The Stack Exchange reputation system: What's working? What's not?

What's broken: uncapped rep per post. Long-time users have had the luck to answer basic questions that are very popular to search, and gather hundreds, thousands, or 10k+ upvotes. This rep gain is completely meaningless - it doesn't reflect greater skill, knowledge, or any other trait that should reflect 'authority'. Meanwhile, newer or more specialized questions are inherently less prone to such a boost, yet at the same time demand much greater skill. From my observation, this is very pronounced. A hard-earned +2 post isn't worth x50 less because it didn't make it to HNQ and get +100.

Solution: cap rep gain per post, and apply retroactively. I don't know where exactly, but +100 net votes seems reasonable.

Solution 2: every upvote after +50 earns 5 rep; 2 after +100; 1 after +200. Catches up with the first solution at +250 while affecting more posts but keeps a permanent incentive. Also retroactive.

It doesn't have to be these exact solutions. I think this should be its own featured Meta Q&A.

This is something I was hoping you all would explore sometime soon, and I'm excited that you've posted about this now.

For awhile now, I've felt that tying moderation privileges to reputation... Doesn't really make a whole lot of sense. Whether one is capable of utilizing particular moderation privileges to better the community should not necessarily be determined by the quality, or quantity, of their answers on a given site. Rewarding users who know how to use particular moderation privileges well with more power in those particular areas is, in my opinion, the better way to go. You guys currently do this with the flag posts privilege, where you get 1 bonus flag per day per every 10 helpful flags, up to a maximum of 100. That's great, but you stop there. No other moderation privileges are expanded upon when you've shown that you're a "good" user of them. Giving close votes and delete votes this treatment could prove to be difficult (the question of "What is a good close vote?" is a difficult question to answer!) but there are some nice folks I know on Stack Overflow who would appreciate the ability to do more without committing to something like becoming a diamond moderator.

I'll take your questions one at a time and give my 2 cents:

What is really great about our reputation & privilege system? What do you think makes it great, and why is it worth keeping?

For the privilege system, super simply put, the biggest contributors get more access to help guide the site the way they see fit. The ability to unilaterally edit, vote to close, review posts and edits by lesser-privileged users, see deleted content... All of that empowers a veteran user to just do more with the site. The reputation levels that are currently chosen for the more powerful privileges (such as access to moderator tools and trusted user) are intentionally high to ensure that community members that have contributed a lot, and likely over a long period of time, have access to them.

This model has worked for a long time, and for good reason. Reputation as a number alone provides incentive to post, and tying the ability to do more with the site provides even further incentive. I would never suggest that we do away with reputation because it's a core part of the Stack Exchange model.

When we're just looking at reputation the number... I mean, it's just overall a great way to incentivize good contribution. There's a degree of accomplishment one feels when they answer a question that leads to resolving a problem someone has, but getting "paid" for it? That never gets old.

What is broken about it, and why? Are there any solutions to it?

Well... As you noted, the more saturated a given site becomes with good quality Q&A content, the more difficult it is for newer users to find a place where they can contribute meaningfully. Someone who is interested in helping with site cleanup efforts has a long way to go to get access to the ability to review, edit posts without review, and vote to close. They do get access to the ability to flag quite early, which is quite important and quite powerful, but the march towards getting anything beyond that is long and, sometimes, difficult when attempting it on a site that's quite saturated with good content. You will eventually experience a situation where a user has thousands of helpful flags, a great suggested edit history, a great knowledge of how the site works, and a good head for user-level moderation... And they simply can't contribute in the way someone who has posted more questions/answers can. I have one particular user in mind as I type this who is an absolutely exceptional example of someone I'd like to see be able to do more. And... That kind of sucks.

This sort of touches on a personal note for me, as well. I've got 500 reputation on Stack Overflow, which allows me to see some review queues, but... I frequently find that there are a lot of situations where I wish I had the privileges I do on this site. Unilateral edits (and the ability to help out with Stack Overflow's suggested edit review problem), the ability to vote to close, the ability to help to delete posts that need it... I mean, if I review any post, I frequently find myself skipping because the post needs an edit but the suggested edit queue is full. I've got thousands of helpful flags and plenty of approved suggested edits, and I'm trying to help improve a question that was presented to me in review. Why am I being stopped here?

The inability for newer users to find questions they can post answers to on saturated sites also encourages answering questions that are either known duplicates or that are duplicates but the duplicate just isn't found (yet?). The thing is, there's no incentive to find a duplicate, whereas there's plenty of incentive to just answer a duplicate. Better duplicate discoverability would be good, but some incentive for finding a good duplicate would be great. However, if you explore this, I would exercise caution, as it may incentivize incorrect duplicate closure, which is an extremely frustrating experience for a new user.

What other systems work really well that we could learn from?

There are not a lot of sites that reward user participation with (moderation) privileges like Stack Exchange does. From a reputation/content scoring level, there are a fair few sites that allow upvoting but remove downvoting, which... I don't know if I particularly care for, since there's no way for old content that may have become obsolete to be reduced in "usefulness-level", but I suppose it does make for an experience free of the typical derision that downvotes receive here on SE.

While I don't feel like you'd particularly appreciate the mention, Codidact does feature an abilities system, all of which reward users who have a good user-level-moderation track record with the ability to do more, which is in addition to a reputation system. Competitor site or not, there's something to respect about users who created a whole platform (out of their love for this platform!) rebuilding the privilege system in their own vision, and in a way that makes sense.

I think you're essentially going to be paving your own way here. My suggestion would be an implementation of something that's side-along to reputation, rewarding the good curation work that many users are interested in performing without invalidating reputation's longstanding meaning. A good idea you've been exploring is roles, but I feel like none of the suggestions tackled a particular pain point that can be useful across the network: The idea of opening up the full edit privilege to more users who have shown to be good editors. That's a great privilege to start this sort of exploration with.

Just a friendly reminder that these proposals exist:

• Reward duplicate closing: +5 for the first voter

• Reward finding duplicate questions - +10, +2, -5

• Duplicates Mark II, Canonicals, more aggressive duplicate prevention

• The top answer to Should there be a deterrent for answering obvious duplicate questions? (on Meta Stack Overflow)

• People MUST be rewarded with reputation for finding duplicates (on Meta Stack Overflow)

• This answer to Create CW answer for every "Vote to close as duplicate" vote

It’s better than nothing… but I have many complaints regardless.

The following is a rough sketch listing some observations and ideas about the reputation system I had accumulated over the years. I didn’t have much time to properly flesh out the explanations here, so expect some simplifications and leaps of logic. (I can also recommend @Karl Knechtel’s answer with a similar list, but I think much better written, and it manages to cover a few issues I overlooked writing this one; yet another case of the FGITW problem.)

A number of answers here express the sentiment that “oh, reputation doesn’t really mean all that much”; suspiciously often, it seems to come from users who have accumulated tens or hundreds of thousands of points. That makes it sound just as believable as a billionaire telling you that money doesn’t buy happiness; well, I am not falling for this trite so-above-it-all and holier-than-thou humblebrag countersignalling. This post takes the stance that reputation influences at least some people’s behaviour (especially those at the bottom of the ladder), and that’s enough for it to matter; it better not be ignored.

So here goes:

• Reputation concentration (the early-vs-late adopter problem): in early stages of a site, users can easily earn reputation by asking (and answering) basic questions, then earn “passive income”, as those questions accumulate views and duplicate links. This means reputation is awarded to early users disproportionately to their level of expertise, merely by virtue of them having asked or answered popular, low-hanging-fruit questions in the past.

