness, whether chronic or acute, is like being dropped into a country completely unfamiliar to you. You don’t know the language, the customs, the cuisine, the people. You feel alone, isolated, and totally out of your depth. Eventually, you start to learn the language, the customs. You find community, fellow travelers, people who can help you understand your new life better. It doesn’t stop being hard, but the learning curve becomes less steep and the isolation less intense. 

However, unlike when you’re immersed in a new country and culture, you’re falling into this new place and experiencing that painfully slow acculturation while you’re trying to still live your regular life in parallel. You’re expected to be a good parent, partner, family member, friend, employee, housekeeper, bill-payer, etc. But you’re living in two different realities now and because of that, it’s easy to feel alienated from your regular life, especially if you don’t feel like you can bring that other part of yourself to your interactions at work, at home, or out with friends. The cognitive dissonance can be jarring.

It’s hard enough to live that double life, but adding in the vagaries of seeking out a diagnosis and often not being believed, not to mention coping with the symptoms themselves, can make life feel completely untenable. Before my autoimmune diagnosis, I spent more than a year seeing medical professionals who didn’t believe there was anything wrong with me other than the normal discomforts of aging. I kept asking doctors if they thought my symptoms could be autoimmune and was told “no” over and over again, though that felt wrong to me. One PA suggested that some of my symptoms might stem from anxiety since I have a history of anxiety (like I wouldn’t know at this point what anxiety feels like). Within five minutes of talking with the first rheumatologist I saw, after waiting five months for the appointment, he said to me “this doesn’t sound rheumatological at all.” Luckily he still did all the standard testing which showed that he was very wrong. But not being believed by so many doctors for so long stays with you. It leaves a scar. Every time I see a doctor now, I feel like I’m going to court and I’m ready to be cross-examined, to be picked apart. I’m a bundle of nerves.

And my experience is painfully common, especially for women, as poet Meghan O’Rourke writes in her amazing book about chronic illness, The Invisible Kingdom (a meditation on and journalistic exploration of chronic illness and how it is positioned in our social fabric):

And so it is a truth universally acknowledged among the chronically ill that a young woman in possession of vague symptoms like fatigue and pain will be in search of a doctor who believes she is actually sick. More than 45 percent of autoimmune disease patients, a survey by the Autoimmune Association found, “have been labeled hypochondriacs in the earliest stages of their illness.” Of the nearly one hundred women I interviewed, all of whom were eventually diagnosed with an autoimmune disease or another concrete illness, more than 90 percent had been encouraged to seek treatment for anxiety or depression by doctors who told them nothing physical was wrong with them. (p. 103)

Once I got on meds for my condition that finally started working (most first-line meds for autoimmune conditions take three months on average to produce any effects on the immune system), I thought I was past the worst of it. Other than occasional much smaller flares, I was essentially in remission. I learned my limits. I protected my spoons with my life. I kept my stress low. I felt like I had it figured out. I felt good even. And then I got sicker with new symptoms that have stolen so much from me, including my sleep. The past 10 months have honestly been a nightmare with a carousel of doctors who all have completely different theories of what is going on and with a condition that is constantly evolving so what they see in the moment isn’t the full picture. And each doctor I’ve seen hyperfocuses on something different and ignores every other aspect of my case. It’s all making me feel like I’m going crazy. 

Each wrong diagnosis brought me to another country, another reality, another identity, that I lived in for a short while. And in each of these countries, I spent countless hours learning, learning, learning all I could, going down research and subreddit rabbit holes, and spending way too much money on products that did nothing because none of those diagnoses were correct. The dermatologist I see who specializes in autoimmune conditions at a research university seems to have given up even trying to diagnose me and she’s basically my last best hope in the state. She wouldn’t even give me differential diagnoses last time beyond “it’s clearly autoimmune given your systemic symptoms.” She’s the one who has put me on a serious immunosuppressant with debilitating side effects, which I guess at least shows she’s taking it seriously, but if she doesn’t know what she’s treating me for, how does she know that this drug that is making me feel terrible is even going to help? Atul Gawande said of doctors that “nothing is more threatening to who you think you are than a patient with a problem you cannot solve” (quoted in The Invisible Kingdom, page 209) and I feel that in how I’m treated at every appointment I go to. With the exception of the charlatan immunologist who was desperate to diagnose me with MCAS though there was no evidence in support of it, not one doctor has seemed at all interested in figuring out what this is – it’s felt more like a game of “not it.”

Of a similar liminal moment in her own illness, O’Rourke wrote, “in my illness I was moored in an unreachable northern realm, exiled to an invisible kingdom, and it made me angry. I wanted to rejoin the throngs. In dark moments I continued to wonder if the wrongness was me” (99). And of course people will feel like that wrongness is them when we live in a culture that views chronic illness as some sort of weakness or something we caused through through our own bad habits. Like O’Rourke, I feel both exiled from my world and forced to be in it at the same time, which is a unique form of torture. Going to work, pretending things are ok, doing my job, meeting deadlines, helping students, smiling, all the while my body is attacking itself, I’m barely sleeping, I’m spontaneously bleeding from my skin and under my skin, and I’m so itchy I sometimes have to wear gloves to bed or I’ll scratch myself raw in my sleep. You feel like you’re play-acting being yourself, being a person in the world, because you’re not really there anymore. And when you’re suffering, and don’t know what your illness is, and you feel abandoned, it’s easy to go down rabbit holes of self-loathing along with those rabbit holes of fruitless research that make you feel like an unhinged obsessive with a murder board and yarn. As Meghan O’Rourke wrote, “your sense of story is disrupted” (p. 259) and you feel like a stranger to yourself.

I started writing about slow librarianship long before I got really sick, but even then, I knew the importance of fostering a work culture where you can be a whole person. I knew how it felt to have a child and feel like you couldn’t prioritize family obligations over work ever (though working during family time? Totally ok, right?). In a workplace that encourages people to be whole people, workers feel like they can prioritize the things in their lives outside of work that are important – their caregiving responsibilities, their health, the people they love, etc. They feel like they can talk about these things – that they don’t make them liabilities. They can be vulnerable and real. And feeling like you can be vulnerable and real about who you are and how you’re doing means that you can also be vulnerable and real in your work, which makes us better employees who are energized to try new things.

I think a lot of people in positions of power might even want a culture like this, but very few actively create it. They might think that saying “take what time you need” when someone is facing illness is enough. But I think two pieces are missing from this. First, managers need to not only say “take what time you need” but work with their direct reports to address the work that would otherwise pile up. If you say “take all the time you need” but all the work with its stressors and deadlines is still there, you’re not really giving people space. Can you take good care of yourself while you watch the work pile up and up and up? How many of us have come back to work while still not fully recovered from an illness because of the work that was piling up or a class they needed to teach? 

Also, managers need to model vulnerability, transparency, and being whole people themselves. If they put up a false front of strength, if they’re not willing to be vulnerable and human and real themselves, if they do not model transparency, there’s no way that others will feel safe doing so. I was lucky to have a boss in my first academic library job who was deeply human in her interactions with her employees, so I got to see what that looked like. And it was her humanity that engendered fierce loyalty in her employees – we all thought the world of her. Even when she made decisions that people didn’t like (which was rare as she really did take our insights to heart), she explained her thinking in a transparent way. Given my later experiences, her way of being feels vanishingly rare. I think a lot of managers feel like they need to project strength, not explain their decisions, not let their direct reports really know them as people with full lives, but I don’t think that’s true. A lot of managers operate out of a place of fear or insecurity, but my first academic library director was confident enough to be her full self, flaws and all. 

In a culture where we don’t feel like we can bring ourselves fully to work, I don’t feel like I can talk about my illness. I feel selfish and weak for even considering it. Like, we all have shit going on, right? The world is pretty awful right now. People’s lives are complicated and messy and there’s probably a lot of suffering I know nothing about happening all around me. If they don’t talk about it, who am I to talk about it? I’m not special. While I’ve mentioned being sick at work in the context of being immunocompromised and needing to protect myself and not participate in large, crowded events, even that has felt really uncomfortable. Everyone should feel like they are important enough to their places of work and valued enough to bring up these things without feeling embarrassed or like they’re asking for “special treatment.” I read recently (can’t remember the source) that close to 60% of people with chronic illnesses have not told people at work about it. Imagine hiding something that is such a pivotal and ineluctable piece of many people’s identities and think about the double life that forces them to live.

