What’s Next for Curation

Why do you conflate downvotes and question closure? These are two different mechanisms with totally different purpose!

Downvotes are a way of ranking content. It is a signal to other viewers whether the question is worth opening and reading.

Question closure is a way of telling the asker that their question cannot or should not be answered on this site. They need to make a significant edit before they can receive good answers.

Last week in the beta announcement, we mentioned that we had a desire to move away from question closure as a primary curation tool.

Yes, you did. You subsequently ignored all feedback, then posted this.

It's clear SE does not care about the community. Curation is seen as a hostile hindrance, to be shoved away into a corner. It's clear SE has made up its mind. It's clear this is only being posted so SE can tell their shareholders that they "communicated" with the community, while trying to keep this sinking ship afloat.

A plan so completely out of touch with how the website actually functions, I wonder if anyone involved has ever asked a question on SO, or touched the review queues.

Needs more focus / low quality

So these closure reasons no longer block answers? Sure, let's open the floodgates. Who cares what the quality of the question is? At least they've got a chance to receive an answer, right? The value of SO isn't in getting a quick answer at all costs. It's repository of knowledge SE seems to be determined to turn into a heap of trash users will have to wade through.

Structured downvote reasons

SO has always been strictly against attaching reasons to downvotes.
How does SE expect this to function? People don't downvote nearly as much as they should already, so don't tell me it's going to be some kind of mandatory option you need to pick if you want to apply a downvote. That's horrendous UX and will kill any incentive to vote.

If that is how it works, do the same for upvotes. If downvotes require a reason, upvotes should be treated the exact same way. Don't just hobble downvotes because they're "unwelcoming".
Low quality content, that is unwelcoming.

Different viewing modes.

What possible incentive would users have to continue curating?

Opinion based

Again, SE seems just to want to open the floodgates. More content does not equal value. It's trash to be sifted through.

Duplicates "We would also offer a curation flow"

Sure. Burden burnt-out curators with more ineffective work. Automatically importing a dupe target's answers is incredibly unreliable. More often than not, dupe targets' answers aren't exact answers to the duplicate question, and require the context of the dupe target's question. Duplicate targets often ask and answer the same concept, but can not be taken out of context.

The problem with duplicate closure is that users just want an exact, tailored answer. They're lazy. They don't want to learn. They want to be served.

Off-topic

SE is way too large to allow any form of automatic migration / linking to other SE sites. These SE sites all require specific domain knowledge, and more often than not, these migrations already do not stick, becasue the question itself is just bad.

Next steps We are early in the development life cycle

Well that's just grand.

SE made a plan to completely upend their website. They pretend to take feedback, but it's painstakingly obvious SE isn't here for feedback.

This is just an announcement that it's going to happen, regardless of what the community thinks.

Frankly, I can't wrap my head around how completely out of touch this whole process is.

• A superfluous redesign that blindly applies the latest design fad.
• Curation being upended in a desperate attempt to hold on to low-value content.
• Meta announcements under the guise of "seeking feedback".

I never imagined I could be this disappointed in this website.

All of your "Curator Experience" bullet points read like a task list to me. If you're setting the expectation that a curator works this way, the curator should also set the expectation of their hourly rate.

Seriously, you expect the average volunteer curator to do

• explain downvotes on their own (something curators have already universally rejected doing)
• hide questions which haven't got any feedback/input (which already happens if a question is downvoted enough)
• take on the burden of classifying questions on their own (and there's already more than enough flak over re-tagging a question, so we're just gonna keep doing that thing that already annoys certain people for whatever reason)
• add context to a duplicate link (which...already exists plenty with the current flow)
• allow an OP to insist that this time, their NullReferenceException is different (and only once in about every 3,000 times have I seen or reopened one of those questions with a gold badge that I've closed because it actually was, so...this flow does exist)
• send questions to other parts of the network when the vast majority have explicitly rejected migrations from Stack Overflow because it's not suitable on their site either

...all for free??

I honestly don't know what to make of this. It's like, avoiding the last decade-plus of Meta because, instead of taking what was discussed there as community perspective, useful context for improvement and the convention on which curators move with, you're saying it's a new dawn and we've gotta do something radical to shake things up, so all the things that made the curators you have now need to be thrown out. (That'll probably scare off the few remaining curators, y'know.)

This sentiment also feels divorced from reality, again:

We want to take each close reason and move it towards a curation workflow that fits in more reasonably with Stack Overflow today. We want to differentiate “this content doesn’t belong on Stack Overflow” from “this isn’t a question I am interested in,” which is different from “this is high-quality content that is worth promoting”. We want tooling that lets those signals exist independently, and we want curator input on what that actually looks like in practice.

If you can find one example of a question which was closed because the curator wasn't interested in the question, then just reopen the dang question and discipline that curator, because that has never, ever, ever been a valid reason for a question to be closed.

Questions can be downvoted simply because it's hard to understand what's being asked, what the problem actually is, or because the question doesn't make contextual sense and wouldn't bring future value, which - I must stress again - is not the same as "I don't like this question".

It's tough to deliver a solution that works if you're not understanding the problem space or domain you're meant to be solving. That's probably why much of the posting here seems to be geared to helping all y'all get back on track, from our perspective, on what we're expecting this to be.

But maybe it's the other way around. Maybe we're the ones in the wrong here. Despite what we've been posting about here on Meta, despite what we've pleaded and recommended, and how we wanted to work together, maybe you have a clear and decisive picture of the domain you need to move toward, and maybe we're the ones who are obstinate.

Either way, I don't think the curators voicing their thoughts here and the company's vision are aligned. No matter how we got here, I think that's what's going on right now.

Feels like this stalemate's been going on for as long as I can remember, and nothing's going to improve if no one does anything to break the stalemate.

This is a giant pile of weasel words, and sounds exactly like yet another company bait-and-switch. Let's keep in mind a few facts:

1. Most closed questions are not going unanswered because we've closed them. It's the other way around, we've closed them because they are unanswerable. While it's true that some opinion-based questions are answerable, those are not the majority of closed questions.

2. Your opinion-based questions experiment has eliminated all closure reasons, and many of those questions are unanswerable. That helps nobody, including the asker, and just pollutes the site. You're going to claim that it's just an experiment, that it's not the company's vision of the final design, that you are looking for input, etc., but from past experiments and the carefully tuned noncommittal language here, I don't believe you. Your opinion-based questions experiment could have simply eliminated the 'opinion-based' and 'asking for off-site resources' closure reasons and kept the rest of the closure reasons and process. In other words, if you really meant to keep curation around, then that's what your experiment would have done. It would have been much easier to implement.

This creates a much stronger feedback signal. Rather than wondering if your duplicate suggestion was helpful, you receive a clear data point. If an author does not accept the proposed answer, it signals to curators that the question may have a unique nuance that requests a new distinct answer.

The thing is, we don't care about that. When we close a question as a duplicate, it's because we are 100% sure that the answers in the linked question are the best solution.

There have been situations when users applied too broad criteria for determining if it's a duplicate or not, or used a too generic duplicate target. But that's not the fault of the system.

What happens if the question asker disagrees that their question is a duplicate? Should they vote to reopen? How is that different from what we have now?

this isn’t a question I am interested in

That is not a close reason and never has been.

We want tooling that lets those signals exist independently, and we want curator input on what that actually looks like in practice.