  Meanwhile, for latecomers it can be hard to accumulate enough reputation to use a site comfortably, as their questions get closed as duplicates of the former… even though the older questions aren’t necessarily all that good themselves once you look closely: they implicitly mix unrelated concerns together, fail to state background assumptions or define circumstances in a sufficiently concrete way, are ambiguously phrased, attract redundant answers and ones addressing completely unrelated circumstances to the asker’s (the “Long Tail of Crap” problem) due to poor wording, and sometimes have bad titles. What is worse is that the moderators and the community tend to be apathetic towards those problems and sometimes actively thwart attempts to remedy them.

  The dearth of opportunities to earn reputation is very visible on Stack Overflow, where the only way to gain enough points to meaningfully participate and not be written off as a newbie is by collecting bounties; which is not all that easy, as bountied questions tend to be ridiculously specialised and demand complicated set-ups to even begin replicating the issue at hand. And even those are slowly drying out at this point: I remember there used to be 16 or so pages of unanswered bountied questions on SO. Now it’s just 9.

• Perverse incentives: Stack Exchange is basically Goodhart’s Law: The Website.

  • The well-known Fastest Gun in the West problem: early answers get the most upvotes and the most visibility, both of which reinforce each other; but usually not in proportion to quality. This encourages writing answers hastily and sloppily.
  • Highly-viewed questions attract redundant, poor-quality answers long after they have already been explained to death. (“Protecting”/answer-locking a question ameliorates this, but is not always available.)
  • Answers to poor questions can nevertheless earn the answerer some reputation, which encourages wild-shot answers, discourages voting to close unclear questions, and removes askers’ incentives to put some effort into improving the question. Additionally, if the question is closed regardless, the fact that it has answers makes it ineligible for automatic deletion.
  • Because downvotes are so underpowered relative to upvotes (more on that below), even a negative-score post will often on net earn the poster reputation. This means that posters, especially those at the bottom of the ladder, are disincentivized from removing posts judged as bad quality by the community. Privileged users may cast delete votes on negative-score answers, but negative-score questions are harder to get rid of (said above).

• Unfair skew towards upvotes:

  • It costs reputation to cast downvotes on answers, which means the few users who can cast downvotes are disincentivized from it anyway. The cost of downvoting was meant to entice users to carefully consider casting downvotes and discourage casting them indiscriminately, but there is no similar countermeasure for indiscriminate casting of upvotes. There is not even a penalty for upvoting a spam post.

  • The threshold to cast upvotes on answers is very low, which means the ratio of upvotes to downvotes is skewed by upvotes cast by low-reputation users, who have not been vetted by the site’s community as having expertise, and are less likely to have internalized how the site is supposed to operate.

  • Hot Network Questions and the association bonus distort voting further. Answers to HNQs get much more visibility and lots of upvotes from people who may not even be experts in the particular site’s subject area, or familiar with its house rules, but happen to have the +100 bonus from participation elsewhere and upvote based on superficial plausibility.

    I often see this dynamic on Retrocomputing. On ELU, answers to single-word requests are an embarrassment to read, also, I suspect, primarily for this reason. But the real poster child for undue HNQ interference is probably the the LangDev site: it used to attract a lot of opinion-based, bikeshedding topics, still kind of does, and did so even before it launched its public beta. The tragedy of LangDev has many aspects, and not all are related to reputation, but to the ones that are, the association bonus has certainly contributed a lot.

  • It takes five downvotes to nullify a reputation gain from a single upvote. It takes three more to nullify a reputation gain from answer acceptance alone, even though the asker is often a newbie, and therefore the least qualified person to judge how good an answer is. This means answerers are rewarded for posting superficial answers, and are not incentivised to correct mistakes discovered by outsiders that may be easily missed, yet important.

  • When democracy fails this badly, you might think plutocracy at least has a chance of working better, right? Unfortunately, the skew towards upvotes manages to spoil even the bounty system. Because of the large number of users who are not discerning with their upvotes, it’s not that uncommon that a mediocre-but-not-outright-awful answer (or worse) accumulates a score of 2, which is enough for it to get the bounty automatically, regardless of the bounty poster’s wishes. This means that bountied questions are subject to popularity-based perverse incentives against answer quality just as much as (if not more than, because of the higher stakes) normal questions, with little possibility for a bounty poster to correct for that. There used to be a way to decline awarding a bounty to anyone else, but now this can be accomplished only through sockpuppetry.

  • Toxic positivity: the nerfing of downvotes that was probably naïvely meant to prevent toxicity and encourage positive interactions has non-obvious deleterious psychological effects in the long term. Readers who are unaware of the quirks of voting mechanics and their long-term effects may have an inflated perception of a given post’s worth, while the posters may similarly have an inflated perception of their own expertise; the latter can breed arrogance. Users who notice problems with content feel powerless to act on their observations as their downvotes are penalized, drowned out by indiscriminately cast upvotes and therefore ineffective; they become frustrated and lash out. The lack of systemic solutions creates the feeling that nasty comments is the only kind of negative feedback that will be noticed. (I am no saint here myself.) And so, despite this ostensible effort to prevent toxicity, Stack Overflow manages to be incredibly toxic place regardless.

  As a particularly egregious example of vote skew: under this answer https://stackoverflow.com/a/14488421/3840170, there is a comment explaining why the proposed solution is a poor idea. When I first wrote this post in March 2023, the comment had 33 upvotes. The answer itself had only 7 downvotes, and 111 upvotes.⁰ Even worse one: this gibberish post https://stackoverflow.com/a/65856842/3840170 (archive link) managed to accumulate 66 upvotes and only 3 downvotes when it was finally deleted after me having to flag it twice, even though it is basically a Milla Jovovich answer.

  It’s the wiki whack-a-mole: invite a lot of newbies to add to the site without first teaching them about research and quality standards, then leave overwhelmed and frustrated curators to drown them in negative feedback, at best wordless or passive-aggressive by virtue of being impersonal canned boilerplate, at worst passive-aggressive by being handcrafted for the occasion. I suspect we are in the worst of both worlds now: newbie askers feel attacked because they see negative numbers next to their questions, while curators feel ignored because their negative feedback, even when thrice as numerous as positive feedback (which is rare), ends up being ineffective and swamped by the latter on the reputation balance sheet anyway.

  The number of people who say they never cast downvotes no matter what convinces me that discouraging downvotes does not really correct for a claimed bias, but it exacerbates an actual opposite bias that would have been problematic even without this measure. The Java fanatic downvoting everything .NET is nothing but a contrived bogeyman – who actually votes like that?

  There is also the bullshit asymmetry principle to take into account: it has always been easier to invent claims that are wrong, but sound plausible to an untrained audience than to conclusively debunk them in a way that is convincing to that same audience (which usually means: from first principles, without relying on shortcuts an expert would take for granted and easily fill gaps within); and with the advent of LLMs, generating large amounts of bullshit has become easier than ever.

• A measure of popularity, not quality, nor information value (aka the dev.to problem): An answer in a highly specialised area where very little information is otherwise available will probably receive just a couple of upvotes. Meanwhile, an answer about basics one might as well look up in Wikipedia can receive hundreds. In itself, it may be argued that this is intended: if an answer is useful to more people, then it’s more useful to the knowledge base, and the author should receive more credit for it. But given that votes are not necessarily cast by experts (see above), this means that posts will receive upvotes merely for covering a popular subject area, without necessarily covering it correctly. And someone actually knowledgeable in a niche area can have a hard time getting their expertise recognised; having barely a couple hundred of reputation points makes them easily equated with a newbie.