At work I feel a lot of shame about being sick and I work more than I should given how I’ve been feeling (this is common). I feel like I need to mask how I’m really doing, that people don’t want to hear it. And it’s true that most don’t. There are three people at work I can talk to about my illness, but others seem so incredibly uncomfortable when I mention it, so I’ve learned to just pretend it’s all ok or say nothing. I know that some of my reticence and shame comes from my own internalized ableism as it’s the water we all swim in, but when I worked in that library where the culture encouraged vulnerability, humanity, and care, I remember how different it felt. How much less distance there was between the person I really was and the person I was at work.  

In a meeting last year, our Dean was talking about making the next all-library meeting in-person only. Previously, they had always offered them hybrid, but she didn’t like that most people were choosing not to come in-person. And I totally get it, even though it sucks to always be the outlier. She wants it to be a team-building experience and that’s really hard to do when most people are participating from home with their cameras off. At that meeting, for the first time, I disclosed my illness in front of a bunch of people and talked about how important it is to always offer an online option for folks who are medically vulnerable or at least find ways to make indoor spaces safer for those who can’t afford to get sick. My boss then asked to meet with me to talk about how to make our spaces safer. I talked about airflow, encouraging and providing masks, and, during temperate months, having the meetings at places where windows and doors can remain open or even holding them outside (we’d had one meeting at a park a few years ago which had been the best one ever from both a health and team-building perspective). I suggested that we make the Winter meeting fully virtual since it’s the height of flu season and you can usually get people participating more in a fully virtual meeting than a hybrid one, where the online people feel like weird lurkers. I was thanked for my feedback and didn’t hear anything after that. This September, the all-library meeting was held in our campus library where the windows don’t open (albeit with a couple of HEPA filters scattered around, but I know we do have large college spaces where doors and windows can be opened because I went to an all-day union meeting in one where they did just that and we’d used that space for library meetings in the past) and our February meeting is going to be held in-person during the worst flu season in decades. Obviously, none of this was personal or intended to cause harm, but, at the same time, how should I feel under the circumstances? Clearly speaking out in that meeting, something that I had to really steel myself to do, had been pointless. Why would I ever bring it up or ask for anything again?

When we come back from winter break, people inevitably ask each other how it was, but do they really want to know? I think they want to hear “good,” “fun,” “restful,” etc. How do you talk about a “vacation” in which you spent most of it doubled over in pain after eating anything thanks to the toxic meds you have to take and your father-in-law was in the hospital dying the entire time? I wish that I felt I could share what an absolute shitshow it’s been, to feel like I could be a full person at work, but when you know no one really wants to hear it, when it just makes people uncomfortable, it becomes so much easier to smile and say “it was good!” and move on. Who wants to be a buzzkill? 

Slow librarianship puts worker well-being over productivity and deadlines, allows workers to be whole people at work, and supports a culture of care. While a radical idea, this even makes good business sense because depleted and burned out workers have been shown to be a major drain on the organization and negatively impact the culture. If you’re a manager and you’re not actively fostering a culture where people can bring their whole selves to work, then you are fostering a culture where people do not feel safe being vulnerable and having needs. Workers, if you know a colleague is struggling with something (an illness, losing a loved one, a difficult caregiving situation, etc.) and you don’t check in to see how things have been going for them, you’re sending the message that you don’t want to hear about these things, that they make you uncomfortable, that they’re not appropriate to take to work. I think we’ve all probably been guilty of this at some point in our lives and maybe we even thought that not asking was the right thing to do. I can imagine some people think that asking is invasive or reminds the person that they are sick, but it is an expression of care. As Philip Hoover writes in his excellent Sick Times article entitled “You know someone with Long COVID. They need you to ask about it genuinely” 

Approach us with empathy and curiosity. Ask questions that show a sincere desire to understand. How are you feeling this week? works because it acknowledges the chronic, fluctuating nature of my health. Or a version of Nunez’s question, and one I’ve longed to be asked since I’ve been ill: What has this been like for you? 

If our response is tough to hear, try not to smother it in optimism. And tread lightly. While some long-haulers may appear okay in public, much of our suffering occurs in private, shade-drawn rooms, across lonely afternoons, stuck in bed. When in doubt, remember that the act of asking never hurts — but never being asked certainly does.

While I don’t have Long COVID, this piece so perfectly encapsulates how I feel as someone with a mostly invisible chronic illness who would just love to be asked “how are you feeling this week?” instead of feeling like I have to pretend I’m okay. I told my manager at the start of Fall term that my autoimmune disease had become significantly worse and that I didn’t know what my capacity might look like going forward. She expressed sympathy, told me to take what time I needed, and never checked in with me after that. That September, I was taking over a very important committee chair role that was made enormously more time-consuming and onerous by the departure of the three colleagues most involved in supporting this work (two of whom had more than 20 years of institutional knowledge locked up in their heads). I didn’t feel like I had the leeway or support to let things drop as more and more kept piling up on my plate related to my chair role and it became clear that I was expected to do a lot of onboarding for the new people in this role even though I was new to my role and was no one’s manager. While I know my boss is extraordinarily busy, the message that not checking in with how I was doing sent me was very different from what I assume she’d wanted to convey. Checking in with a colleague or direct report seems like such a small thing, and it is in terms of the effort it requires, but the impact it can have in making someone feel cared for and less like they have to live a painful double life is enormous.

These are some things I’ve wandered across on the web this week.

Congratulations to All three Dr. J's from the @NirdsLab Dr. Bathsheba Jackson @sheissheba , Dr. Gavindya Jayawardena @Gavindya2 , and Dr. Yasith Jayawardana. @yasithdev. @WebSciDL PhD Crush board updated, just waiting in the border of alumni cloud until the paperwork is done! pic.twitter.com/dwEMMzOpy5

— Sampath Jayarathna (@OpenMaze) July 10, 2024

After the defense, @lesley_elis continued @WebSciDL's tradition of students sharing a hometown meal with us. Lesley's Philly roots showed up in the cheesesteaks (made authentic with Cheez Whiz), @Tastykake, birch beer soda, and @GrittyNHL decorations. Congrats again, Lesley! 🎉🎓 pic.twitter.com/WP8Ny1d2BY

— Michele Weigle (@weiglemc) July 29, 2024

I’m glad we’ve reached a new Public Domain Day, and that the works I’ve been featuring in my #PublicDomainDayCountdown, and many more, are now free to copy and reuse. I’ve been posting about works joining the public domain in the United States, which include sound recordings published in 1925, and other works published in 1930 that had maintained their copyrights. (Numerous works from 1930, and later, that had to renew their copyrights, and did not, were already in the public domain, though many of the best-known works did renew copyrights as required.) This is the eighth straight year Americans have seen a year’s worth of works join the public domain, after a 20-year freeze following a 1998 copyright extension.

I intend my countdown not just to celebrate the works joining the public domain, but also to celebrate what people have done with those works. In some posts, I note later creations based on those works. In nearly all my posts, I link to things that people have written about those works. Like the works themselves, those responses may have flaws or quirks, but I value them as human reactions to human creations. Whether they’re reviews, personal blog posts, professionally written essays, scholarly analyses, or Wikipedia articles, they’re created by people who encountered an interesting work and cared about it enough to craft a response to it and share it with the world. Those shared responses in turn pique my interest in the writers and the works.

It wasn’t always easy for me to find such responses online. Sometimes I’d go searching for responses to a promising-sounding work, and only find sales listings on e-commerce sites, social media posts not easily linkable or displayable without logging into a commercial platform, paywalled articles that many of my readers can’t view, or generic-sounding pages that read like they were generated by a large language model or a content farm, but not by anyone who I could clearly tell cared about or even read the work in question. Some works I initially hoped to feature got left off my countdown, replaced by other works where I could more readily link to an interesting response.

The people publishing the responses I link to are often swimming against a strong current online. Many online writing systems– including the one I’ve been writing these posts on— are now urging their users to “improve” their posts by letting “AI” write them. Some writers may be tempted to allow it, when facing an impending deadline or writer’s block or anxiety, even when the costs can include muffling one’s own voice, signing onto falsehoods confidently stated by a stochastic text generator, or abusively exploiting existing content and services. Other writers may feel pushed to put their work behind paywalls or other access controls that makes them less likely to be plagiarized or aggressively crawled by those same “AI” systems. And most writers, myself included, find it easy to dash off a quick short take on a social media platform, be quickly gratified by some “like”s, and then have it forgotten. It’s harder to take the time to craft something longer or more thought-out that will be readable for years, and that might take much longer for us to hear appreciated. The easy alternatives can discourage people from devoting their time to better, more lasting creations.