There is no need for an “I’m not interested” signal, just don’t look at the question. We already have voting and closure, and you know very well how both work in practice.

We want to decouple quality signaling from the asker’s ability to receive an answer

If no one can understand the question enough to give an answer, it is definitionally low-quality. It’s utterly useless. Additionally, downvotes and close votes are different things!

We want to prevent the “hard lock” failure mode where questions are closed against a “duplicate” that may not actually solve the asker’s specific problem.

The entire point of duplicate closure is that the linked question solves the asker’s problem. If it doesn’t, then whoever closed it was just wrong and it should be reopened. No system changes are necessary.

Hoid, I have to assume you know these things already, and someone told you to put that in. It would be very helpful though if you and other CMs could relay this and other basic information about how curation works to whoever is making these decisions.

Our long-term vision is smarter backend routing. A system that identifies where a question would be best suited, whether that be adding certain tags or considering other Stack Exchange sites. In the near term, we are consolidating off-topic flags into the existing flag workflow. This reduces the number of clicks and menus you have to navigate and removing the friction and redundancy currently found in the off-topic sub-menus.

This sounds like a good idea on paper, but have you done any research into it? There's a reason why moderators decline most requests for migration.

The network doesn't have a dedicated site for every question. Some questions simply cannot be answered by the community. For example, many cloud questions (Azure, AWS, etc.) have been closed as off-topic, not because they belong on a different site in the network, but because they can only be answered by the cloud provider and that information is fully in their control. Answers provided by community members could be unreliable and cause the information seeker to suffer financial repercussions.

This reason has also been used for posts that cannot be categorized as anything else. Things that aren't questions or requests for knowledge. Posts that do not belong on any Q&A site. Even the most intelligent system could not route it to a place where it belongs. Sure, they could be dealt by asking the mods to delete it straight away, but it's not something that a mod needs to deal with when we have appropriate close reasons.

We are looking to adapt the successful feedback mechanisms from the Staging Ground to work at the scale of the main site.

First of all, there is no "scale of the main site" anymore. The number of asked questions is extremely low and there are enough curators who more than capable of handling them all.

Now, to the main point.

You already have Staging Ground. But you never fully made it work, regardless of the feedback you received for it.

Main problem with SG is that not enough experts in the subject are able to see any particular question and give feedback on it.

First you made SG optional, so not all users see those questions at all. Considering that without SG those questions would end up being directly asked on the main site, there is absolutely no reason that SG is optional.

Next, SG question which has been reviewed and not published will vanish from the timeline. this makes those questions even less visible and less likely to be published if the question is sufficiently edited by the author.

For some questions early feedback can be given by many curators, but when question gets edited, only those who did initial review will be notified. But if the reviewer is not SME, it may happen that they are not in a good position to review the question again. Publishing question prematurely can have a negative impact on its reception. On the other hand, not publishing it means it will be stuck in SG without an answer.

If you would show those questions in the timeline, just like any other question, regardless of its review status, then actual experts would be able to see them, give feedback or publish and answer them.

All this would be extremely easy to implement.

Also you could give all users a choice to publish questions through SG, so those that really want feedback could get it.

Instead of just putting a post notice on a duplicate question, we would also offer a curation flow where one or more of the canonical duplicate’s answers would appear directly on the duplicate question, with added context if needed.

This is a good change and I believe it's something we have been asking for. A way to point to a specific answer with some extra context to help see the OP how it applies to their specific problem. You don't need to overcomplicate this. The provided context could also be visible just to the OP, the people who voted to close and moderators.

Or you know... that extra context could also be provided in the comments after voting to close.

I'm going to go ahead and answer each of the sections in your heading individually, just to get my thoughts written down:

Needs more focus / low quality

A lack of focus has nothing to do with quality; a question can be very well asked but lack focus. You have confused 2 very different issues here

Goal: We want to decouple quality signaling from the asker’s ability to receive an answer. We want to create more space to ask questions that may not be perfectly formed on their first attempt.

Quality "signaling_[sic]" (votes?) has only affected a users ability to get answers by the fact that a low quality questions may not be shown to as many people. Being Low Quality, by itself, does not stop something getting answers, however, it might mean people aren't interested in answering.

Idea: We are thinking about structured downvote reasons. Preset categories that provide context beyond a negative score. We are looking to adapt the successful feedback mechanisms from the Staging Ground to work at the scale of the main site.

Downvotes, with reasons, has been a long discussed topic. I don't think we need to cover that here for the nth time. You do, however, need to ensure protections of the voters; responses to providing my reasoning for why a post might be getting downvotes (even when I haven't voted, or might have even upvoted), are some of the most toxic experiences a user can have on Stack Overflow.

Curator Experience: To prevent “review fatigue” we want to think about different viewing modes. Curators could toggle between seeing questions that have not received any feedback or hiding questions that have already been addressed by other curators. This would lead to the question eventually being deleted if no efforts were ever made to improve it by the author.

Users are already limited on how many reviews they can do in each queue, and they can easily bail out if they want to. I don't see "review fatigue" as an issue.

Opinion-based

Goal: We recognize that well-formed, opinion-based questions have value, but we also know that not everyone cares for them. Our goal is to make this content opt-in/out based on individual preference.

Opinion based contact can have value, I don't disagree, however, the implementations Stack has done so far have been bad. If your goal was to make these experiences better, why not actually improve those features, rather than break the entire site?

Idea: There could be a classification layer using categories using question types like “Tooling Recommendation,” “how do I,” or “implementation guidance." These would allow users to ignore, follow, etc., just like other tags, and only see content that they are interested in.

We have this, and it's broken... Posts can't be migrated, and users are tricked into selecting the wrong category. For the 100th time, where is the "How to?" category?! Fix the solution(s) you've tried.

Curator Experience: Rather than voting to close, curators could instead have the ability to reclassify the posts. This could mimic the workflow of changing tags on a post that we already have.

Why not give curators, and moderators, the power to curate/moderate the solution you have? We've been asking for this and you CONTINUE to ignore us; this isn't your idea, it's the community's.

Duplicates

Goal: We want to prevent the “hard lock” failure mode where questions are closed against a “duplicate” that may not actually solve the asker’s specific problem.

This already exists; we have reopen votes. That users don't edit their question, after it is closed, to explain why a question doesn't answer theirs, is the author's fault, not the community's. Those that do, often get met with success; either with an explanation of why in the comments or by reopen votes.

Idea: Instead of just putting a post notice on a duplicate question, we would also offer a curation flow where one or more of the canonical duplicate’s answers would appear directly on the duplicate question, with added context if needed. Then the asker would still have the opportunity to accept the proposed duplicate answer or leave context explaining why it doesn’t answer their question. We would also want to think of a feedback loop for the question author to accept that proposed duplicate answer before the question is locked to new answers.

This is what we expect to happen already, you know that, right?

Curator Experience: This creates a much stronger feedback signal. Rather than wondering if your duplicate suggestion was helpful, you receive a clear data point. If an author does not accept the proposed answer, it signals to curators that the question may have a unique nuance that requests a new distinct answer.

Again, this is what is meant to happen; that authors don't do this isn't the community's fault. Educate your new users; we have also been asking for better onboarding for years.

Off-Topic

Goal: Maintain Stack Overflow’s focus on programming without making “Off-Topic” feel like a shut door for users with legitimate technical questions that may belong elsewhere.