• Naïve scoring of individual posts. Computing post score as upvotes minus downvotes is nonsense. An answer having no votes at all is not worth the same as one having two upvotes and two downvotes: the latter is likely (given the previous point) to be something highly misleading which happens to have been upvoted by a couple of newbies who just don’t know better.

• No reputation for moderation or curation. There is no reputation to be gained from moderation tasks (raising flags, reviewing posts, identifying duplicates) or updating outdated answers, which discourages people from performing those otherwise rather dreary tasks. Only low-reputation users can earn reputation from suggesting edits to posts, and it’s a measly 2 points per approved edit. This also encourages writing redundant answers (see above), as there is more reputation to be earned from writing your own answer than from editing an existing one, even as a full answer may be not warranted.

All those problems mean that reputation is not a very meaningful or accurate measure of either expertise or contribution to the site, and not a very good incentive towards meaningful participation.

My proposed remedies:

• Make separate downvote and upvote counts on answers visible to everyone, instead of making it a privilege. Especially that a userscript can circumvent this anyway: "View Vote totals" without 1000 rep. There is a good reason people hate YouTube for removing the dislike count.
• By default sort answers based on some kind of Wilson score (shameless plug), or other more meaningful statistic, not naïve upvote minus downvote score.
• Raise the reputation loss from receiving a downvote on an answer. Perhaps introduce an incremental penalty: first downvote subtracts 2 reputation, second downvote subtracts 6, third and all subsequent ones subtract 10. (Not making them equal from the start should soften the blow from tactical downvotes cast for spurious reasons, which I expect to be rare.)
• Raise the reputation threshold to cast upvotes on answers to the same level as for casting downvotes.
  • At the very, very least: make the reputation threshold to upvote answers not take the association bonus into account. The bonus should only grant privileges like commenting and flagging, that are subject to moderation review.
• Implementing the previous point will likely incentivise now-“disenfranchised” users to upvote questions more instead (which we might want to encourage anyway: questions tend to be under-upvoted), which may skew question scores upward; to counter this, lower the reputation gain from a question upvote back to 5 points. Maybe to a bit higher level, but probably no more than 8 points.
• Lower the reputation reward from answer acceptance to just 5 points. Unpin the asker-selected answer from the top on all sites.
• Implement sub-linear reputation gain from upvotes, e.g. logarithmic (the first n upvotes give you 10 reputation each, the next n upvotes give you ¹⁰��₂ = 5 reputation each, the next n upvotes give you ¹⁰⁄₃ reputation each, etc.) or square-root (getting the first n upvotes gives you 10n reputation, the next 2n upvotes give you 10n reputation, the next 3n upvotes give you 10n reputation, etc.). This should solve the “too much reputation from just one very popular post” problem better than the daily reputation cap.
• Deny all reputation to answers on negative-score questions; possibly also on positive-score closed ones.
• (Much later edit:) In the answer author’s “signature box” at the bottom, display the user’s highest absolute score and relative rank in the particular question’s tags, preferably more prominently than overall reputation score.
• (Even later edit:) Bring back the option to decline awarding a bounty. Automatic awarding should only be used as a measure against inattentive bounty posters.
• (Much, much, much later edit:) Introduce a reputation penalty for upvoting a post that is later deleted, whether by community vote or unilateral moderator action: a smaller one (say, −2) for questions and larger one (maybe −4 or −6) for answers. Or perhaps a smaller one for community-vote/automatic deletions, and a larger one for moderator deletions. Or perhaps have the penalty scale with the number of downvotes instead of having a fixed one per post. This should make the system fairer by taking into account that if you upvote a post that turns out to be useless or downright harmful, it reflects poorly on your judgement and expertise. It would balance out the cost of downvoting, and it should also disincentivize bleeding-heart, indiscriminate upvoting and encourage users to review posts more thoroughly before upvoting them.

Not sure what to do about the moderation bit. Also, perhaps there should be separate reputation scores for subject-matter expertise and for moderation tasks, as those are definitely separate kinds of skills, but I only have the vaguest ideas how that would look like.

(Even later edit:) Also, we could somehow encourage closing earlier questions as duplicates of later ones. After all, just because a question was asked first doesn’t mean it’s a better one: questions on Stack Overflow are often formulated badly and mix unrelated concerns in one post, which sometimes gets unnoticed right after it’s asked, and only becomes apparent when answers accumulate. This could also transfer some attention, and therefore reputation, to newer users, instead of having it accumulate with long-inactive users, thereby ameliorating the “reputation concentration” issue.

These are, I admit, rather bold proposals. Do I expect any of them to be implemented? No, I don’t. I expect Stack Exchange Inc. is going to do with them exactly the same thing as with all community proposals: ignore completely, or perhaps even implement their complete opposite (given that they attempted to force through the 1-reputation voting experiment, despite immense pushback to the contrary). In fact, even if they were to be implemented exactly as laid out above, I would expect them to be pretty unpopular (despite the rather high number of upvotes this answer has received); they would turn lots of people who feel entitled to their meaningless Internet points very upset. The best case, I think, is that they will be read and taken into consideration by whoever builds the next Q&A website from scratch.

⁰ Since then, as of late October 2023, the answer has earned three more upvotes, only one downvote, while the comment received five more upvotes. If the meta effect has made any difference here at all, it seems to have made the problem worse.

What is really great about our reputation & privilege system? What do you think makes it great, and why is it worth keeping?

You need some way to sort the noise and votes are a decent way to do that. We also want to reward good posts and posters. You can earn privileges this way. The system does well here.

What is broken about it, and why? Are there any solutions to it?

How long you got?

There's some odd distortions the system creates with regard to reputation. The first would be the false dichotomy of "high score/rep means good quality". I've seen users in the last month who insisted that their decently scored post indicated it was good quality. I've seen terrible posts that got lots of upvotes. I've even seen rep used to feed posts that are blatantly off-topic (the worst would be service outages).

Then there's the people for whom reputation = life. I've suspended sock puppet masters, only to have them come begging for the lost rep back. One literally told me that his getting a job in India depended on having high rep on Stack Overflow. Wat?

There are too many people who want reputation for... reasons. As I've noted elsewhere, I have seen a direct pattern of people who get into vote fraud, get caught, and then switch to plagiarism (or vice versa). Maybe they like the big number. Maybe they want privileges. ChatGPT has just piled onto that list. It's hard to say. There's not always a clear motive there.

I'm not sure there's any solutions that don't exacerbate the problem.

What other systems work really well that we could learn from?

• Non-rep privileges. Want to do some rep-based things? Maybe take a test. We have no shortage of people who earn privs and then lose them through bad usage. it's not a great model. Maybe create a way to train anyone who wants to review
• Tag badges - We give you a nifty privilege for earning a gold one. What about silver and bronze?

What is really great about our reputation & privilege system? What do you think makes it great, and why is it worth keeping?

I think it's a great system for encouraging participation. If there was nothing to signify your long-term participation on the site, a lot of people who are here simply to provide answers would grow bored with it pretty quickly.

What is broken about it, and why? Are there any solutions to it?

What's broken about it, particularly on Stack Overflow, is it encourages users to answer posts that should be closed as a duplicate or off topic rather than voting to close because one is rewarded and the other isn't and is increasingly difficult to use with each additional one that is answered. It encourages users to ask for upvotes or to beg users to accept their answers, and it encourages users to fight against any moderation that goes against that gain of reputation. The incentives need to be balanced out to counter this conflict. I don't think simply adding reputation to moderation actions is the solution, rather, remove the negative rep of receiving and giving downvotes and reduce the incentives of accept votes and upvotes. Adjust how quickly users are introduced to these systems and allow feedback from anonymous users to be visible, perhaps through a new helpful/not helpful UI element that's available to everyone and doesn't impact reputation (not a replacement of voting, an additional feedback source without barriers).