As I’ve noted before, both copyright and the public domain serve important purposes in encouraging the creation, dissemination, sharing, and reuse of literature and art. One reason I write my public domain posts is to promote a better balance between them, particularly in encouraging shorter copyright lengths to benefit both original creators and the public. Similarly, as I’ve noted in another recent post, I value both human creation and automated processes, but I increasingly see a need to improve the balance between those as well, especially as some corporations aggressively push “generative AI”. While I appreciate many ways in which automation can help us create and manage our work, I treasure the humanity that people thoughtfully put into the creation of literature and art of all kinds, and the human responses that those creations elicit.

Today I’m thankful for all of the people, most no longer with us, who made the works that are joining the public domain today. I’m thankful for the new opportunities we have to share and build on those works now that they’re public domain. I’m thankful to all the people who have responded to those works, whether as brief reactions or as new works as ambitious as the works they respond to. And I hope you’ll keeping making and sharing those responses with the world when you can. I look forward to reading them, and perhaps linking to them in future posts.

Personal Note, January 1 2026: I have a new job: Executive Director of the Project Gutenberg Literary Archive Foundation. Here's what I wrote for PG's January Newsletter.

Greetings from the new Executive Director

On October 21, Project Gutenberg lost its longtime leader, Greg Newby, to pancreatic cancer. I had agreed to step up as Acting Executive Director so that Project Gutenberg could continue the mission that had become Greg's life work: to serve and preserve public domain books so that all of us can use and enjoy them without restrictions. Although I've been doing development work for Project Gutenberg for the past 8 years, I did not really understand what Greg's job entailed, or how many tasks he had been juggling. Three months in, I'm still discovering mysterious-to-me aspects of the organization. I've also been amazed at the dedication and talent of the many volunteers behind Project Gutenberg and our sister organization, Distributed Proofreaders. And at the large number of donors who make the organization financially viable and sustainable. So as of 2026, with your support, I'm continuing as Executive Director.

In the past three months Project Gutenberg has proven to be resilient; we took a heavy blow and managed to keep going. My top priority going forward is to make Project Gutenberg even more sustainable as well as resilient. In other words, my job is be one runner in a relay race: take the baton and make sure I get it to the next runner. That's what we all have to do with public domain books, too. We want them to still be there in 50 years! Whether you're already a volunteer or booster, an avid reader, or just someone curious about what we do, I hope you'll help us pass that baton.

Source

showed that (i) a successful block-reverting attack does not necessarily require ... a majority of the hash power; (ii) obtaining a majority of the hash power ... costs roughly 6.77 billion ... and (iii) Bitcoin derivatives, i.e. options and futures, imperil Bitcoin’s security by creating an incentive for a block-reverting/majority attack.

Source

90% of transaction volume on the Bitcoin blockchain is not tied to economically meaningful activities but is the byproduct of the Bitcoin protocol design as well as the preference of many participants for anonymity ... exchanges play a central role in the Bitcoin system. They explain 75% of real Bitcoin volume.

The Attack

The key idea behind this strategy, called Selfish Mining, is for a pool to keep its discovered blocks private, thereby intentionally forking the chain. The honest nodes continue to mine on the public chain, while the pool mines on its own private branch. If the pool discovers more blocks, it develops a longer lead on the public chain, and continues to keep these new blocks private. When the public branch approaches the pool's private branch in length, the selfish miners reveal blocks from their private chain to the public.
...
We further show that the Bitcoin mining protocol will never be safe against attacks by a selfish mining pool that commands more than 1/3 of the total mining power of the network. Such a pool will always be able to collect mining rewards that exceed its proportion of mining power, even if it loses every single block race in the network. The resulting bound of 2/3 for the fraction of Bitcoin mining power that needs to follow the honest protocol to ensure that the protocol remains resistant to being gamed is substantially lower than the 50% figure currently assumed, and difficult to achieve in practice.

Given that the rule of thumb followed by most practitioners is to wait for 6 confirmations, a fork that goes 6 levels deep can very likely diminish the public’s trust in Bitcoin and cause a crash in its market price. It is also widely accepted that a prolonged majority attack (if it happens) would be catastrophic to the cryptocurrency and can cause its downfall.

The conventional wisdom in the blockchain community is to assume that such block-reverting attacks are highly unlikely to happen. The reasoning goes as follows:

1. Reverting multiple blocks and specifically double-spending a transaction that has 6 confirmations requires control of a majority of the mining power;
2. Having a majority of the mining power is prohibitively expensive and requires an outlandish investment in hardware;
3. Even if a miner, mining pool or group of pools does control a majority of the mining power, they have no incentive to act dishonestly and revert the blockchain, as that would crash the price of Bitcoin, which is ultimately not in their favor, since they rely on mining rewards denominated in BTC for their income.

Source

In short, an attacker can first use the Bitcoin derivatives market to short Bitcoin by purchasing a sufficient amount of put options or other equivalent financial instruments. She can then invest any of the amounts calculated above, depending on the timeline of the attack, to obtain the necessary hardware and hash power to perform the attack. If the attacker chooses to obtain a majority of the hash power, her success is guaranteed and she can revert the blocks as deeply as she wishes. However, she also has the option of a smaller upfront investment in hardware in exchange for longer wait times to achieve a high probability of success. In any case, as long as her earnings from shorting Bitcoin and then causing an intentional price crash outweighs her investments in hardware, there is a clear financial incentive to perform such an attack. The numbers above show that the annual trade volume in Bitcoin derivatives is more than three orders of magnitude larger than the required investment in hardware. Thus, it is possible and profitable to perform such an attack.

Assumptions

• We only consider the cost of hardware at the time of writing. We assume the attacker is buying the hardware, rather than renting it and do not consider potential discounts on bulk orders.
• We ignore electricity costs as they vary widely based on location.

The justification for the first assumption is that it keeps our analysis sound, i.e. we can only over-approximate the cost by making this assumption. As for the second assumption, we note that electricity costs are often negligible in comparison to hardware costs and that our main argument, i.e. the vulnerability of Bitcoin to majority attacks and block-reverting attacks, remains intact even if the estimates we obtain here are doubled. Indeed, as we will soon see, the trade volume of Bitcoin derivatives is more than three orders of magnitude larger than the numbers obtained here.

Goal

Attacker

• Outsiders: someone who has to acquire or rent sufficient hash power.
• Insiders: someone or some mining pool who already controls sufficient hash power.

• Obtaining and maintaining for the duration of the attack sufficient hash power without detection.
• Obtaining and maintaining for the duration of the attack a sufficient short position in Bitcoin without detection.

Hash Power

Outsider Attack

• Mining rigs.
• Power to run the rigs.
• Data center space to hold the rigs.

Mining rigs

• Could they acquire mining rigs sufficient to provide 30% of the combined insider and outsider hash power, or ~43% of the pre-attack hash power?
• How long would it take to acquire the rigs?
• Would their acquisition of the rigs be detected?

Source

Power

Data centers

Hyperion

Operational cost

Insider Attack

Source

Short Position

• When the attacker's chain is within one block of being the longest, they have ten minutes to purchase the shorts. There is unlikely to be enough liquidity in the market to accommodate this sudden demand, which in any case would greatly increase the price of the shorts. I will ignore this possibility in what follows.
• At the start of the attack the attacker gradually accumulates sufficient shorts. Even assuming there were enough liquidity, and that the purchases didn't increase the price, the attacker has to bear both the cost of maintaining the shorts for the duration of the attack, and the risk of the market moving up enough to cause the position to be liquidated.

At the time of writing, the open interest of BTC options is a bit more than 20 billion USD. Thus, a malicious party performing the attack mentioned in this work would need to obtain a considerable amount of the available put contracts. This may lead to market disruptions whose analysis is beyond the scope of this work. This being said, if the derivatives market continues to grow and becomes much larger than it currently is, purchasing this amount of contracts might not even be detected.

• Regulated exchanges such as the CME offering options on Bitcoin and stock exchanges with Bitcoin ETFs and Bitcoin treasury companies such as Strategy.
• Unregulated exchanges such as Binance offering "perpetual futures" (perps) on Bitcoin.