We have a tool for technical questions that belong elsewhere; it's called migration. We can't, however, migrate older posts; maybe you should allow us to do so. Off-topic questions are normally more that they aren't suitable on any site. Also, don't forget, users have a small selection of sites that things can be migrated to; give them more options if you want those questions to go elsewhere.

Idea: Our long-term vision is smarter backend routing. A system that identifies where a question would be best suited, whether that be adding certain tags or considering other Stack Exchange sites. In the near term, we are consolidating off-topic flags into the existing flag workflow. This reduces the number of clicks and menus you have to navigate and removes the friction and redundancy currently found in the off-topic sub-menus.

Making people aware that other sites exist, before they ask, isn't a bad thing, but tags aren't always a good indicator. They are often misused, and there's plenty of cross over. docker is an absolute mine-field and I have no idea what is and isn't on-topic for it.

Curator Experience: Curators would be able to monitor this system and implement mechanisms to better fine-tune the decision-making process. When necessary, they would also be able to intervene in the event of something being routed somewhere incorrectly.

There's not a lot here to really go on, so I can't really comment.

The short here is that several places here you seem to mistake downvote reasons for closure reasons; these are not the same. Often questions that are close further are downvoted, but that doesn't necessarily mean they should be (a high quality question that needs migration shouldn't be downvoted). Likewise a downvote doesn't mean closure; a common reason for closure is due to a lack of research.

Unclear

You also completely miss the unclear closure reason; the question is on-topic, however, what they actually want to ask is completely unapparent. These are problematic, and often result in bad answers if they are answered; the answers guess at what the requirements are, often can completely miss the mark and when (if) the question is clarified it can frequently invalidate the answer(s). This is why closure for these questions is important, because it stops those answers.

Off-topic:

Our long-term vision is smarter backend routing. A system that identifies where a question would be best suited, whether that be adding certain tags or considering other Stack Exchange sites. In the near term, we are consolidating off-topic flags into the existing flag workflow. This reduces the number of clicks and menus you have to navigate and removes the friction and redundancy currently found in the off-topic sub-menus.

Two thoughts:^{(for the record most, of the upvotes so far were before I added the second)}

1. Please don't turn "Off topic" into a single opaque category, in case that's what you mean. I think the close reasons help guide curators.

   As an aside, I would say in fact the current (non-beta) UI fails to display those reasons enough (by not showing them in the Reopen queue).

2. I absolutely agree with routing people to other sites. Thank you for thinking about it! I really want to see this implemented, on both Stack Overflow and other sites, ideally as something that happens before the question is posted. (Migrations are difficult, and you can point people directly to the site-specific asker advice.) I think it's a significant lack, and that its lack is driving people away from the sites and making off-topic curation seem like more of a problem than it needs to be.

   I think a rough version of this could be more easily and quickly implemented than reworking curation, because at its core it only needs to be helpful links shown before the "Post Question" button, similar to the current "maybe this is a duplicate of" section. It could be incrementally improved over time, to be smarter, with machine learning. I think it's a better near-term goal than the changes you're prioritizing.

   And please don't forget that this also addresses the opinion-based issue. In the past, we've split off sites to handle different styles of subjective and objective questions and answers. Route people to them! Use what we already have!

   Once we're better directing new users to the resources and communities we already have, I'd rather see you, the company, engage with the communities about tweaks to site scopes. Scopes can change; they have in the past. It feels almost like changing the interface is a way to force a scope change that you expect we'd argue too much about. But look around, a lot of arguing happened anyway. If scope changes are what's needed, propose them as scope changes instead of as side effects of your UI development. Write up the arguments in favor, the stories of how you'd like to see the expanded-scope questions handled. Be clear if there are non-negotiable details, but don't make the system more opaque, or toss it out.

We want to take each close reason and move it towards a curation workflow that fits in more reasonably with Stack Overflow today. We want to differentiate “this content doesn’t belong on Stack Overflow” from “this isn’t a question I am interested in,” which is different from “this is high-quality content that is worth promoting”.

This is so far removed from how closure works that I cannot meaningfully address it. Which close reason, exactly, do y'all think means "this isn't a question I am interested in"?

Goal: We want to decouple quality signaling from the asker’s ability to receive an answer. We want to create more space to ask questions that may not be perfectly formed on their first attempt.

You already have a product for this: https://stackoverflow.com/staging-ground/ If you want fewer questions to be closed, put more questions through that process.

Goal: We recognize that well-formed, opinion-based questions have value, but we also know that not everyone cares for them. Our goal is to make this content opt-in/out based on individual preference.

You already have a product for this: https://softwareengineering.stackexchange.com/ That's literally why this site was created; for programming questions that are just a bit too subjective or open-ended for Stack Overflow.

Goal: We want to prevent the “hard lock” failure mode where questions are closed against a “duplicate” that may not actually solve the asker’s specific problem.

You already have a solution for this: question askers can edit and vote to reopen their own questions. Editing it gives them an option to automatically put the question in the reopen vote queue.

Idea: Instead of just putting a post notice on a duplicate question, we would also offer a curation flow where one or more of the canonical duplicate’s answers would appear directly on the duplicate question, with added context if needed. Then the asker would still have the opportunity to accept the proposed duplicate answer or leave context explaining why it doesn’t answer their question. We would also want to think of a feedback loop for the question author to accept that proposed duplicate answer before the question is locked to new answers.

I'm not opposed to this, but 1) how will voting on the copy work? Will it be clear that it's a copy; will you be able to vote on the same answer multiple times; will rep be doubled? and 2) please note that you are using terms incorrectly here. Duplicate questions are closed, not locked. Please do not use the term locked to refer to anything other than an actual lock that prevents interactions by non-mods or non-staff.

Goal: Maintain Stack Overflow’s focus on programming without making “Off-Topic” feel like a shut door for users with legitimate technical questions that may belong elsewhere.

Cool, for the hundredth time, the community has given you a hand-crafted solution for this, myself included: Auto-direct users during the ask wizard step to the appropriate site based on machine learning/generative AI scanning. Y'all love genAI tech now, right? Here is one actually good use for it; making askers (and the company, apparently) aware that Stack Exchange already has lots of other sites besides Stack Overflow.

We want to differentiate “this content doesn’t belong on Stack Overflow” from “this isn’t a question I am interested in,” which is different from “this is high-quality content that is worth promoting”.

May I propose the following differentiation:

• "Content that does not belong on Stack Overflow": close, and delete if it's unsalvageable.
• "This isn't a question I am interested in": click away, add any uninteresting tags to your ignore list if they aren't there already, possibly edit the question to add those tags if they aren't already present.
• "This is high-quality content that is worth promoting": upvote, copy a short-link to share it off-site, possibly use a bounty to reward an especially good answer.

Estimated cost of implementing these features is zero, since they are the status quo.

If it is your allegation that the community is wrongly using the close/delete features on content simply because we aren't personally interested in it, then that allegation should be substantiated. The solution would be better education for appropriate use of these existing features, or perhaps revocation of privileges (e.g. temporary block to the site) for persistent abuse.

Goal: We want to decouple quality signaling from the asker’s ability to receive an answer

This is a bad goal which will harm the quality of content on the site. If a question is too unclear, vague or open-ended in its current state by our standards, then there is a likelihood that any answers written will not address what the asker actually wants, and that subsequent edits to the question to make it clearer and more focused will obviate the already-written answers. That is why we prevent answers being written on questions while they are in such a state.