What other systems work really well that we could learn from?

A lot of other systems leave out downvotes entirely. They certainly foster a more... "friendly" atmosphere, but where that hurts is with quality and the ability for the avg user to indicate that a given post is wrong/out of date. Limiting that kind of feedback to users leaving comments is a barrier that stops most people from leaving feedback at all. I think we could learn from these systems and better capture feedback from users, possibly in a way that isn't perceived so negatively but still accomplishes the goal.

I was asked to contribute this as an answer, so here goes. I'm going to directly quote from a blog post I wrote in 2020:

The problem with Stack’s approach

Stack Exchange — which Codidact largely grew out of — awards various privileges based on your reputation score, which is calculated largely by votes on your posts.

This… is a bit problematic, in a couple different ways.

First off, reputation itself is wearing way too many hats. It’s simultaneously supposed to be acting as:

…a general indication of a user’s proven expertise on the site’s subject matter. (More simply put: It’s supposed to give you a rough idea of how much that user knows about whatever that specific site is about. A high reputation score on a site such as Math shows that you probably know what you’re talking about with math-related stuff.)

…a general indication of a user’s familiarity with the site mechanics. This is why reputation unlocks privileges, and why users who have 200 reputation on another site are automatically granted 100 reputation when they join another site on the network.

…a gamification incentive. Also known as “Pointz”. People love earning points for their contribution; reputation serves as a gamification incentive by awarding people points when they get votes. (See also: Reddit karma; Imgur Internet Points.)

As a result of trying to do all these things at once, it winds up doing all of them… less than ideally.
Anyone who’s paid any attention to Stack Overflow Meta will notice that there are users with thousands of reputation points who have absolutely no idea how site mechanics or policies work. Sure, they’re probably good coders, but they don’t know how to edit a post or what the heck a “review queue” might be.
You can also earn up to 1,000 reputation by suggesting edits on a site. I’ve personally earned a bunch of reputation across various sites that I know nothing about, just by suggesting edits… but if I decide to post, my reputation score might lead people to think I’m more familiar with the subject than I actually am.
It works for gamification pretty well, I suppose. Not perfectly, but good enough… and getting into that is really off-topic for this article.

Circling back up to what I just said, I want to emphasize the important part: “…there are users with thousands of reputation points who have absolutely no idea how site mechanics or policies work.”
That is the problem… because reputation points give privileges.

As users with thousands of reputation points, they have access to moderation tools such as voting to close posts, voting to delete posts, unilateral edits, and more… because they proved they know how to answer questions.

They don’t have those privileges because they’ve showed they know the rules about when to vote to delete answers. A lot of them have no idea what the policies for deleting posts are, or how the privilege (that they have) works. They earned reputation through answering questions, and suddenly have the ability to vote to delete posts.

There’s another way

So now we’ve identified the problem: Users earning privileges through actions that are entirely unrelated to the privilege itself.

And now we have an obvious solution.

Users should earn abilities and privileges through actions that are directly related to the ability they are earning.

The rest of the post goes on to detail how we went about implementing this idea over there, but since we're talking about changing things here on SE, I can contribute a few more thoughts in that direction.

Rep is wearing too many hats; let's take at least one of them off.

Privileges, for the most part, should not be tied to reputation.

This is a point others have raised, mostly focusing on edits. I'd like to expand on that somewhat.

Perhaps the best example in the system we have on SE today is flags. You earn the privilege to raise a flag at 15 reputation^[1]. When you earn the privilege, you have a maximum of 10 flags per day^[2]. As you raise flags, you earn more flags to raise per day if your flags are being marked helpful, while you run into warnings and flag-bans if your flags are being declined.

This is how the system should work; as you prove that you know how to properly use each system, you should be trusted more within that particular system.

Extending that thought, if enough of your suggested edits are being consistently approved, you should earn the privilege to edit directly. If your close flags are being marked helpful and the posts that you flagged are getting closed, you should be able to vote to close directly. If your NaA and VLQ flags are consistently helpful, you should be able to VTD directly - and as your close votes and delete votes result in consensus, you should earn more per day.

While it makes sense to set a baseline level of reputation to be able to begin earning those abilities, in order to ensure some level of familiarity with the site and not to overrun people with poor review items^[3], we're currently locking a bunch of people out of helping with moderation just because they aren't prolific contributors to Q&A, without taking into consideration any effort they've put into other aspects of the site.

_{^[1]Aside from custom mod flags on your own posts, which you can raise at any rep, IIRC.}

_{^[2]Well, 10 flags for posts and 10 for comments, since they're counted separately.}

_{^[3]Although it's a fair argument that that ship has already sailed.}

Reputation is meant to be a gamified system - do a good, get a shiny. That already motivates some people to play the game to get shiny, not because they want to do good.

Add to that, reputation is meant to reflect the quality of action a user has taken, most significantly in the content provided for the sites, and it seems clear that the shiny you get for doing good must be balanced by the shiny you lose for doing bad.

Upvotes and downvotes must have the same impact on reputation, otherwise it becomes meaningless at any scale worth discussing.

Consider some situations that are extremely common.

A user who posts on a small site with a +1/-1 will have a net gain of 8 rep. Is this post really contributing the same value as four distinct approved edits, or most of the same value as a post with +1/0 score? A post on a larger site with a score of +10/-10 will have a net gain of 80 rep. Do we really believe such a post is at the same level of quality as one with +8/0 score? I absolutely don't think so. This only encourages controversy and appealing to a popular interest instead of fact-based informative content.

A user with enough rep earned off historic posts has cleared any barriers to posting. They won't get Q-banned without a moderator intervening, which most are loath to do to someone who's been around for a while. They have a silly idea pop into their head, and they ask a question. +1/-3 and, oh, wait, they've gained rep for posting a bad question?! And that's just a small site. Getting +3/-7 on some sites isn't hard, but that person is still gaining half the rep of a post at +3/0 score, despite being clearly of the opposite quality by the standards of the site.

A user finds a question that is definitively off-topic by the consensus policies established. They disagree and they want to answer it. +1/-4 because the unhelpful question asker is happy they got what they want, site standards be damned, and that's a net gain in rep for going against the community. Even larger if the answer was accepted.

If we're meant to be about using rep to gauge value created, value added quality generated, then we have to be doing that in both directions. We can't even get people into negative number reps anyway, so it's not like we know anybody below a certain threshold is a net drain on content and quality (and an increase in curation and moderation workload). The only way we can see a user's net contributions are not valuable, is if the only corresponding public measure is reduced.

And that requires losing the same shiny for doing bad as one gains for doing good. Anything imbalanced as we have now, makes rep into a certificate for participation, not quality assurance.

Two things to add:

• A few users on my site (I'm a mod on Academia) have done a lot of community moderation work (retagging, posting on meta, voting to close, editing, etc.) and would be good candidates for moderators. But, they have relatively low reputations, which effectively makes them unelectable. I realize "incentivizing" moderation actions is potentially problematic, but our current solution (not offering any reward at all) is perhaps a bit extreme.
• There is a huge first mover advantage -- answers posted in the first few hours will usually outperform answers posted a day later in terms of votes (and therefore visibility, and reputation). The green checkmark counteracts this to some extent, but this introduces its own tyranny -- askers sometimes make non-sensical selections, or do not select anything at all. There are several possible improvements here, the simplest of which is to introduce a post sorting order that will rank posts based on a combination of newness and upvote-to-view ratio, and allowing alternative ways to award the checkmark (or killing it entirely).