Unregulated Exchanges

Instead of all of a particular futures vintage settling on the same day, perps settle multiple times a day for a particular market on a particular exchange. The mechanism for this is the funding rate. At a high level: winners get paid by losers every e.g. 4 hours and then the game continues, unless you’ve been blown out due to becoming overleveraged or for other reasons (discussed in a moment).

Consider a toy example: a retail user buys 0.1 Bitcoin via a perp. The price on their screen, which they understand to be for Bitcoin, might be $86,000 each, and so they might pay $8,600 cash. Should the price rise to $90,000 before the next settlement, they will get +/- $400 of winnings credited to their account, and their account will continue to reflect exposure to 0.1 units of Bitcoin via the perp. They might choose to sell their future at this point (or any other). They’ll have paid one commission (and a spread) to buy, one (of each) to sell, and perhaps they’ll leave the casino with their winnings, or perhaps they’ll play another game.

Where did the money come from? Someone else was symmetrically short exposure to Bitcoin via a perp. It is, with some very important caveats incoming, a closed system: since no good or service is being produced except the speculation, winning money means someone else lost.

Perp funding rates also embed an interest rate component. This might get quoted as 3 bps a day, or 1 bps every eight hours, or similar. However, because of the impact of leverage, gamblers are paying more than you might expect: at 10X leverage that’s 30 bps a day.

In a standard U.S. brokerage account, Regulation T has, for almost 100 years now, set maximum leverage limits (by setting minimums for margins). These are 2X at position opening time and 4X “maintenance” (before one closes out the position). Your brokerage would be obligated to forcibly close your position if volatility causes you to exceed those limits.

Binance allows up to 125x leverage on BTC.

One reason perps are structurally better for exchanges and market makers is that they simplify the business of blowing out leveraged traders. The exact mechanics depend on the exchange, the amount, etc, but generally speaking you can either force the customer to enter a closing trade or you can assign their position to someone willing to bear the risk in return for a discount.

Blowing out losing traders is lucrative for exchanges except when it catastrophically isn’t. It is a priced service in many places. The price is quoted to be low (“a nominal fee of 0.5%” is one way Binance describes it) but, since it is calculated from the amount at risk, it can be a large portion of the money lost. If the account’s negative balance is less than the liquidation fee, wonderful, thanks for playing and the exchange / “the insurance fund” keeps the rest, as a tip.

In the case where the amount an account is negative by is more than the fee, that “insurance fund” can choose to pay the winners on behalf of the liquidated user, at management’s discretion. Management will usually decide to do this, because a casino with a reputation for not paying winners will not long remain a casino.

But tail risk is a real thing. The capital efficiency has a price: there physically does not exist enough money in the system to pay all winners given sufficiently dramatic price moves. Forced liquidations happen. Sophisticated participants withdraw liquidity (for reasons we’ll soon discuss) or the exchange becomes overwhelmed technically / operationally. The forced liquidations eat through the diminished / unreplenished liquidity in the book, and the magnitude of the move increases.

Risk in perps has to be symmetric: if (accounting for leverage) there are 100,000 units of Somecoin exposure long, then there are 100,000 units of Somecoin exposure short. This does not imply that the shorts or longs are sufficiently capitalized to actually pay for all the exposure in all instances.

In cases where management deems paying winners from the insurance fund would be too costly and/or impossible, they automatically deleverage some winners.

So perhaps you understood, prior to a 20% move, that you were 4X leveraged. You just earned 80%, right? Ah, except you were only 2X leveraged, so you earned 40%. Why were you retroactively only 2X? That’s what automatic deleveraging means. Why couldn’t you get the other 40% you feel entitled to? Because the collective group of losers doesn’t have enough to pay you your winnings and the insurance fund was insufficient or deemed insufficient by management.

In theory, this can happen to the upside or the downside. In practice in crypto, this seems to usually happen after sharp decreases in prices, not sharp increases. For example, October 2025 saw widespread ADLing as (more than) $19 billion of liquidations happened, across a variety of assets.

Source

Source

Regulated Exchanges

Conclusions

Outsider Attack

• They could not acquire 43% or more of the hash power.
• Even if they could it would take so long as to make detection inevitable.
• Even if they could and they were not detected, the high cost of the rigs makes the necessary shorts large relative to the open interest, and expensive to maintain.
• These large shorts would need to be leveraged perpetual futures, bringing significant risks of loss of collateral through liquidation, and of the potential payoff being reduced through automatic de-leveraging.
• The attacker would need more than the peak aggregate open interest in Bitcoin futures over the past year.

Insider Attack

Source

Source

mining-company stocks are still flying, even with cryptocurrency prices in retreat. That's because these firms have something in common with the hottest investment theme on the planet: the massive, electricity-hungry data centers expected to power the artificial-intelligence boom. Some companies are figuring out how to remake themselves as vital suppliers to Alphabet, Amazon, Meta, Microsoft and other "hyperscalers" bent on AI dominance.
...
Miners often have to build new, specialized facilities, because running AI requires more-advanced cooling and network systems, as well as replacing bitcoin-mining computers with AI-focused graphics processing units. But signing deals with miners allows AI giants to expand faster and cheaper than starting new facilities from scratch.
...
Shares of Core Scientific quadrupled in 2024 after the company signed its first AI contract that February. The stock has gained 10% this year. The company now expects to exit bitcoin mining entirely by 2028.

Counter-measures

Source

• The 6-block rule is just a convention, there is no dial that can be turned.
• Much of the access to the Bitcoin blockchain is via APIs that typically have the 6-block rule hard-codded in.
• Many, typically low-value, transactions do not wait for even a single confirmation.
• Even it were possible, changing from a one-hour to a four-hour confirmation would have significant negative impacts on the Bitcoin ecosystem.

Acknowledgements

Happy New Year

Blackeye peas and collard greens for us.

I don’t think any of us went into 2025 thinking this would be an easy year, and boy wasn’t it. Rather than restate our collective challenges, I’m going to stick to my own little life and keep it short with bullet points and pictures. I’ll also list as many positives as I can, because even in the darkest of times, there are sources of joy and reasons to be grateful.

Real life, in no particular order

• (OK, there’s some order, let’s do the hard one first) My mom passed away in late October – our relationship was complicated, and so are my feelings, but I’m working through it all with a therapist
• (Bittersweet) I adopted Mom’s green cheek conure, Tutu (which we jokingly spell “Teauxtu,” thanks to my brother), who bonded to me quickly and thoroughly, but who is currently plucking all of his feathers anyway — we’re working on it with his vet, and I’m hopeful we’ll get him past this (though I’ll love him anyway, even if he’s a naked bird forever; I cannot overstate how sweet a little guy he is!)
• My brother got married to a really nice lady, and I’m so happy for them!
• In early October, we moved from Midcoast Maine to Western NY, where there is a larger and more active covid-conscious community – because the economy was so unstable, it took months to sell our house in Maine, but we did finally succeed, the day before Christmas
• We joined the mask bloc here for a D&D night and had a lot of fun – looking forward to getting involved in mask distribution and attending more events in 2026
• I finally read all of the Murderbot Diaries series by Martha Wells, spurred on by the release of the TV show – then I listened to the audiobooks, and it became my comfort listen for the whole latter part of the year
• I ran an accessible birding event here in Rochester, and I’m hoping to run more next year
• I took American Sign Language 101 and 102, which I’ve really enjoyed. I’ll be retaking 102 in early 2026 (it was a rough autumn/winter), and then I hope to move on to 103 and 104.
• I learned to darn (as in, mending damaged knitted items), and I’m working on expanding my mending skills repertoire – this is well-timed, since, besides his own feathers and our hair, Tutu loves to bite on shirts until they have holes in them
• We saw the most vivid aurora we’ve ever seen, including during our time in Alaska
• My aunt sent me her scone recipe, and I made so. many. scones. this year

Bird Buddies! (Outdoor birds captured with our feeder cameras.)

The three baby bluebirds in the video below brought us so much joy this spring! Here, they appear to have a little conference, to decide whether it’s time to fly away or not, before a starling shows up.

Bird Buddies who live indoors

Miscellaney!