Goal: We recognize that well-formed, opinion-based questions have value, but we also know that not everyone cares for them. Our goal is to make this content opt-in/out based on individual preference.

This is not as bad as a goal, but it's still bad and will harm the quality of content on the site. In practice, it is much harder to set clear standards for which opinion-based questions are acceptable and which aren't, and this will lead to more low quality questions being written and not removed.

Moreover, the userbase who currently curate questions and answers are probably not much interested in curating opinion-based questions if there are standards which allow them, so having an opt-in/out section of the site where most curators don't go is a recipe for slop and spam.

Goal: We want to prevent the “hard lock” failure mode where questions are closed against a “duplicate” that may not actually solve the asker’s specific problem.

This is a bad goal because it is based on a false premise. Duplicate closure is not a "hard lock" any more than any other closure reason. If a question is not really a duplicate, this can be resolved in comments or by editing the question, and then the question can be re-opened.

If you think it should be easier to get such questions reopened, that's fine, we can talk about that. Allowing duplicate questions to stay open for answers is not a suitable alternative, and will only result in the 99% of genuine duplicates receiving duplicative answers from rep-chasers.

Goal: Maintain Stack Overflow’s focus on programming without making “Off-Topic” feel like a shut door for users with legitimate technical questions that may belong elsewhere.

I must say I don't really understand what this goal is. If it's really just about making people feel better when their questions are closed as off-topic, I don't see how expanding the set of allowed topics (as your idea suggests) would do anything towards this goal. There will still be topics that aren't allowed, and questions on those topics will still be closed, and the people who wrote those questions will still feel bad about it.

If I understand the goal correctly, it seems the best thing to do is to soften the wording shown to users when their question is closed due to being off-topic, and explain to those users what their alternative options are.

Needs more focus / low quality

• "Quality" is "the asker's ability to receive an answer". We close questions that need focus, lack details, or lack MREs because we don't have enough information to answer them. When that happens, it is important that the question remains closed so that bad answers that lack context do not flood the question page.
• "Structured downvote reasons" already exist. They're called close reasons. The only reason to switch from close reasons to "downvote reasons" would be to make downvotes less unfriendly, which I don't think will work. The perception of unfriendliness comes from the big negative number; adding text to that probably won't help. (Which we know because "on hold" didn't really work.)

Opinion-based

• This is potentially fine in theory.
• In practice, it will be bad for all the reasons that the experiment is bad. I am not going to rehash them here; there are way too many and (given company track records) I have insufficient faith that they would be addressed.

Duplicate

• Duplicate questions still need to be closed. Otherwise, people will only notice the one transcluded answer from the canonical and inadvertently post duplicates of the canonical answers that weren't transcluded. If that happens, duplicates will stop serving their purpose: reducing duplication by pooling answers in one place.
• Incorrect duplicates are not a "hard lock". There is already a recommended pathway for resolving them: editing the question and voting to reopen.
• "[A]n author does not accept the proposed answer" is a very bad feedback signal. There are too many confounding variables: the author could've lost interest in the problem, solved it independently, lost access to their account, etc.

Off-topic

• There are a lot of off-topic close reasons. How exactly do you plan on expanding them out? It seems like any time gained in clicks will be lost in scrolling.
• Migrations by automated systems or non-moderator users seem like a really bad idea. We try to avoid migrations in general since experienced Stack Overflow users often do not understand the asking guidelines on other sites. In other words, I wouldn't trust the "smarter backend routing" to not. Migrate. Crap.

I'm going to start out by saying that what's below is just some stuff I feel like needs to be said and not all of it is directed at SO. I've probably said some of it before but I see these little nuggets all over and I'd like to push back some because I think they're getting in the way of any viable discussion between the company and the community.

The Question

While closure was an effective tool to manage higher volume in the past, we believe it is now a significant cause of friction to authors and curators.

The data doesn't support that closure/review has ever been an effective tool on high-volume sites and, I'd argue, is actually better-suited to medium sized sites. The close review queue is (was, anyway) tuned to have an upper number of questions in the queue and (if I'm remembering correctly) actually has a variable in the "aging" formula to kick questions out of review sooner if the queue gets too big. Five digit numbers are demotivating for reviewers, y'know?

Historically there have been significantly more posts nominated for closure but aging out of review than getting closed. This led to the number of votes needed to close being reduced from 5 to 3 back in 2019, as only 36% of questions that received a first vote to close would complete the process when 5 votes to close were required. Even after the change, nearly half of questions nominated for closure aged out of review.

Unfortunately, for small sites, the review queue indicators don't always light up to draw attention to close reviews, leading to those sites also struggling with closure. Review works best on sites with enough questions getting reviewed for indicators to work but not so many that the community of reviewers can't handle them before they age out.

As such, I'd say SO is probably nearly at the sweet spot for review to actually work again, ironically. Though that assumes there are reviewers to handle those questions.

We want to differentiate “this content doesn’t belong on Stack Overflow” from “this isn’t a question I am interested in,” which is different from “this is high-quality content that is worth promoting”. We want tooling that lets those signals exist independently, and we want curator input on what that actually looks like in practice.

I don't actually have a big problem with this, though you're outlining three axes with statements that relate to one end of each axis instead of looking at the entire axes as a whole.

More accurately, the three axes are:

1. Scope - ranges from out of scope to in scope. This can be indicated through comments, votes and - in the case of out of scope content - closure/deletion.
2. Interest - ranges from disinterest to interest. Interest can be indicated through "saves" and following, though disinterest can only be indicated on a tag-wide basis, through ignoring tags. I'd say it's not uncommon for users to indicate strong disinterest with downvotes.
3. Quality - ranges from low quality to high quality. The primary indicators here are comments and votes, though flags/closure/deletion are options in some cases.

Scope is really broad and often subjective. Why doesn't the content belong? Is it about edible cookies? Is it a post type that isn't in-scope here? Is it in a non-English language? Do community members even agree about whether the question belongs here? This is generally determined by the community and the goal should be to be as inclusive as is reasonable. That said, the community is empowered to determine questions - or often question types - that don't perform well in the Q&A format.

Interest is... interesting because the platform doesn't generally put much value in whether or not someone finds a specific question interesting. The (unstated) expectation is that votes and closure shouldn't be used to indicate that a question is or isn't "interesting" - many basic questions may be "uninteresting" or even "useless" to an expert but that doesn't mean they should be downvoted - but I'd say they are used for this in many cases, couched as "too basic" or "doesn't show research effort". That said, I'd argue that in some cases, duplicate closure is overused for "basic" or "uninteresting" questions far more often than is warranted.

Quality is where votes should come in but this is also subjective, particularly without adequate guidance for when to vote. For example, the hover text for voting heavily relies on "useful" without qualifying who should find it useful, which leads to people calling things they personally find uninteresting "not useful" and then downvoting it. The other element of voting on questions is "shows research effort", which is quite subjective and also used problematically at times. I'm not quite sure how we ended up with these two metrics being how we define "quality" but, here we are.

Here's the thing. You are welcome to propose as many different methods for giving feedback as you like but you can't actually force people to use them "correctly". People will use whatever tools they have at hand to make content they don't like go away, even if that's not the purpose of the tool. Another answer mentions the fact that sufficiently downvoted posts are removed from the front page - has that feature incentivised people to downvote more to hide content they don't want? Probably.