Make downvote criteria the same as upvotes.

Downvotes are not worse or any less important than upvotes. On the contrary, with the current system, downvotes are more important than upvotes, as they can only be cast by more experienced users and there are fewer of them.

Remove the -1 rep cost for downvoting answers.

Increase the reputation necessary to upvote to 125.

The only thing which I would leave the same is the reputation loss from receiving downvotes. With -10, users could feel like participation is too difficult. With the abundant revenge downvotes, this problem would be bad for psychological health of SE's users.

Reputation should not be gained from closed (non-duplicate) questions and their answers.

A closed question means that it should not receive answers. If it has received answers nonetheless, then these answers should be treated as invalid. Users should not be rewarded for asking popular off-topic questions. Users should not be rewarded for answering questions that should not be answered.

The only exception to this are questions closed as duplicates that serve as useful signposts.

(This answer is primarily driven by my experience with Stack Overflow, and my understanding of the Stack Overflow tour page; but I have confidence that the same principles should apply more or less network-wide.)

Whoo boy. I have so many complaints about the reputation system generally (and even more about how voting impacts on the Q&A, as opposed to users). I actually have a blog post planned about it for the somewhat near term, but I stumbled upon this Meta post so I might as well take the opportunity to gather some thoughts.

Perverse incentives

• Getting your question or answer upvoted awards 10 reputation; getting it downvoted only takes away 2. That incentivizes people to be controversial rather than boring. Boring is usually higher-quality and more encyclopedic.

• Getting your answer accepted awards an additional 15 reputation on top of the +10 from OP's upvote. This incentivizes tailoring answers to OP's specific circumstances, rather than trying to write in a broadly-applicable way that will be helpful to people who find the Q with a search engine. It also incentivizes tolerating questions that are clearly too broad (if you help OP with multiple issues, and make code with multiple issues work, that increases the likelihood of OP being positively overwhelmed with the helpfulness). Also, anything with that effect, has the ripple effect of discouraging editing (since Q&A are now more tightly coupled, the minimum unit of coherent editing increases).

• Accepting the answer also gives 2 reputation to OP. This incentivizes OP to accept an answer even if none of the provided answers actually get it right.

• Reputation is awarded the same way for answers even if the question is downvoted and/or closed. This incentivizes answering questions that are clear duplicates, because of the likelihood that OP will upvote and accept that answer. Such answers interfere with the deletion of low-quality signposts (although I am personally of the opinion that keeping around duplicate questions is useful because it helps establish the importance of their canonicals). It also incentivizes trying to mind-read OPs who don't express themselves clearly.

• There's an especially obnoxious cumulative effect here: the first upvote on a "fastest gun in the west" answer, along with the accept vote, nets a total of +25. It takes thirteen downvotes from concerned curators to make it unprofitable for unscrupulous reputation-seekers to answer questions that they ought to know shouldn't be answered. I've interacted with users like that many times. They often have 10+ user accounts and hundreds of thousands of rep; and their response to comments explaining the damage they're doing to the site ranges from a shrug to extreme indignation.

• There's a lock-in for votes on Q&A that weren't edited, which doesn't respect the fact that sometimes people genuinely change their mind. (One thing that's happened to me a few times is that I upvoted something because I thought it was the best canonical available, then found something much better and actually discovered serious issues with the first option.) This incentivizes 2k+ users to make dummy edits in order to reflect that changed viewpoint. (I try to make my edits substantial, but there have been times where I improved a post while also changing my upvote to a downvote... because I really don't know what I was thinking with the original upvote. Sometimes the upvote was literally years ago, and I still had to edit because of the lock-in.)

• Getting an edit approved awards reputation, but making unilateral edits does not. This means that people who reach the 2k threshold have a sudden shift from extrinsic to intrinsic motivation for improving the site. (This makes me want to launch into a whole separate diatribe about the edit queue, but I don't think it's on topic here.)

• Downvoting costs a point. I doubt that many people are seriously discouraged by this, but it bears noticing that the stats show a heavy upvote bias among logged-in users. Last time I checked the statistics, votes from users were more than 90% up, while "anonymous feedback" is close to 60%. (This is, admittedly, skewed by the fact that users between 15 and 125 reputation can upvote normally but have attempted downvotes logged as feedback.)

• Meanwhile, there is no reputation incentive for most curation activities. As above, that includes unilateral edits (2k+ users), and also:

  • closing duplicates

  • casting close votes on things that should be closed

  • casting delete votes on things that should be deleted

  • using the review queues (in any capacity whatsoever) - the only thing one can get from using the review queues is a ban from the review queues

(There are finer points I could make about the curation experience, but again I think they drift off topic.)

• Comments can be upvoted, but not downvoted. While this doesn't award reputation, it has psychological implications. I've NLN-flagged quite a few old comments now that had a significant vote tally on them - typically, comments on an answer complaining that the problem shouldn't exist or should be simpler to address, and typically ones that show a serious (but common) lack of insight or understanding.

• I'm not entirely sure how this works, but it's clear from my experience that moderators are disincentivized to delete duplicate answers on old questions if they're contemporary with the post. I don't get why, because this apparently would not significantly impact on the answerer's reputation - but I do think it bears mention in this section.

• Currently, Q&A get rewarded the same regardless of the topic matter (or the inherent difficulty of the problem). This incentivizes people to stay in already popular tags and allows less popular tags (where answers are often more needed) to languish in obscurity. While the "no sense of urgency" culture is a net positive, good questions really should get answered eventually. As it stands, our only workaround for this is an extremely clumsy "bounty" system. Which segues nicely into...

Linearity

In my mind, it makes very little sense to count each upvote and downvote on a Q or A the same, and it makes barely more sense to tally per-post reputation to give an overall score to a user.

• In my experience, both Q and A aren't really any higher quality past a score of +10 or so; they're just older. Often they're lower quality because a) the information has become deprecated or b) it's an A for a Q that used to meet site standards, but now it doesn't. There are any number of psychological barriers to downvoting things that already have a lot of upvotes, and even with "trending sort" it can be quite hard to lower them on the page. Personally, my highest-voted answer is at +346/-2 on a Q that's +444/-5 - and I've thought more than once about deleting it, because the question is still blatantly two unrelated questions, even after extensive editing to make sense of a barely coherent mess.

• In my experience, questions below -3 aren't really any lower quality than ones at -3 (except for spam/vandalism/trolling, which we do an excellent job of cleaning up). Even from -1 to -3 is a rather shallow gradient. A -3 score makes posts eligible for manual deletion; there's little point in going further than that.

• On the other hand, in my experience, answers at +2 are much higher quality than ones at +1 on average - because the +1 category is swamped by things that were upvoted by OP and nobody else. A +2 typically entails some kind of confirmation from someone with relevant knowledge and understanding. Most people asking don't have that (or they would have self-answered).

• Summing reputation gained across multiple questions doesn't make a lot of sense. For people who didn't get in early (and thus have the ability to gain "dividends" on old questions that come up a lot in search engines), the easiest way to game the system is quantity over quality. The first upvote on a question ought to count less (per the previous point), but it counts more (because of the accept system). But being able to write a lot of answers each day isn't evidence of expertise; it's evidence of addiction to a particular mode of using the site. (I shudder to think what would happen if the daily reputation cap had not been implemented.)

One-dimensionality

Even if the system worked as intended in the above regards, and incentivized people to write high-quality Q&A (and put A only on Q that should have them), and take positive steps towards curating the site:

• The ability to be helpful or insightful has little to do with the ability to be responsible - but we use the reputation system to award moderation-like privileges. (This has a knock-on effect that off-site critics mistakenly describe non-diamonds as "moderators".)