Work, in no particular order

• I ran the first part of a major, potentially multi-year LibGuides accessibility remediation project (“potentially” because there’s a task force considering how important LibGuides are to us and our patrons, and maybe we will shift focus away from them or have a single accessibility remediator or … something I haven’t thought of, who knows?)
• I co-ran a task force focused on accessibility training for library staff
• We survived a university-wide “Cybersecurity Alignment” (or the first part of one?)
• I learned the developer side of Figma (kind of, I still find parts of it baffling)
• I learned a ridiculous amount about authentication to electronic resources (which would be really satisfying if I didn’t keep running into things I still don’t know)

Coming up in 2026, an incomplete list

• I’ll take more ASL classes
• I’ll mend some things
• I’ll ride my bike more, because there are more safe places to do so where I live!
• I’m hoping to work through more herbal studies and practice more of those skills – it may become vital in the near future
• There will be an IT “alignment” at work mid-year – it could mean anything from “nothing” to “now I report to Central IT (who doesn’t care about my MLIS and might actually consider it a negative) instead of the Library (who doesn’t require an MLIS for my position, but knows it’s a strength, and also my team is so good, I don’t want to lose them, OK, yes, I am very stressed about this possibility)”
• Our landlord will be selling our house, something I think (hope) he didn’t realize he’d be doing when we started renting here – we’ll get first right of refusal on buying it, and he’s willing to write us a lease for as long as we want one if we choose not to buy (NY law means the new owner will have to honor it), so while it’s a stressful situation, it’s not as bad as it could be
• A valued colleague (everyone in my department is valued, truly, but his institutional knowledge is unmatched) will be retiring, and because of budget constraints, that will probably mean some shuffling of responsibilities semi-permanently
• If we decide we aren’t moving (so, we decide to buy this house or to sign a longer lease), I’ll set up some garden beds and grow some things
• I want this to be a thing, but financially, it may not: I really very much want to build or buy a small (like, Class B, or C at the highest) camper so that I can travel again

Ninety-five (or 100) years is a very long time for copyrights to last. But Olaf Stapledon saw a much longer future for us in Last and First Men (reviewed here and here). His book tells a story of successive human species over the next 2 billion years.

Stapledon died in 1950, and his work is already public domain most places outside the US. Tomorrow a copy in Australia will be among the first books to be relisted in my new books listing, finally free for all. #PublicDomainDayCountdown

In January 2021, I began a journey that would span nearly five years, three children, countless late nights, and a singular focus: teaching machines to extract data from complex scientific tables with confidence—and to know when they're uncertain. On October 29, 2025, I successfully defended my dissertation titled "SCITEUQ: Toward Uncertainty-Aware Complex Scientific Table Data Extraction and Understanding" at Old Dominion University. This milestone represents not just the culmination of intensive research but a testament to perseverance, family support, and the power of focused determination.

Finding My Path at LAMP-SYS

When I joined the Lab for Applied Machine Learning and Natural Language Processing Systems (LAMP-SYS), part of ODU's Web Science and Digital Libraries Research Group (WSDL) under the guidance of Dr. Jian Wu, I knew exactly what problem I wanted to solve: making scientific table data extraction both accurate and trustworthy through uncertainty quantification.

Scientific tables are ubiquitous in research papers, containing critical experimental data, statistical results, and research findings. Yet extracting this data automatically from PDF documents remains surprisingly difficult. Unlike the simple, well-structured tables you might see on Wikipedia, scientific tables are complex beasts—featuring multi-level headers, merged cells, irregular layouts, and domain-specific notations that confound even state-of-the-art machine learning models.

But here's the real problem: existing methods don't tell you when they're wrong. They extract data with the same confidence whether they're processing a simple table or struggling with a complex one. For scientific applications where accuracy is paramount, this means researchers must manually verify every single extracted cell—a task that doesn't scale when you're dealing with thousands of tables.

The Research Challenge

My research addressed a fundamental question: How can we build systems that not only extract data from complex scientific tables but also quantify their uncertainty, allowing us to focus human verification effort only where it's needed?

To tackle this challenge, I formulated four research questions:

RQ1: What is the status of reproducibility and replicability of existing TSR models?

Before building something new, I needed to understand what already existed. I conducted the first systematic reproducibility and replicability study of 16 state-of-the-art Table Structure Recognition (TSR) methods. The results were sobering: only 8 of 16 papers made their code and data publicly available, and merely 5 had executable code. When I tested these methods on my newly created GenTSR dataset (386 tables from six scientific domains), none of the methods replicated their original performance. This highlighted a critical gap in the field. This work was published at ICDAR 2023: "A Study on Reproducibility and Replicability of Table Structure Recognition Methods."

RQ2: How do we quantify the uncertainties of TSR results?

To address this, I developed TTA-m, a novel uncertainty quantification pipeline that adapts Test-Time Augmentation specifically for TSR. Unlike vanilla TTA, my approach fine-tunes pre-trained models on augmented table images and employs ensemble-based methods to generate cell-level confidence scores. On the GenTSR dataset, TTA-m achieved an F1-score of 0.798, with over 80% accuracy for high-confidence predictions—enabling reliable automatic detection of extraction errors. This work was published at IEEE IRI 2024: "Uncertainty Quantification in Table Structure Recognition."

RQ3: How can we integrate uncertainties from TSR and OCR for holistic table data extraction?

I designed and implemented the TSR-OCR-UQ framework, which integrates table structure recognition (using TATR), optical character recognition (using PaddleOCR), and conformal prediction-based uncertainty quantification into a unified pipeline. The results were compelling: the accuracy improved from 53-71% to 83-97% for different complexity levels, with the system achieving 69% precision in flagging incorrect extractions and reducing manual verification labor by 53%. This work was published at ICDAR 2025: "Uncertainty-Aware Complex Scientific Table Data Extraction."

RQ4: How well do LLMs answer questions about complex scientific tables?

To evaluate the QA capability of Large Language Models on scientific tables, I created SciTableQA, a benchmark dataset containing 8,700 question-answer pairs across 320 complex scientific tables from multiple domains. My evaluation revealed that while GPT-3.5 achieved 79% accuracy on cell selection TableQA tasks, performance dropped to 49% on arithmetic reasoning TableQA tasks—highlighting significant limitations of current LLMs when dealing with complex table structures and numerical reasoning. This work was published at TPDL 2025: "SciTableQA: A Question-Answering Benchmark for Complex Scientific Tables."

The SCITEUQ Framework

Putting it all together, SCITEUQ (Scientific Table Extraction with Uncertainty Quantification) represents a comprehensive solution to uncertainty-aware scientific table data extraction. The framework achieves the state-of-the-art performance while providing essential uncertainty quantification capabilities that enable efficient human-in-the-loop verification.

Each component contributes to a more reliable approach:

• GenTSR provides rigorous cross-domain evaluation
• TTA-m quantifies uncertainties in structure recognition
• TSR-OCR-UQ integrates structure and content extraction with uncertainty maps
• SciTableQA enables systematic evaluation of reasoning capabilities

Publications and Research Impact

My research resulted in five first-author publications at top-tier conferences and journals:

• ICDAR 2025: "Uncertainty-Aware Complex Scientific Table Data Extraction."
• TPDL 2025: "SciTableQA: A Question-Answering Benchmark for Complex Scientific Tables."
• IEEE IRI 2024: "Uncertainty Quantification in Table Structure Recognition."
• Nature Scientific Data 2023: "DeepPatent2: A Large-Scale Benchmarking Corpus for Technical Drawing Understanding."
• ICDAR 2023: "A Study on Reproducibility and Replicability of Table Structure Recognition Methods."

Each of these papers faced initial rejection before ultimately being accepted. This taught me an invaluable lesson: rejection is not failure; it's an opportunity to refine and improve your work.

Industry Experience: From Azure to Alexa to Microsoft AI

While my research focused on scientific tables, my internships at Microsoft and Amazon broadened my perspective on applying machine learning at scale.

Microsoft (Summers 2022, 2023, 2025)

My first two summers at Microsoft were with the Azure team, where I worked on infrastructure optimization problems far from my research area. I developed an AI-human hybrid LLM-based multi-agent system for AKS Cluster configuration, reducing cluster type generation time from 2 weeks to 1 hour (link to blog post). I also designed ML anomaly detection systems on Azure Synapse that reduced hardware maintenance costs by over 20% and formulated new metrics for characterizing node interruption rates that decreased hardware downtime by 25% (link to blog post).

In Summer 2025, I joined the Microsoft AI team under the Bing organization, working on problems at the intersection of large-scale search and AI—which is what I'll be doing when I return to Microsoft full-time in January 2026.

Amazon (Summer 2024)

At Amazon, I worked with the Alexa Certification Technology team in California, where I drove 10% customer growth by designing LLM-based RAG systems with advanced prompt engineering techniques and increased the revenue by over 5% by developing LLM Agents on AWS to improve Alexa-enabled applications (link to blog post).