Please - try to make this better, more clear, more "useful" but don't expect people to follow those intentions, particularly if you continue to do nothing to reduce the volume of content that needs curation.

Consider the following:

A user has an excellent answer explaining concept A. It's highly regarded and has been around for several years. Over the years, they see the same question asked over and over by different users who didn't find the excellent answer that already exists. They dutifully duplicate close and leave comments clarifying how their existing answer solves the problem. Over time, they get frustrated that users frequently fail to find their answer, which they've kept updated as the technology has changed.

With the question vote hover text in mind, they start downvoting these questions in addition to duplicate voting because if the asker had put in any "research effort", they would have found the existing answer and never asked the question in the first place and it's not "useful" since there's a duplicate question. Even well-asked, fully-explained questions get the downvote simply because they failed to find the existing answer. The user's once-helpful comments take on a tinge of frustration, caused by years of pointing people to a resource that already exists. Maybe the comments even stop entirely, with the questions just being duplicate closed and downvoted.

The thing is, a well-asked duplicate should be closed and it should get a helpful comment but it shouldn't get a downvote for not showing "research effort" or not being "useful"... but what do you think led to those votes? Was it the asker? Or was it the platform failing to prevent the question being asked - not just once - but repeatedly over many years?

Now, you're in a position of having these curators who are already misusing votes because the system is overburdening them and you're discussing changing the curation system that is all they have - but you literally frame it as

"we plan to retire certain curation workflows, such as close votes and most review queues, in the new design."

... and then, when the community goes on high alert you come here and, instead of asking how the curation system could better serve the community, you put out this "thought exercise" that - as many have pointed out - seems to indicate an incomplete comprehension of how curation on the platform functions.

We are early in the development life cycle with these changes, and we're actively looking for experienced community members to pressure-test these ideas in a workshop type of way with the product manager and designer that has been working on close reasons.

You're going at it backwards. Stop asking the community to review your solutions and give us a chance to actually get your PM/designer up to speed without having to tear down a preconceived idea of how to "solve" this issue. This is a huge ecosystem and solving curation issues without solving (or at least fully understanding) the content creation issues that necessitate the curation tooling in the first place will lead to failure and frustration.

Remember, closure isn't the end of things, either - it leads to deletion. Without question closure, there will be a ton more content on the site than there currently is. If you look at closed questions over a year old with no answers (non-duplicate), there are only 9.6k, most of which have a score of 2 or higher. Compare that to these 1.6 million zero score open questions with no answers over a year old that have been saved from deletion by simply having two or more comments - that's over half of all unanswered open questions! The platform would be full of significantly more cruft without closure or a path to deletion for content that isn't engaging to people.

Comments/Answers

Votes vs closure

Why do you conflate downvotes and question closure? These are two different mechanisms with totally different purpose! ~ Dharman

It's completely reasonable for the company to consider these two together because the user experience for the asker whose question is closed almost always includes downvotes. Heck, Anerdw even draws the equivalence:

"Structured downvote reasons" already exist. They're called close reasons.

There are 277 non-deleted questions that were asked in the last 10 days that have been closed for non-duplicate reasons. Of them, 205 have a negative score, 161 have a score of -2 or less, and 31 have a score of -5 or less. Of the 51 duplicate closures of questions asked in that time, only 16 have a negative score but all 16 have a score of -2 or less.

These two mechanisms have significantly intertwined usages since being close-worthy generally leads to downvotes and - even if closure can be overcome - the downvotes often lead to it feeling like it's not worth trying to fix the problems. Remember, while they may be signals to two different people, one person gets the full impact of both signals.

This is one of the reasons I've batted around the idea of whether downvotes on questions make sense at all - beyond their functional purpose of leading to post deletion. It's not an easy problem to solve but let's not pretend the two are unrelated.

Duplicates

When we close a question as a duplicate, it's because we are 100% sure that the answers in the linked question are the best solution. ~ Dharman

You don't even agree with this statement since your next sentence acknowledges that duplicates have been used too broadly in the past and others on this page have pointed out that duplicate closures can be wrong or point at overly-broad canonicals.

My biggest problem is probably with omnibus canonicals being used as a duplicate closure. Which of the 50+ answers will be useful to the asker? Point it out to the asker. If the question is too unclear to point it out, vote to close as unclear. It's ing stupid to tell someone with an unclear question that "Your answer is in here. Somewhere. Go dig." ~ user4581301

This is how I've felt for a long time. Canonicals can be extremely helpful but using them as a duplicate target is not a good UX for anyone. Tim Post once talked about an "answer as duplicate" option where someone could cite excerpts of an existing answer on another question while closing to ensure the germane parts were visible. This sounds similar to what's being mentioned as a solution in the question.

While user4581301 and I share similar feelings, others say things like:

The problem with duplicate closure is that users just want an exact, tailored answer. They're lazy. They don't want to learn. They want to be served. ~ Cerbrus

This frustration is understandable but it's very uncharitable to assume that these askers are "lazy", particularly after having just stated:

Duplicate targets often ask and answer the same concept, but can not be taken out of context.

Not everyone on SO is enough of an expert to take a concept and apply it to their specific situation - an "answer" that they can't use, isn't an answer. If we find that existing answers aren't helpful to users, we need to investigate whether the answers adequately generalize the solution to help users who have their question closed as a duplicate recognize it as such.

We need to stop dupe closing and pointing to questions with 50+ answers and then calling askers "lazy" for not being able to sift through thousands of characters to find the bit that actually solves their problem.

The entire point of duplicate closure is that the linked question solves the asker’s problem. If it doesn’t, then whoever closed it was just wrong and it should be reopened. No system changes are necessary. ~ Starship

It is technically true that any closed question can be reopened but it almost never actually happens. I've personally had a question closed as a duplicate here on SO and I'm pretty sure the only reason it was reopened was because I went to the Blue Room and whinged at the SO mods about my poor question being closed. Most users don't have that access.

If you have 10k, go to the Question Closed Stats and just look at the numbers:

I also, at least in my experience, find that duplicates are one of the easiest to get reopened. [...] they can ping the closer in the comments and they can reopen it if they evidenced appropriately. ~ Thom A

Closed questions almost never get reopened and, over the last 90 days anyway, duplicates get reopened far less often than other closure reasons despite being the second most common reason for closure. Considering it's the only closure type that can be unilaterally undone by a gold tag badge user, the percentage being so low seems to indicate users find little success in getting duplicate closed questions reopened.

Remember, askers - who are often newer users and low-rep - can't see who closed their posts without looking at the timeline and they don't have a way to ping the close voter unless the post was unilaterally closed... but how would they even know that? I forgot that you can ping unilateral close voters and I wrote this five years ago, so I clearly knew it in the past.

We need to stop insisting that the status quo is sufficient and transparent.

Uninteresting content

"this isn’t a question I am interested in" - Closure was never about not being interested in some particular questions. Dalija Prasnikar

"this isn’t a question I am interested in" If a user is not interested in a question they won't open it, or simply close the page after reading it. They may also downvote it after reading, but they definitely should not vote to close. Using that as a reason to close a question is an abuse of their close voting privileges and could result in a warning from the mods. Dharman

"This isn't a question I am interested in": click away, add any uninteresting tags to your ignore list if they aren't there already, possibly edit the question to add those tags if they aren't already present. kaya3

However much we might say people shouldn't downvote or vote to close questions they don't find interesting, as a community, let's not claim it doesn't happen or that people aren't using "lacks research effort" and "not useful" as code for "uninteresting" in the same way some use duplicate closure too broadly to point at omnibus answers that would require hours to digest.