• Clear communication is totally separate from expertise, but we reward both on the same scale. A reputation score can't tell you who's an X guru vs. who is obsessed with X and wants to find a lot of things to do with it vs. who is skilled at creating simple demonstrations for common X gotchas.

• The site already has a separate system for recognizing expertise: tag badges. Why is, for example, editing of a tag wiki tied to overall reputation, and not to experience with that tag? Which leads me to...

Quirks of the privilege tower

• It takes 15 reputation to upvote, but 125 to downvote. I mentioned earlier that this contributes to a bias in the voting/feedback stats. It's also inconsistent with the widely-held view (at least on Stack Overflow meta) that downvoting is a vital curation tool. In particular, that makes it that much harder to sink old content that is deprecated (assuming it can't be salvaged; but trying to add up-to-date information to old As that aren't CW often gets you yelled at) or which no longer meets site standards (which changed since the content was posted).

• The site association bonus is enough to allow people to create chat rooms and edit CW posts, but not enough to downvote. This strikes me as very backwards. It shouldn't take subject matter expertise to recognize that a question or answer is unclear or unhelpful (or, say, politically inflammatory - and it's usually very hard to get site moderators to agree with R/A flags for that); but it might well require an SME to recognize a technical inaccuracy or an opportunity to rephrase something in proper domain language, and it definitely requires SME to chat about the subject matter of that SE site.

• It takes 50 reputation to comment on other users' questions. This means that a lot of new users who know a little and want to be helpful, end up trying to use the answer section to give feedback on questions (that should be closed). Even flagging takes 15 reputation, and doesn't let these new users explain specifically how they think the question should be improved. (Speaking of which, the flagging system is also a mess, insofar as low-rep users have additional flagging reasons that copy the close vote reasons, but work differently and independently from actual close votes, and are hidden in that interface.)

• It takes 3000 reputation to vote to close questions. That's absurd. It's saying that "I think this question has issues that need to be fixed and shouldn't be answered in its current state", and getting two other people to agree with you before anything is actually done about it anyway, is somehow a greater and more dangerous responsibility than unilaterally attempting to fix it yourself (2000 reputation). (It's especially absurd in an environment like Stack Overflow, where the large majority of incoming questions are terrible and the site would be much better off if new questions started in a closed state and had to petition to get opened.)

• It takes only 1500 reputation to create new tags, but 2500 to synonimize them. This is another bias towards creating junk and away from curation.

• It takes only ten reputation to answer protected questions. This is definitely not enough, at least for Stack Overflow. Lots of old, famous questions have way more answers than they need - and we don't close them because they meet site standards and closure is seen as a black mark, but we still let almost anyone contribute additional redundant garbage to them. There have been times when I had something useful to contribute to a question that already had dozens of answers - but it's accompanied by a sincere wish that half those existing answers would just disappear.

In my not so humble opinion (active Stack Overflow user for 10+ years, active forum user for 20+), you're coming at this from entirely the wrong direction. Paraphrased, this is your concern:

"We want more users to like our site, so they stick around and make our numbers bigger, because eternal growth is totally not a capitalistic fable and we're not yet big enough and still don't make enough money for our shareholders".

You can then tweak the reputation system, but people (most notably the ones you need to keep the site alive) don't come nor stay for the reputation system. People stay because they have time to spend and want to hang out with likeminded people, and/or they want to display their knowledge, and/or they want to help people who run into the same challenges they have.

And then there's people who use Stack Overflow as their resume, so they think they have to have a profile full of badges and points so potential employers will notice them.

Just accept that there are only so many people wanting and capable of giving meaningful, helpful, constructive, qualitative answers that actually improve the value of your Q&A site, and you're not scaring those away with any reputation system, and that there never has been and never will be a lack of people daring to ask questions.

Well, at least in my own experience, reputation as a motivator is a bell curve. You'd need to get pulled into being an SE regular for reputation to matter. I turned up on Super User because a site about computer stuff seemed interesting and branched off from there. And the last time I actively cared about reputation was when I was trying to work my way to 10k to see a deleted answer of mine (the system changed halfway through me trying to get it... but I got my 10k anyway!).

Most of this isn't actually about reputation. Bear with me; I'll come to a point eventually ;)

TLDR: Reputation is overrated.

Unfortunately, some of the actions taken to try to fix new user onboarding - I think Joel spelled out the problems really well 10 years ago but to be really blunt, quite a lot of what was done was rather hamfisted and... didn't work quite well. He is right about the community dynamics here, just not how it was handled.

To quote Joel

After a few years, an insider group of old-timers forms. They get to know each other. They know the rules. They know the history and the legends of the community. And it’s only natural to get little bit irritated when newbies show up who don’t know the rules.

And there's so much more here than the reputation. Even if a new user gets past the reputation hump - growing that circle of 'insiders' is essential. The newbies see the number and think that's important. It’s an entirely arbitrary number, and frankly past a point gets you nothing.

The problem isn't as much 'reputation' as much as how newcomers being onboarded. A new user with quality posts, or maybe even an intuitive understanding of the system is going to have an easier time than someone who just walked in via Google and/or is expecting a forum. There's different user stories for either as a new user and... neither is going to go "Hey! How do I make this number bigger?"

They are going to remember helpful comments, people editing/improving their posts and guiding them through their fledgling stage.

But as we’ve seen, our system and community in its current state has a high barrier to entry. Asking or answering a question back in 2009 was a lot easier as we were building the library of knowledge that exists today. There were a lot more questions yet to be asked!

Not entirely true. Our standards were much lower and we ended up often needing to change scope as time went on. Even as a very experienced user, there are new technologies I'm messing with (I'm currently working on and having trouble with Nebula VPN for example) and software and software UXes evolve over time. That there are fewer questions to ask, or we'll ever be done with new questions is a misnomer.

The definition of a good question and answer , and the barrier to that has gotten higher. As an oldtimer, working/asking mostly about personal projects on my techie sites, I'm happy to refine a question in my head for weeks before asking. I go back and work on old answers and write new ones over several writing sessions because I've had years to work this out.

I'm pretty sure if I started a new account, picked up a few projects, and started using the same tricks I used as a new user, I'd probably be able to hit 10k in a year. It’s not a reputation learning curve; it’s a skills and spoons learning curve.

We still want users to be able to participate on our site and feel as though they are a part of this community. But unfortunately, we’ve heard from so many of our users, particularly newer ones, that the learning curve is steep and it’s difficult to find ways to engage on the site.

I don't think anyone comes in here worrying about the depth of the mechanics and reputation. Much of the friction is not mechanical and about reputation; it’s cultural, with a mismatch of expectations between new users and experienced one.

The 'learning' curve is as much about learning to deal with how to ask 'good'/better questions, dealing with criticism (textual and votewise) and the mechanics of the site. Worrying about votes is very much an early/middle lifecycle thing for a user—when you're wondering 'can I ask my next question' or 'can I unlock the next privilege?'

I've often said that SE builds communities in spite of itself. Healthy communities need to go beyond the utility of knowledge sharing. As much as main site use, things like chat and meta are also really good, neglected onboarding tools.

Reputation isn't broken other than the attention put to it. 5 reputation vs 10 for a question upvote didn't 'break' the network - the reasonings behind it did.

Reputation as a number isn't 'unfriendly'; it’s the actions and meanings behind, say, a downvote that matter more. There are folks who tend to be a little too focused on reputation, but people are going to come here to find answers and stay for the community.

By the time you 'worry' about reputation, you're already hooked, so any new user who complains... isn't really new, just someone who's engaged already, but who is still finding their feet. It’s better to teach them where the ground is, than move the ground to their feet.