These internships, while not directly related to my dissertation research, taught me how to apply ML thinking to diverse industrial problems and to work effectively in large, complex organizations.

Balancing PhD Life with Family

Perhaps the most challenging aspect of my PhD journey had nothing to do with research—it was combining my studies with raising three young children. My youngest son, Daniel, was born just six month after I was enrolled in the PhD program. Managing research deadlines, experimental runs, paper submissions, and the demands of parenting three boys (Paul, David, and Daniel) required discipline and sacrifice.

I developed a strict routine: work from 9 AM to 3 PM every day at my research lab, then pick up my kids from school and be fully present for them. This meant no late nights in the lab, no weekend marathons of coding—just consistent, focused work during designated hours. It wasn't always easy. Conference deadlines sometimes meant asking my wife, Olabisi, to take on even more, or my mother, Beatrice, to provide extra support. But this routine kept me grounded and taught me that quality of work matters more than quantity of hours.

The Defense

On October 27, 2025, I defended my dissertation before my committee:

• Dr. Jian Wu (Director) - Old Dominion University
• Dr. Michael L. Nelson - Old Dominion University
• Dr. Michele C. Weigle - Old Dominion University
• Dr. Sampath Jayarathna - Old Dominion University
• Dr. Yi He (Co-advisor) - College of William & Mary

Their thoughtful feedback, probing questions, and constructive critiques throughout my PhD journey were instrumental in refining my research and pushing me to think deeper about the implications and limitations of my work.

Lessons Learned

Looking back on nearly five years of doctoral work, several lessons stand out:

1. Embrace Rejection as Refinement

Four of my papers were initially rejected. Each rejection stung, but each one ultimately led to a stronger paper. The review process, while sometimes frustrating, forced me to clarify my arguments, strengthen my experiments, and address weaknesses I hadn't noticed. My TPDL 2025 paper on SciTableQA went through two rounds of revisions, but the final version is significantly better than the original submission.

2. Establish Non-Negotiable Boundaries

My 9 AM to 3 PM schedule wasn't just convenient—it was essential for maintaining my sanity and my family relationships. While some might argue that PhD students need to work 80-hour weeks, I proved that focused, disciplined work during reasonable hours can produce quality research. Those boundaries also made me more efficient: when you only have six hours a day, you learn to prioritize ruthlessly.

3. Build for Reproducibility from Day One

My systematic study on TSR reproducibility taught me the hard way how difficult it is to reproduce other people's work. This experience shaped how I approached my own research. Every framework I built—TTA-m, TSR-OCR-UQ, SciTableQA—comes with comprehensive documentation, publicly available code, and clear instructions for replication. Future researchers shouldn't struggle to build upon my work the way I struggled with others'.

4. Choose Problems That Matter to You

I entered my PhD knowing I wanted to work on table extraction with uncertainty quantification, and I never wavered from that focus. This singular vision helped me navigate the inevitable setbacks and distractions that come with doctoral research. When experiments failed or papers got rejected, I could always return to the core question: How do we make scientific data extraction both accurate and trustworthy?

5. Internships Broaden Your Perspective

While my Microsoft and Amazon internships didn't directly contribute to my dissertation, they fundamentally shaped how I think about research. Working on production systems with millions of users taught me to think about scalability, robustness, and real-world constraints in ways that academic research rarely emphasizes. These experiences make me a better researcher because I can now evaluate my work not just on benchmark performance, but on whether it could actually be deployed at scale.

Looking Forward

In January 2026, I'll be joining Microsoft as a Data Scientist 2 with the Microsoft AI team at the Redmond campus in Washington state. My family and I are excited about this new chapter—moving from Norfolk, Virginia, to the Pacific Northwest, and transitioning from academic research to industry applications.

While I'll be working on different problems at Microsoft, the skills and mindset I developed during my PhD—rigorous experimentation, systematic evaluation, uncertainty quantification, and reproducible research—will continue to guide my work. I'm particularly excited about the opportunity to apply research-driven thinking to real-world problems at a scale that can impact millions of users.

Acknowledgments

This journey would have been impossible without extraordinary support:

To Dr. Jian Wu, my advisor, mentor, and guide—thank you for believing in my research vision, for pushing me to think bigger, and for your patience during the inevitable frustrations of doctoral research. Your mentorship has not only shaped my research but also my approach to solving complex problems.

To Dr. Yi He, my co-advisor at William & Mary, your expertise and thoughtful feedback greatly enriched this research. Thank you for your guidance and support throughout this journey.

To my dissertation committee—Drs. Michael Nelson, Michele Weigle, and Sampath Jayarathna—your constructive critiques and expert insights were essential in refining my ideas and strengthening this work.

To my colleagues in WSDL and LAMP-SYS, the collaborative environment, intellectual exchanges, and camaraderie made this journey both enriching and memorable.

To my wife, Olabisi—you walked beside me every step of this journey with unwavering devotion and love. Your patience during the long hours, your understanding through the challenges, and your constant encouragement when the path seemed difficult made this achievement possible. This accomplishment is as much yours as it is mine.

To my sons—Paul, David, and Daniel—you are my greatest blessings and my constant source of joy and motivation. I hope this work serves as an example that with dedication and faith, you too can achieve your dreams.

To God Almighty, who is the source of all wisdom and strength, I give thanks and praise.

-Kehinde Ajayi (@KennyAJ)

Recorded in a Ohio service center somewhere along I-80.

The muted sound of cars and trucks passing at high speed about 200 feet away.

This is a snippet cut and pasted a few times in REAPER:

Yale’s 1905 commencement ceremonies included a honorary doctorate for British composer Edward Elgar, and a portion of his first “Pomp and Circumstance” march. It’s been a staple of graduation processions ever since. The full suite of five marches that Elgar finished takes about 30 minutes to play, but took nearly three decades to complete. The first march has long been in the US public domain; the last, published in 1930, joins it there in two days. #PublicDomainDayCountdown

David Seubert writes that more than 2500 records from 1925 digitized by the UC Santa Barbara Library will soon be freely downloadable there. A full listing of these recordings is online (though note that recordings made in 1925 but not released that year won’t be public domain yet).

A highlight of the collection is “St. Louis Blues”, sung by Bessie Smith with Louis Armstrong on cornet. One of the top selling records of 1925, it will be public domain in 3 days. #PublicDomainDayCountdown

Marlene Dietrich enjoyed success on stage and screen in 1920s Berlin, but became an international star in 1930. That year she came to the United States to star in Morocco alongside Gary Cooper. Her performance was nominated for an Academy Award. So was the direction by Josef von Sternberg, who also directed her in The Blue Angel and several other films. Morocco was inducted into the National Film Registry in 1992, and will be inducted into the public domain in 4 days. #PublicDomainDayCountdown

Source

A gauge of risk on Oracle Corp.’s (ORCL) debt reached a three-year high in November, and things are only going to get worse in 2026 unless the database giant is able to assuage investor anxiety about a massive artificial intelligence spending spree, according to Morgan Stanley.

A funding gap, swelling balance sheet and obsolescence risk are just some of the hazards Oracle is facing, according to Lindsay Tyler and David Hamburger, credit analysts at the brokerage. The cost of insuring Oracle Corp.’s debt against default over the next five years rose to 1.25 percentage point a year on Tuesday, according to ICE Data Services.

The company borrowed $18 billion in the US high-grade market in September. Then in early November, a group of about 20 banks arranged a roughly $18 billion project finance loan to construct a data center campus in New Mexico, which Oracle will take over as tenant.

Banks are also providing a separate $38 billion loan package to help finance the construction of data centers in Texas and Wisconsin developed by Vantage Data Centers,

Source

When you run using the latest, say, an Nvidia chip for training a model, those things are being run flat out, 24 hours a day, seven days a week, which is why they're liquid-cooled, they're inside of these giant centers where one of your primary problems is keeping them all cool. It's like saying ‘I bought a used car and I don't care what it was used for.’ Well, if it turns out it was used by someone who was doing like Le Mans 24 hours of endurance with it, that's very different even if the mileage is the same as someone who only drove to church on Sundays.

These are very different consequences with respect to what's called the thermal degradation of the chip. The chip's been run hot and flat out, so probably its useful lifespan might be on the order of two years, maybe even 18 months. There's a huge difference in terms of how the chip was used, leaving aside whether or not there's a new generation of what's come along. So it takes us back to these depreciation schedules.