I'm not saying that everyone does this but that doesn't mean it doesn't happen. I can't possibly understand how else you explain the misuse of duplicates that has persisted here and the attitudes of some when it's suggested that a little more effort be put in to ensure askers can actually find the relevant content in the duplicate.

The reality is, when you've been answering questions for years, most questions won't be novel and you're going to be as exhausted as the user described in the "consider" section above. Despite that, we need to remember that it's the platform at fault when users fail to find answers that already exist or ask questions that need improvement or otherwise qualify for closure. We need to be more willing to throw the blame at the platform and give the askers grace and remember that our expertise includes knowing how the platform works, which most people using the site don't have.

The company needs to improve the platform so that all users have a better experience but they seem unable to see themselves as the source of the problems and seem to require community members to pick up more of the effort as they revamp the system without actually making - or even understanding - the curation process.

Stop asserting that the platform works fine as it is - it doesn't! Question whether the reality aligns with the intention rather than assuming it does or being blind when it does not. We don't have to roll over and accept the solutions the company is providing but when we can't recognize the flaws in the system and accept that things need to change, the company has no reason to try and partner with us to solve the problems.

Idea: There could be a classification layer using categories using question types like “Tooling Recommendation,” “how do I,” “implementation guidance. These would allow users to ignore, follow, etc., just like other tags, and only see content that they are interested in.

When are we going to get to hash out what these classifications should be and what should happen when they are chosen? Out of the gate we were given classification options that shoved perfectly valid every day Q&A questions into a discussion format that is ill-suited to answering them. Despite this being a known issue, it's still the case today, months later.

While closure was an effective tool to manage higher volume in the past

That's besides, well, reality. We never closed question because we've had enough to answer elsewhere. We close questions that simply doesn't target the goal of this site.

We want to take each close reason and move it towards a curation workflow that fits in more reasonably with Stack Overflow today.
...
We want tooling that lets those signals exist independently, and we want curator input on what that actually looks like in practice.

I do agree that the tools provided are sometimes too simple and do not balance the interests of asker and answerer enough.

We are thinking about structured downvote reasons.

If that means a bar of buttons below the question, where answeres can simply click downvotes FOR a reason why this question is bad, I would welcome it. It would be nicer and more effective than downvoting that on the first level is just a vote AGAINST the question (in its current state).

A preset of pre-formulated ones and the ability to add a custom button would be great. Sometimes questions don't receive answers because we look for an interesting title click on the question and find a huge code-block or a huge text you have to work yourself through, even though the question implies a reasonable scope and you get uninterested in the question.

I remember my first question and today I know how silly it must have looked to an expert. However it took awhile before I received an answer, watching the amount of views grow, maintaining a positive score and no other feedback at all. Today I would hesitate to address that question myself, not because I can't answer, but it requires a lot of context for a good answer. I would guess old-timers here would have even closed the question with the asker lacks of a fundamental understanding of the problem. Maybe some would say it lacks of focus, but I would say the question isn't well scoped which you can see on the amount of views it received. No one really cares about it.

So providing feedback without the need of engaging with the user directly which can be time consuming and not in the interest of an answerer is a good compromise in my opinion.

We recognize that well-formed, opinion-based questions have value, but we also know that not everyone cares for them. Our goal is to make this content opt-in/out based on individual preference.

I agree here, additional space for additional content that may not be requested by the SO-community but requested from the outside to attract people to this site is a legit interest of the company, in my opinion. You have to pay for the hosting and have to please others than the community of SO as well. That's fair.

In addition I think there is a fairly big amount of people who wants to exchange their views and grow by it. A bar below [newest] [active]... with a selection of [all] [debugging].. would sort things out.

The thing that annoys me the most about that feature, is that it is just been mixed with my answers. The answers must be sorted out as well. I don't know if we want that to be as comment or a different reputation system or whatever, please go ahead and figure this out. But I'm sure that this thingy is mixed with my answers does not just hold me personally back from using that feature.

We want to prevent the “hard lock” failure mode where questions are closed against a “duplicate” that may not actually solve the asker’s specific problem.

Fair enough.

we would also offer a curation flow where one or more of the canonical duplicate’s answers would appear directly on the duplicate question, with added context if needed.

As a first automatically generated feedback, that's good enough and the ability to write extra context, for people who wanna do the hard work of pointing the finger on it, that's great.

Then the asker would still have the opportunity to accept the proposed duplicate answer or leave context explaining why it doesn’t answer their question.

IF that means a veto-right by the asker, thats a really bad idea. While nowadays it's less common, but we also have had trolls here they just want to annoy you by interacting with you.

Besides, we do have the meta-site for it, but many of us don't bother to go here, searching a ton of meta question to find the subject it is discussed. If we automatically generate a meta-post and post a link below, where askers should go anyway if they disagree, I'm in. That just helps people to navigate in the process in which is agreed upon.

This also could make the closing/reopening wars more transparent.

Maintain Stack Overflow’s focus on programming without making “Off-Topic” feel like a shut door for users with legitimate technical questions that may belong elsewhere.

Great one. Even veterans have sometimes issues knowing, where the question belongs. If it's about server, Linux-specific or whatever. This also supports smaller sites and again navigates users through the fairly complex structure of SE.

PS: I still wish I could see the related questions of smaller sites here in one questions feed. But I'm unsure how one would implement it. An orange tag featured by CodeReview or even something in the title, still maintaining to sort it out, would been really cool in my opinion.

We won’t rehash it too much. We think the argument we made in our modernizing curation proposal covers it well enough.

The only thing that is clear is that your proposal to modernize community curation isn’t supported by the community which suggests the proposal is going against the direction of the community (similar to the grain of wood). Going against the grain of wood results in the wood splintering. I definitely feel the community is struggling and starting to splinter.

It’s extremely frustrating to see that every new experiment and user experience results in absolutely no community moderation and no consequences to users asking out of scope questions using those new experiments and workflows. You release a beta website without any ability to downvote or upvote the content, or at least a way to see the score of that content, and the threaded conversations make it difficult to tell the difference between a useless comment and a helpful answer.

We want to take each close reason and move it towards a curation workflow that fits in more reasonably with Stack Overflow today. We want to differentiate “this content doesn’t belong on Stack Overflow” from “this isn’t a question I am interested in,” which is different from “this is high-quality content that is worth promoting”. We want tooling that lets those signals exist independently, and we want curator input on what that actually looks like in practice.

In my opinion, the only individuals who perceive users closing questions they are not interested in are those who lack the motivation to enhance the quality of their questions and are consequently upset that the community deems their question to require improvement. Every “blog” about the decline of Stack Overflow, has one key similarity. They are from users who, asked a question, and refused to improve the quality of their question once they were told the issue was exactly.

Additionally, that tooling you describe, is long overdue. It is very difficult for the community to moderate today, the current tools are absolutely abysmal, and are long overdue for a complete overhaul.

Before making major changes to curation, the boundaries of the "things" we are talking about should be clarified. This includes the types of content to be handled and where they will be handled.

The identity of "Stack Overflow" is unclear. The new logo, "open-ended-questions", the "redesign", and the "what next for curation" are worsening this problem. The opening of the chat to everyone is also worsening it.