As an old user - well, what's the value in my 145k reputation here?

At this point I have name recognition on the site, so my reputation isn't just reputation. I've not unlocked any additional benefits from being a high rep user. My motivation isn't reputation for most parts - it’s what I can do to try to, depending on the time period, protect my other sites (where its the same story), push (what I think are) positive changes...

Reputation isn't as important as people think it is. Reputation is broken because people focus on it too much. It’s also a handy, mildly arbitrary way to gauge how well people know the system, even if it’s how to break it.

And because of that, well, I don't think 'major' changes are needed. Our privilege system, tied to reputation, could use some consideration though. What would be positive actions that could be rewarded outside reputation gain? What are actual pain points for new users? (Downvotes? Not getting the structure of a good post? Not being able to find help and mentorship?)—these feel more important than trying to 'fix' the reputation system.

As a mature system, the problem with the reputation system starts with the name, as "reputation" is related to trust, but the current reputation system is actually a points system. It's a single bucket for points that might represent different things, such as effort (among other things). A high reputation value doesn't always mean that someone has a high level of trust from the community (understanding community as the real people that use the site, not just as an abstract concept).

Also, the current system assumes that there is a direct relationship between these reputation points and values, skills, and knowledge. It's ok that the points system give access to privileges/benefits, but tasks and especially moderation responsibilities deserve to be handled in a better way.

The designation of site moderators already have defined things for this and experience alonge more thang 10 year on multiple sites. There might be things that require to be improved there things that could be extended for certain tasks an perhaps not necesarily on a per-site basis.

Tasks requiring skills

For tasks that requires skills, one way to handle them better might be to limit the rate of tasks that one user that can do according to the their experience level.

Examples:

gold tag badge holders

Those at the beginner level might use the dup hammer once a week, those at the intermediate level might use it once a day, and those at the advanced level could use it without restrictions.

To be promoted to the next level, the posts they closed by using the dup hammer should not be reopened within one month.

tag creators

• At the beginner level, they might create one tag.
• At the intermediate level, they might create up to five tags.
• At advanced level, no restrictions

To be promoted to the next level, the tags they create should have a tag wiki approved by an advanced tag creator. A diamond moderator might also approve tag wikis, but only when there are not advanced tag creators available.

Trust

For tasks that trustiness is a very important element, I think that we should elevate trust above a mechanical points system. It might be related and might be required to have access to have earned a certain level of reputation points, but privileges/benefits strongly promoted as being related to trust, IMHO, should be reserved for very specific privileges and preferably avoiding "mechanical" access to them (i.e. they might be assigned/removed by an election process and/or by site moderators/CMs).

One example might be participation on disputes of moderation actions. Anyone might start a dispute and anyone starts with the privilege to participate on disputes According to participation history in disputes their participation rate of participation in current disputes might have some limit. Let say, someone that derails conversations / present deceiptful content frequently in disputes might be limited to make one post a day instead of being allowed to post freely.

Gap

The current system already has several features that allow moderators to suspend specific privileges – i.e. when a user is caught repeatedly approving evidently very low-quality posts, that user could be suspended from doing further revisions, but could continue posting questions and answers, which is fine.

Something that might need to be reviewed regarding trust is the way that moderation disputes are handled. They might deserve their own trust point scale, as failing in conducting a dispute probably should lead to a whole site participation ban.

The core mechanics work; don't change them.

• Upvotes and downvotes to earn and lose reputation.
• Earning basic privileges: it's like a fizzbuzz test.
• Downvotes cost something. This makes them meaningful.
• A culture of learning by doing. (But not being an asshole about it.)

There are lots of comments here about reputation economics. If this needs fixing, please don't change the core mechanics to do so.

Illustrative examples that don't change the core mechanics:

• OverLordGoldDragon suggests a decay mechanism

• Multiple answers hint at separating "moderation rep" from "quality rep". I don't think this conflicts with my suggestion.

• Multiple answers suggest adjusting particular values (e.g. value of a question upvote).

Based on comments on this answer, maybe I should clarify that I believe the core "gameplay loop" provides necessary and sufficient short-term rewards for someone already committed to quality questions and answers. In my opinion, onboarding and longer-term economics are separate concerns (which I am eager to get involved in, e.g. when Staging Ground becomes available).

What is broken about it, and why? Are there any solutions to it?

It might not be "broken", but reputation (and voting, to a lesser extent) have an ongoing identity crisis where it is not clear what they mean. For example:

A user having a higher reputation than another user could mean any of these:

• Has answered more questions
• Has given more useful answers
• Has answered fewer, but more popular questions
• Has given more answers, but on tags that are not relevant to the current question.
• Answers questions that are easier to understand (and therefore, more likely to be voted on)
• Was first to help someone fix a common syntax error 15 years ago

The fact that every question and answer puts the user's reputation in your face encourages people to notice and interpret that number in some way. The problem is that within any random Q&A thread, this number has no reliable, coherent interpretation. I think most would agree that if you rely on reputation numbers to make choices about voting, you are using the site wrong, and maybe even harming the community. And yet, the question/answer page is designed as though we want you to notice to them.

There is nothing wrong with having cool Internet points for fun, but we have an incoherent blend of a merit system and an engagement system, with magical numbers displayed everywhere. In my opinion, this seems to be dragging against the goal of having high-quality, self-moderating information.

We still want users to be able to participate on our site and feel as though they are a part of this community. But unfortunately, we’ve heard from so many of our users, particularly newer ones, that the learning curve is steep and it’s difficult to find ways to engage on the site.

That's quite hard to parse for me. What "learning curve" does one need to search for and read answers, or post a question? Or is it the "learning curve" for gaining reputation points while not adding any value to the site?

Also, "find ways to engage": does one not know how to answer questions or edit posts to improve their quality? Or is it about, again, gaining rep points, for... what?

I guess my point is, how are the modifications of the SE reputation system can help "find ways to engage" and/or the "learning curve"?

Perhaps we would want to ask those users who think the learning curve is steep and can't find ways to engage to tell us more?

I want to add that in my main site, Earth Science, many of the users with privileges to moderate are not participating or doing moderation tasks (neither upvoting). Earlier users have answered the good questions and left the site, and the new users find it difficult to obtain the reputation that the users that have been there since the beginning and are still active (a few) have (they see they obtain a few upvotes for their detailed answers, while top users have 23K).

So as well as a loss in users, we have moderation troubles. For example, after we banned id-my-rock questions, it took months to completely delete the existing questions. I needed to open a post asking for delete votes so that the site would stop appearing first for searches as "Help identify my rock" in Google, whose status took two months to complete (while we continued receiving a flood of id-my-rock questions). There is also an accepted proposal of merging synonym tags from 2016, but the tags were never merged and we do not have people with privileges to vote to merge active in the site (you need to have answered some question with the tags).

I would say then, at least for Earth Science, the reputation gains for upvotes on answers and accepted answers should increase, or the moderation privileges reputation needed should reduce, if the site wants to survive (it is giving signs of dying).

One of the problems on Stack Overflow and many other sites is that it rewards with reputation answers to questions that instead should be closed as duplicates. To fix this:

• Provide reputation gain to the user who correctly marks the question as duplicate, for example +10 ("correctly marked" means that the question is finally closed as a duplicate). The reputation gain should be comparable to that from answering a question, because finding the duplicate requires non-trivial effort.
• Provide badges for users who correctly mark many questions as duplicates ("correctly marked" means that the question is finally closed as a duplicate).