73% of Ethereum miners have just given up: “About 10.6 million RTX 3070 equivalents have stopped mining since the merge.”

We strongly recommend that you do not hit eBay for a cheap video card, despite the listings reassuring you that this card was only used by a little old lady to play Minecraft on Sundays and totally not for crypto mining, and that you should ignore the burnt odor and the charred RAM. Unless you’re poor, and the card’s so incredibly cheap that you’re willing to play NVidia Roulette.

How well do miners treat their precious babies? “GPU crypto miners in Vietnam appear to be jet washing their old mining kit before putting the components up for sale.” There are real cleaning methods that involve doing something like this with liquid fluorocarbons — but the crypto miners seem to be using just water.

Conduit debt financing is a structure where an intermediary entity (the “conduit”) issues debt securities to investors and passes the proceeds through to an end borrower. The key feature distinguishing conduit debt from regular corporate bonds is that the conduit issuer has no substantial operations or assets beyond the financing transaction itself. The conduit is purely a pass-through vehicle, the debt repayment relies entirely on revenues or assets from the ultimate borrower.

Think of it this way: Company A wants to borrow money but doesn’t want that debt appearing on its balance sheet or affecting its credit rating. So it works with a conduit entity, Company B, which issues bonds to investors. Company B takes that capital and uses it to build infrastructure or acquire assets that Company A needs. Company A then enters into long-term lease or service agreements with Company B, and those payments service the debt. On paper, Company A is just a customer making payments, not a debtor owing bondholders.

The structure creates separation. The conduit issuer’s creditworthiness depends on the revenue stream from the end user, not on the conduit’s own balance sheet (because there isn’t really one). This is why conduit debt is often referred to as “pass-through” financing, the economics flow through the conduit structure to reach the underlying obligor.

Legal risks when things break: Substantive consolidation (court merges conduit with sponsor), recharacterization (lease treated as secured financing), and fraudulent transfer challenges. The structures haven’t been stress-tested yet because hyperscalers are wildly profitable. But if AI monetization disappoints or custom silicon undercuts demand, we’ll discover whether bondholders have secured claims on essential infrastructure or are functionally unsecured creditors of overleveraged single-purpose entities.

Why would Meta finance this via the project finance markets? And why does it cost $6.5 billion more?

That’s how much more Meta is paying to finance this new AI data center using the project finance market versus what they could have paid had they used traditional corporate debt. So why on earth is this being called a win? And even crazier, why are other AI giants like Oracle and xAI looking to copy it?

Meta data center

Construction on the project was well under way when Meta announced a new financing deal last month. Meta moved the project, called Hyperion, off its books into a new joint venture with investment manager Blue Owl Capital. Meta owns 20%, and funds managed by Blue Owl own the other 80%. Last month, a holding company called Beignet Investor, which owns the Blue Owl portion, sold a then-record $27.3 billion of bonds to investors, mostly to Pimco.

Meta said it won’t be consolidating the joint venture, meaning the venture’s assets and liabilities will remain off Meta’s balance sheet. Instead Meta will rent the data center for as long as 20 years, beginning in 2029. But it will start with a four-year lease term, with options to renew every four years.

This lease structure minimizes the lease liabilities and related assets Meta will recognize, and enables Meta to use “operating lease,” rather than “finance lease,” treatment. If Meta used the latter, it would look more like Meta owns the asset and is financing it with debt.

The joint venture is what is known in accounting parlance as a variable interest entity, or VIE for short. That term means the ownership doesn’t necessarily reflect which company controls it or has the most economic exposure. If Meta is the venture’s “primary beneficiary”—which is another accounting term of art—Meta is required to consolidate it.

Under the accounting rules, Meta is the primary beneficiary if two things are true. First, it must have “the power to direct the activities that most significantly impact the VIE’s economic performance.” Second, it must have the obligation to absorb significant losses of the VIE, or the right to receive significant benefits from it.

Blue Owl has control over the venture’s board. But voting rights and legal form aren’t determinative for these purposes. What counts under the accounting rules is Meta’s substantive power and economic influence. Meta in its disclosures said “we do not direct the activities that most significantly impact the venture’s economic performance.” But the test under the accounting rules is whether Meta has the power to do so.

The second test—whether Meta has skin in the game economically—has an even clearer answer. Meta has operational control over the data center and its construction. It bears the risks of cost overruns and construction delays. Meta also has provided what is called a residual-value guarantee to cover bondholders for the full amount owed if Meta doesn’t renew its lease or terminates early.

Today, Nvidia's rack systems are hovering around 140kW in compute capacity. But we've yet to reach a limit. By 2027, Nvidia plans to launch 600kW racks which pack 576 GPU dies into the space one occupied by just 32.

To get ahead of this trend toward denser AI deployments, Digital Realty announced a research center in collaboration with Nvidia in October.

The facility, located in Manassas, Virginia, aims to develop a new kind of datacenter, which Nvidia CEO Jensen Haung has taken to calling AI factories, that consumes power and churn out tokens in return.

Another judgment call: Under the accounting rules, Meta would have to include the residual-value guarantee in its lease liabilities if the payments owed are “probable.” That could be in tension with Meta’s assumption that the lease renewal isn’t “reasonably certain.”

If renewal is uncertain, the guarantee is more likely to be triggered. But if the guarantee is triggered, Meta would have to recognize the liability.

Ultimately, the fact pattern Meta relies on to meet its conflicting objectives strains credibility. To believe Meta’s books, one must accept that Meta lacks the power to call the shots that matter most, that there’s reasonable doubt it will stay beyond four years, and that it probably won’t have to honor its guarantee—all at the same time.

David Sacks Nov 6

OpenAI explicitly requested federal loan guarantees for AI infrastructure in an October 27 letter to the White House—which kindly refused the offer, with AI czar David Sacks saying that at least 5 other companies could take OpenAI’s place—directly contradicting CEO Sam Altman's public statements claiming the company doesn't want government support.

The 11-page letter, submitted to the Office of Science and Technology Policy, called for expanding tax credits and deploying "grants, cost-sharing agreements, loans, or loan guarantees to expand industrial base capacity" for AI data centers and grid components. The letter detailed how "direct funding could also help shorten lead times for critical grid components—transformers, HVDC converters, switchgear, and cables—from years to months."

Trump unveiled the “Genesis Mission” as part of an executive order he signed Monday that directs the Department of Energy and national labs to build a digital platform to concentrate the nation’s scientific data in one place.

It solicits private sector and university partners to use their AI capability to help the government solve engineering, energy and national security problems, including streamlining the nation’s electric grid, according to White House officials who spoke to reporters on condition of anonymity to describe the order before it was signed.

• Mr. Sacks has offered astonishing White House access to his tech industry compatriots and pushed to eliminate government obstacles facing A.I. companies. That has set up giants like Nvidia to reap an estimate of as much as $200 billion in new sales.
• Mr. Sacks has recommended A.I. policies that have sometimes run counter to national security recommendations, alarming some of his White House colleagues and raising questions about his priorities.
• Mr. Sacks has positioned himself to personally benefit. He has 708 tech investments, including at least 449 stakes in companies with ties to artificial intelligence that could be aided directly or indirectly by his policies, according to a New York Times analysis of his financial disclosures.
• His public filings designate 438 of his tech investments as software or hardware companies, even though the firms promote themselves as A.I. enterprises, offer A.I. services or have A.I. in their names, The Times found.
• Mr. Sacks has raised the profile of his weekly podcast, “All-In,” through his government role, and expanded its business.

Steve Bannon, a former adviser to Mr. Trump and a critic of Silicon Valley billionaires, said Mr. Sacks was a quintessential example of ethical conflicts in an administration where “the tech bros are out of control.”

“They are leading the White House down the road to perdition with this ascendant technocratic oligarchy,” he said.

David Sacks Nov 24

“The way this works”, said an investor friend to me this morning: “is that when Nvidia is about to miss their quarter, Jen Hsun calls David Sacks, who then gets this government initiative to place a giant order for chips that go into a warehouse.”

I obviously can’t confirm or deny that actually happened. My friend might or might not have been kidding. But either way the White House’s new Science and AI program, Genesis, announced by Executive Order on Monday, does seem to involve the government buying a lot of chips from a lot of AI companies, many of which are losing money.

And David Sack’s turnaround from “read my lips, no AI bailout” (November 6) to “we can’t afford to [let this all crash]” tweet (November 24) came just hours before the Genesis announcement.