What could be the guidance to choose where to post an inquiry for information and help? Should it be posted in the chat or on one of the "main sites"? How should we tell the user that there is no place in the Stack Exchange network for their inquiry when that is appropriate?

Goal: We want to decouple quality signaling from the asker’s ability to receive an answer. We want to create more space to ask questions that may not be perfectly formed on their first attempt.

The entire point of the "too broad" close reason is that some questions can't be reasonably answered as formulated because they require too much guidance or far too long of an answer. I don't understand exactly how allowing answers to questions that can't be reasonably answered in their current form is supposed to help.

If a question can be reasonably answered as formulated, it probably shouldn't have been closed in the first place.

Idea: We are thinking about structured downvote reasons. Preset categories that provide context beyond a negative score. We are looking to adapt the successful feedback mechanisms from the Staging Ground to work at the scale of the main site.

Variants of this have already been proposed and rejected a zillion times. The reasons that this is a bad idea have already been enumerated in Why isn't it required to provide comments/feedback for downvotes, and why are proposals suggesting this so negatively received?

Goal: We recognize that well-formed, opinion-based questions have value, but we also know that not everyone cares for them. Our goal is to make this content opt-in/out based on individual preference.

I think that there's a misalignment on what constitutes a "well-formed, opinion-based" question. Other sites, like Literature Stack Exchange and The Workplace, have proven that questions that are at least somewhat subjective can work in a Q&A format, subject to the constraint that it's possible to reasonably defend answers with evidence. Answers are also expected to be supported by evidence.

Chitchat type discussions that are not expected to answer the question or be supported with evidence may be entertaining for participants but are largely useless to other people.

Idea: There could be a classification layer using categories using question types like “Tooling Recommendation,” “how do I,” or “implementation guidance." These would allow users to ignore, follow, etc., just like other tags, and only see content that they are interested in.

There's already a tool recommendation site, so I don't understand what the point of this proposal is. If people want to ask these questions, we should be more aggressive about directing them there. Also, these questions were banned on Stack Overflow because they simply don't work well on the site and it's not clear to me why they would go better now. See: Q&A is hard, let's go shopping!

Our long-term vision is smarter backend routing. A system that identifies where a question would be best suited, whether that be adding certain tags or considering other Stack Exchange sites.

I would strongly support that. Picking the correct site can be a little overwhelming for new users (especially with more obscure sites like tool recommendations).

Well I can admit that the shift in curation tools (and curation in general) aims at proposing a more friendly access to the site to beginners. After all we once had been beginners, and helping them is fine.

But the actual result is that the new design is awful not only for curators but also for answerers. I am around for quite a while, and had always tried hard to be kind to newcomers. I consistently left comments when downvoting to explain why the question was poor and what should be done to allow others to post useful answers. Or just passed my way when a question had already be closed or heavily downvoted. But as a result, I could always find questions deserving answers or at least useful comments. But now, most of the questions have just no interest at all. Having someone explaining that they have a compilation problem, simply because they failed to compile their C++ source is not the kind of question I want to answer. With the old system, I used to vote to close and leave a comment explaining that the solution was to use a beginner oriented tutorial. With the new interface, I just move along without even a comment, because I do not want to post a true answer.

What I mean is that the SO site contained high quality answers and the SO community had a number of specialists in many programming domains. And others who just received a lot of help from the site and were ready to in turn give some help. But my opinion is that with the new design, many more questions will be accepted, but few will receive good answers because after reading 10 questions with no interest, most people will just leave the site.

I'll address one category at a time.

Needs more focus / low quality

We want to decouple quality signaling from the asker’s ability to receive an answer.

The worst thing that can happen to a question is for it to be posted, and then later receive edits so significant that it essentially becomes a different question. This will invalidate existing answers. All existing comments and quality signals, both on the question and on the answers, are now meaningless and misleading. Moderators go to great lengths to prevent or roll back these types of edits since they completely break the site's quality signals. Most of the questions that are closed in this category are closed because there is not a way to answer them without edits of this magnitude.

Closing these questions serves several important purposes. It disables interaction with the question, which prevents the creation of quality signals until after the question is in its final form. This is good for answerers because it prevents them from wasting time posting an answer that will likely be off-base and downvoted/deleted. This is good for future readers since the "closed" banner provides a clue that the answers may be based on a misinterpretation of the question and may not be relevant. It's also good for the asker because now they can take their time and revise their question without getting downvoted into oblivion in the meantime.

Leaving these questions open would only cause our quality signals to be less trustworthy, and without that trust there's no reason for someone to visit this site. Put another way, people don't use this site to get "an answer". They use this site to get "a correct answer", and it's not possible to give them that unless answerers can understand the question accurately.

Opinion-based

There could be a classification layer using categories using question types like “Tooling Recommendation,” “how do I,” or “implementation guidance." These would allow users to ignore, follow, etc., just like other tags, and only see content that they are interested in.

Migrating the question to the Software Recommendations sub-site would be a better solution since they're more likely to receive a quality answer on a site dedicated to that sort of question. Your broad examples of "how do I" or "implementation guidance" would generally be perfectly acceptable at the moment and it's unlikely they'd be closed as "opinion-based" unless they were particularly poorly written (which goes back to the point above).

In general, tag filters are not a good solution for stuff like this. Those only work for logged-in users, and most site users are not logged in. A significant portion of this site's traffic comes via search engines, and those search results will not take my tag filters into account regardless of whether I'm logged in or not. Tag filters are meaningless for the vast majority of site use cases.

Duplicates

We want to prevent the “hard lock” failure mode where questions are closed against a “duplicate” that may not actually solve the asker’s specific problem.

That's not how closing as a duplicate currently works. The workflow you propose already exists. The asker can provide clarification about how the proposed duplicate isn't the correct answer and flag the question to be reopened. I see this happen on sites like Worldbuilding all the time. New users may not be aware of this workflow, but that's just an onboarding problem.

I highly recommend that you do not include content from the proposed dup target inside the question that's being closed. Answers should always be read within the context of the exact question that was asked. Answers will frequently reference variable names, functions, etc. that were mentioned in the question. When you blindly pull in an answer from some other question, the reader will have no idea what these names represent and the answer will be on the spectrum between confusing and a complete non sequitur. There's even a possibility that the current question and the dup target use the same names for completely different things, making the transcluded answer impossible to interpret correctly. I've seen this done on other platforms, and it's absolutely frustrating for readers. New users quickly get the impression that the site doesn't give very clear or accurate answers and don't come back.

Off-topic

Maintain Stack Overflow’s focus on programming without making “Off-Topic” feel like a shut door for users with legitimate technical questions that may belong elsewhere.

The easiest solution would be to give us the tools to migrate questions to the appropriate site. Not just a fixed list of a handful of pre-programmed migration targets. Any site. This is something that we've been requesting for a long time.

There will always be some number of questions that don't really have a home anywhere and will be closed as off-topic. "Off-topic" doesn't mean that we don't want to answer it, it means that it's very unlikely that this community has someone who can provide a quality answer to that question within a reasonable amount of time. If I'm asking a question, I would absolutely prefer to have someone be honest and let me know that my question is unlikely to receive a response so that I can seek assistance elsewhere. If we don't close them, they just stay open for an indefinite amount of time and attract no answers. The asker checks the site every day in vain, eventually giving up and feeling ignored. It's not fair to the asker to leave them hanging when we can tell quite quickly that the topic is not a good fit for the set of subject-matter experts that use that particular site. From a reader's standpoint, old questions with no answers are one of the most frustrating things to find when searching for a solution (example). Our relative lack of those is one thing that makes this site so attractive to users, and closing off-topic questions is a big part of how we do that.