• OpenAI, Anthropic and xAI, none of whom have a viable business model.
• Meta, Google and Microsoft, all of whom are pouring the cash from their viable business models into this non-viable business,

Update: 28^th December 2025

Source

If I were advising Xi, I’d counsel him to go for the jugular by engaging in AI-dumping, a repeat of their aughts steel-dumping playbook. It’s already underway — and working. Eighty percent of a16z startups use open-source Chinese models. Same story at Airbnb. China is registering similar or better performance as the American LLM leaders, but with a fraction of the capex. Flooding the market with competitive, less-expensive AI models will put pressure on the margins and pricing power of the Mag 7, taking down a frighteningly concentrated S&P and likely sending the U.S., possibly the globe, into recession.

Inspired by Tom Whitwell's 52 things I learned in 2022, I started my own list of things I learned in 2023 and repeated it last year. Reaching the end of another year, it is time for Things I Learned In 2025. Part way through the year I had the brilliant idea of putting a learning at the bottom of my weekly newsletter, and that worked well until the middle of the year when I stopped publishing newsletter issues. So here is a half year of learnings.

What did you learn this year? Let me know on Mastodon or Bluesky.

In Ethiopia, time follows the sun like nowhere else

This could have easily gone in the Thursday Threads on time standards. There are 12 hours of daylight, numbered 1 through 12. Then 12 hours of night, numbered 1 through 12. What could be easier?

From Thursday Threads issue 104 on Long Term Digital Storage.

A biographer embedded with the Manhattan Project influenced what we think about the atomic bomb

The quote above comes from the transcript of this podcast episode. I've thought about this a lot in the past week as the Trump administration's flood-the-zone strategy overwhelms the senses. In a valiant effort to cover everything that is news, I can't help but wonder about the lost perspective of what isn't being covered. And I wonder where I can look to find that perspective.

From Thursday Threads issue 105 on Facial Recognition.

The origin of the computer term "mainframe" comes from "main frame" — the 1952 name of an IBM computer's central processing section

"Mainframe" is such a common word in my lexicon that it didn't occur to me that its origins was from "main frame" — as in the primary frame in which everything else connected. I've heard "frame" used to describe a rack of telecommunications equipment as well, but a quick Kagi search couldn't find the origins of the word "frame" from a telecom perspective.

From Thursday Threads issue 106 on How much do you know about the credit card industry?.

It takes nearly 3¢ to make a penny, but almost 14¢ to make a nickel

I knew pennies cost the U.S. mint more than one cent to make, but I didn't realize that the cost of nickels is so much more out of whack. I also learned a new word: seigniorage — the difference between the face value of money and the cost to produce it.

From Thursday Threads issue 107 on the humble battery. Also this year, the U.S. mint stopped pressing pennies in November.

It is much harder to get to the Sun than it is to Mars

I suppose it that headline above needs some nuance. It is easy to get to the Sun...just escape Earth's gravity and point yourself there. It is hard to get to the Sun in a controlled way that means you won't burn up along the way.

From Thursday Threads issue 108 on Educational Technology.

There are now 23 Dark Sky Sanctuaries in the World

The pictures that accompany this article from the New York Times are stunning (gift link). And to think that there are only 23 places in the world that have reached this level of commitment to the environment.

From Thursday Threads issue 109 on Generative AI in Libraries.

Mexico has only one gun store for the entire country

And not only is there one gun store, the single store in Mexico is located on an army base and is run by soldiers, according to an article in the Associated Press from 2016.

From Thursday Threads issue 110 on Research into Generative AI.

Plants reproduce by spreading little plant-like things

From Thursday Threads issue 111 on End-to-end Encryption.

Most plastic in the ocean isn't from littering, and recycling will not save us

Contrary to popular belief, most plastic in the Great Pacific Garbage Patch stems from the fishing industry, with only a small fraction linked to consumer waste. The video highlights that mismanaged waste, rather than individual littering, is the primary contributor to plastic pollution, with 82% of macroplastic leakage resulting from this issue. It emphasizes the ineffectiveness of recycling as a solution, noting that less than 10% of plastics are currently recycled, and the industry has perpetuated the myth that recycling can resolve the plastic crisis. Microplastics, which are increasingly recognized as a major problem, originate from various sources, including tires and paint, with new data suggesting that paint is a significant contributor.

From Thursday Threads issue 112 on Social Media Research.

"But where is everybody?!?" — the origins of Fermi's Paradox

This retelling of the Fermi Paradox coms from this story about why, despite the vastness of the universe, we have yet to encounter evidence of extraterrestrial civilizations. Enrico Fermi famously posed the question, "Where is everybody?" suggesting a disconnect between the expectation of abundant intelligent life and the lack of observable evidence. With this comes the Great Filter notion...proposing that there may be significant barriers preventing intelligent life from becoming spacefaring. The article goes on to speculate where we are relative to the "Great Filter" — are we past it, or is it yet in front of us? In other words, have we survived the filter or is our biggest challenge ahead of us?

From Thursday Threads issue 113 on Copyright and Foundational AI Models.

The pronoun "I" was capitalized to distinguish it from similarly typset letters

I'm not buying the opinion author's underlying premise (capitalizing “they” in writing when it refers to a nonbinary person), but the origins of why we capitalize "I" and not other pronouns are fascinating.

From Thursday Threads issue 114 on Digital Privacy.

The word "scapegoat" originated in a 1530 bible translation

From Thursday Threads issue 115 on Public and Private Camera Networks.

"Leeroy Jenkins!!!!" was staged

First posted on May 10, 2005, this year marks the 20th anniversary of this bit of internet folklore. I remember when this first came out, and I totally believed it was real until earlier this year.

From Thursday Threads issue 116 on Government Surveillance.

Ammonium chloride may be the 6th basic taste

From Thursday Threads issue 117 on Local Government Surveillance.

Struggle over academic freedom in Texas state universities has a long history. Today it’s often over race and gender; in the 1940s, it was over things like John Dos Passos’s USA trilogy. When the University of Texas Board of Regents banned it from classrooms, university president Homer Price Rainey objected to their interference. After they fired him, thousands protested on campus. The first part of the USA trilogy, The 42nd Parallel, joins the public domain in 5 days. #PublicDomainDayCountdown

The Penn Libraries, where I work, has first editions of many of the works featured in my #PublicDomainDayCountdown . From today through Public Domain Day, the Libraries social media will feature photos of some distinctive books from 1930.

One of featured photos, also shown here, is of the 1930 edition of W. H. Auden‘s Poems, which made his work known to readers worldwide. Glynn Young writes about the “brilliant collection” of 30 poems, and one verse drama, joining the public domain in 6 days.

In 1925 the Associated Glee Clubs of America put on a concert like no other. 15 choral groups, with over 850 singers in all, came together in New York’s Metropolitan Opera House to sing a program broadcast on radio across America. Portions were electrically recorded, including “Adeste Fideles”, where the audience of 4000 joined in the carol. Lloyd Winstead writes about the record, which is now in the National Recording Registry, and joins the public domain in 7 days. #PublicDomainDayCountdown

Moss Hart wrote the first draft of Once in a Lifetime, a comedy about Hollywood’s transition to “talkies”, as a 25-year-old unknown. Established playwright George S. Kaufman helped revise it into a Broadway hit. Steve Vineberg calls it the “the finest comedy ever written by Americans”, and discusses how it began a long successful collaboration between the two writers. Still making theatergoers laugh in the 21st century, the 1930 play joins the public domain in 8 days. #PublicDomainDayCountdown

Given how much “Dick and Jane” have been used sardonically, one might think Zerna Sharp’s schoolbook characters were already public domain. But you can’t copyright names, expressive style, or stock situations, and fair use allows limited copying for purposes like criticism, analysis, and parody.

In 9 days, the 1930 Elson Basic Readers introducing Dick and Jane join the public domain, and their stories’ full texts and original artwork can be reused without limit. #PublicDomainDayCountdown

Four writers get credit on the 1930 copyright registration for “Body and Soul”: composer Johnny Green, and lyricists Robert Sour, Edward Heyman, and Frank Eyton. But many more artists shaped the perennial jazz standard we know today. Among its more than 1700 recorded interpretations and variations are covers by Coleman Hawkins, Frank Sinatra, Billie Holiday, Ella Fitzgerald, Tony Bennett and Amy Winehouse. The song surrenders itself to the public domain in 10 days. #PublicDomainDayCountdown