General

Overall, this seems like you're trying to take down Chesterton's fence. These systems were all designed very intentionally to solve specific problems that have existed in the site's history. Making the proposed changes would undo those fixes and bring back those problems. The proposal reads like it was designed by someone who has very little experience using the site and doesn't have a deep understanding of why things are designed the way they are. Those are the most dangerous people to listen to because they're willfully ignoring years upon years of experience, hindsight, and lessons learned.

Please don't lose focus on the most important people: readers. A question has but one asker, but can have tens of thousands of readers over time. Those readers are the ones whose ad revenue pays your bills. Some site processes may not give the preferred experience to the asker or the answerer, but that's because the process is maximizing value to the readers. Making changes that degrade our quality signals or that make it harder for readers to find accurate, trustworthy content will have a significantly larger negative impact on the site than the small positive impact that comes from tweaks to the asker experience.

While closure was an effective tool to manage higher volume in the past, we believe it is now a significant cause of friction to authors and curators.

It is no more friction than before. Nothing has changed that would have increased the friction. The friction was the quality bar. It was always a quality bar, and it always caused friction. All quality bars cause friction.

"Removing the friction" is not possible with "reducing the quality bar", that friction is a necessary part of keeping quality high.

The truth is that SE sells question and answer pairs as data to LLM providers. And for that income it does not matter if a question was duplicate, off topic, opinion based or "in need of improvement".

In fact, you might even get money out of the last category. Because LLMs chatbots need also to handle confusing unclear user input, and if this community invests much time in making sense of badly asked questions, that help train LLMs to do the same.

You want to milk the goodwill of volunteer contributors by increasing their efforts for your gains. If a volunteer contributor could answer 10 clear on-topic different questions in the past, you figure they might also answer 4 duplicate questions yet again, provide some opinion on 2, answer off topic on 2, try hard to make sense on gibberish for 2, and answer 2 genuine new questions.

Since the overall traffic went down, that's not a big ask to those volunteers, they answered 10 questions before, they'll answer 10 questions now, and you can monetize on that effort spent.

Needs more focus / low quality

Goal: We want to decouple quality signaling from the asker’s ability to receive an answer. We want to create more space to ask questions that may not be perfectly formed on their first attempt.

SO already works like this, at least under C++ tag that I observe. Even the most noobie question (e.g. Why are these random numbers printing in the terminal? Everything else seems to be working fine) gets plenty of feedback, comments explaining what went wrong, links to duplicates, even an answer before it was closed. Sure, it was downvoted (personally I upvoted, because it has mcve and for a beginner it's not easy to google the problem), but answered nonetheless, so I would say there is no strong coupling between quality and ability to get an answer. The only time you don't get an answer is when no one can understand what you are asking for, e.g. here Are there any quick sort libraries that expose the current pivot element?, but note the effort experienced users put to pry for the details from the OP.

For the imperfect questions on a first attempt we already have a solution: Staging Ground. And again, new users are getting plenty of feedback. E.g. here: https://stackoverflow.com/staging-ground/79899849, experienced users are trying to help as much as possible, even though crucial details (environment configuration, CLI commands to compile the projects) from OP are missing.

From that point the rest of the effort necessary to get an answer is 100% on the OP side, and no amount of changing SO interface to make it "friendlier" will help, when someone doesn't want to be helped.

Opinion-based

Goal: We recognize that well-formed, opinion-based questions have value, but we also know that not everyone cares for them. Our goal is to make this content opt-in/out based on individual preference.

Thank you, good for me as long as I can filter these questions out from my feed.

Duplicates

Goal: We want to prevent the “hard lock” failure mode where questions are closed against a “duplicate” that may not actually solve the asker’s specific problem.

It's like that already, though. When we vote to close as a dupe user has the ability to edit the question with additional details explaining why the duplicates don't solve his problem. What else can be done here?

Off-topic

Goal: Maintain Stack Overflow’s focus on programming without making “Off-Topic” feel like a shut door for users with legitimate technical questions that may belong elsewhere.

Yeah, it's like that already. There is a significant overlap between StackOverflow, ServerFault, SuperUser, we usually don't "off-topic close" the questions on SO when someone asks e.g. how to do something in Bash, or run something in Docker. What could be improved is to allow close reason "belongs to another StackExchange site" with the list of ALL sites in the network. Currently we only have meta.stackoverflow.com, superuser.com, tex.stackexchange.com, dba.stackexchange.com, stats.stackexchange.com.

To summarize, I'm not sure why to put effort in "modernizing curation" if the currently available tools already satisfy most of your requirements.

Would it be best to do this more systematically? E.g., for each of these topics, make a list of all theoretically possible solutions (don't worry about whether or not they're "good" at first), and when we have a fairly comprehensive list, run them through the gamut of meta, objectively identfying the pros and cons (and not just automatically assume "status quo = best").

E.g. for "duplicates" alone, there are many plausible mechanisms:

1. no change;
2. label re-posts as duplicate, but don't close them (i.e., allow answers);
3. allow users to both "label as duplicate" (don't close) and "close as duplicate";
4. differentiate between "exact same question" and "in the same ballpark";
5. differentiate between "frequently asked duplicates" and "one-off duplicates";
6. close the older question as the duplicate of the re-post;
7. wholesale allow re-posts (abolish closing as duplicates altogether);
8. add a time delay between "label as duplicate" and "close as duplicate" (temporarily allow answers);
9. don't close duplicates if it's a re-post of something older than [threshold date];
10. allow re-posts at privleged user discretion;
11. maintain a shortlist of repeatedly asked questions and close those.

"We've considered many options, and we chose the best of those" seems like it will result in a more optimized decision.

We are thinking about structured downvote reasons. Preset categories that provide context beyond a negative score. We are looking to adapt the successful feedback mechanisms from the Staging Ground to work at the scale of the main site.

There is another suggestion how it can be done: attach text highlight tool to votes with the main aim to make downvote feedback fine-grained.
Highlight presets could be just 5 colours "strongly agree, agree, too unclear, disagree, strongly disagree".

Advantages: voting is not hindered, feedback is anonymous.
Disadvantages: "main problem of most questions is the lack of information, so there is nothing to highlight" (from Basilevs's comment).

If this tool will be used, in many cases instead of "for one of the dozens of possible reasons Q or A was downvoted" feedback will be "after reading this specific sentence multiple users decided to downvote". Or instead of just upvoted post, there will be fast instrument to say "upvote, but strongly disagree with that paragraph".
And since aim (not necessarily community aim, but SE management aim) is to improve experience of downvoted users, some incentive can be added to make feedback not only negative. For example, +1 (total -1+1=0) for marking at least one sentence as positive and at least one as negative when downvoting.

Design wise default representation could be just an additional summary line similar to citation line, but coloured (similar to IDE's). With an option to switch to extended view to check exact highlights.
Handling edits of already highlighted texts will be non-trivial, but still seems possible, especially on downvote part (on downvotes feedback is mostly for the poster, who will see it before edits).