A more specific proposal for handling per-question reputation

Working within the reputation system generally, several previous answers (including my own) seem to have highlighted some points of agreement:

• +10/-2 for up/downvotes is bad, and has negative consequences
• However, downvoting things into oblivion isn't that helpful in terms of correcting user behaviour, and tends to feed resentment
• Super-high-upvoted questions aren't really that much better than sort-of-high-upvoted ones
• However, we should incentivize answerers to aim for quality, rather than mass-producing answers that get +1s

Some citations:

There are also some pertinent comments as part of the discussion, particularly on Kevin B's answer.

To me, the conclusion is clear - we should count upvotes and downvotes the same, but scale per-post reputation in some non-linear way based on the net vote total. Specifically, the curve should be steeper in some low-positive total range, and shallower elsewhere.

I wanted to spitball some numbers and see how people feel about them:

Of course, it could be done with a mathematical formula of some sort rather than a table lookup, but I think this illustrates the sort of curve I have in mind.

The choice of 25 as the maximum marginal/delta of 35 is partly to compensate for re-introducing upvote-downvote parity and partly to compensate for the rep withdrawal on super-popular questions; but mainly to make the numbers nice and have more resolution to work with. I'm assuming that other numbers could be scaled at the same time if need be; but I see how that would have ripple effects on an established userbase.

Allow low-rep post feedback to be useable

When talking about limits on participation for new users due to low reputation, an important one is the ability to vote. Low-rep users are allowed to leave feedback, which could signal question askers and answerers that posts could be improved.

However, this feedback is nearly entirely inaccessible to authors of posts, which means new users can't properly indicate that posts are unclear or low quality to the author. Comments don't work but open an avenue for spam, but votes don't really allow for spam.

I understand that allowing 1-rep users to cast votes that affect reputation is undesirable since it could enable voting fraud, so:

Please make post feedback easily visible to the author of posts.

Currently, the only way I know of to view post feedback on your own posts is through SEDE. It would be great if your user profile could reveal post feedback by low-rep users, so it could be used to adjust questions. Ideally I'd have it split between logged in and logged out users, but I understand that might be asking for a little much.

Something to call out here before we dive into it:

...[W]e’ve heard from so many of our users, particularly newer ones, that the learning curve is steep and it’s difficult to find ways to engage on the site.

This is orthogonal to reputation. Being able to participate on a site is a different problem than what the numbers mean next to your avatar and what you get for having your numbers grow.

Getting users to participate and feel like they're engaging on the site means that you have to give them an incentive to engage. A lot of people use Stack Overflow as a means to an end - ask a question, get an answer, move on with their day. The folks who want to engage in a community may stay for longer. The folks who want to share their knowledge may stay for longer. None of this has anything to do with reputation.

With that bit out of the way:

What is really great about our reputation & privilege system? What do you think makes it great, and why is it worth keeping?

A lot of this boils down to familiarity, but it's really the sentiment that the system we have accomplishes a few things:

• Strong bands/divisions based on either participation levels or quality of answers
• Placing tools in the hands of power users, should they choose to use it, when they've had enough winters under their belts
• Something to strive for; while no one will really come close to Jon Skeet, they can at least shoot for the stars

What is broken about it, and why? Are there any solutions to it?

It is quite fragile, since it's the main measure of two things in my mind:

• How active you are, and
• How expert you are

To that end, these are the things that have broken it in the past:

• Giving questions 10 reputation, which inflates the amount of reputation that users actually get/earn from participating in the site
• Previous efforts which rewarded reputation in an unchecked fashion which were never recalculated (looking right at you, Documentation)

What other systems work really well that we could learn from?

While it's not perfect, I think the sentiment is that it's a more social variant of an Elo rating for everyone here. That is to say, the total reputation you have is a factor of how well your content is received against the wave of everyone else. This does make participation slightly daunting (which fine, you've caught me, I've come a bit full circle on this haven't I?), but it does mean that the system has a very explicit goal in rewarding those who are participating and whose content is comparable or better to others.

You don't explicitly state what your goals are for reputation, so it's not really feasible for me to say that something needs to change here. But I really don't think the system is particularly bad.

We just have to be very judicious when giving reputation out.

I think that the reputation penalty for downvoting answers is (became) a broken part of the system.

This was hinted to in one of the prior answers, but I would want to explicitly focus on this. To start with, I believe it worked reasonably well until the recent few months - maybe not quite perfect, but not worthy of raising serious concerns about either. Until recently, I could not even imagine that I would ever complain about this.

As it turned out though, this has been working based on a certain implicit barrier to posting answers. First, you had to have a certain level understanding in the topic of the question and second, you had to put some effort into well... just writing the answer - you know, figuring what to write, finding words, typing them, and stuff like that.

This barrier is no longer with us now that one can just copy the question into ChatGPT and copy its response into the answer field.

This has led to an unprecedented flood of low-quality answers that are in the sore need of downvotes to rate, and rep penalty makes this impossible - at least at the scale corresponding to that of the ChatGPT flood.

If needed, refer more detailed discussion at MSO

This looks like the most basic and natural way to establish trust model, proven to work reasonably well on questions and well integrated into the system (rate limits etc). This is what we miss for answers...

answer downvotes should be free unless these are cast on competing answers...

...to start with, multiple downvoted answers will throttle low quality posters (bot or not) - this is embedded in the system of rate limits. Besides, all the usual quality mechanisms associated with downvotes and negative score will start working at last: option of 20K-deletion votes, roomba etc etc. The last but not the least, people will just start feeling better getting the ability to rate content as they want and there will be less of bitter feelings (I believe this is very important)

First of all, thanks for bringing this up to discussion with the community before jumping the gun and implementing some change.

• What is really great about our reputation & privilege system? What do you think makes it great, and why is it worth keeping?

As noted many times, some people care a great deal about imaginary internet points and so it works well for attracting those kind of people. They might be people who contribute with high quality content and/or people who just answer endless duplicates, but overall the system works somewhat well for attracting users. Essentially it is the same thing as receiving a golden star sticker from your teacher back in elementary school - it appeals to some basic human need to be recognized.

The rep system works quite well for discouraging low quality and incorrect answers. In fact down votes work much better than up votes for filtering out quality over low quality.

What works somewhat well is that moderation tools and privileges are gradually unlocked. Things like review queues should obviously not be unlocked immediately. However, there is not much of a relation between high reputation and moderator suitability.

• What is broken about it, and why? Are there any solutions to it?

So many things.

• As noted above, reputation is a very poor metric for moderator suitability. It is also a somewhat poor metric for community trust in one particular user. If anything, reputation is a metric for how active someone is on the site.

• Reputation is completely out of proportion for Q&A under popular tags. Looking at the top users at SO, pretty much everyone in top 50 are active in the most mainstream, active tags. So besides being a metric of how active someone is, it is also a metric of how popular tags the user is following.

• For these reasons, reputation is also a very poor metric of technical knowledge. The bronze, silver and gold badges are somewhat more relevant.

• Reputation is also a measurement of "I happened to answer some question that went viral". Some who've asked and answered one particular up-voted post can gain ridiculous amounts of rep just because of that.

  As a perfect example from SO for those who know programming: What is the "-->" operator in C++? This is the single-most up-voted C and C++ question and only because it is about a feature which is curious, intriguing and exotic. Not because it is about relevant technical knowledge or a canonical duplicate. Looking at my own by far most up-voted answers on SO, the top 2 are also about such curious and strangely behaving code - while they are far from my highest quality answers posted on the site.

• Reputation encourages answering duplicate questions over and over again. Some users purposely do this to "grind rep" instead of closing obvious duplicates. It is always the same users.

  This in turn encourages low research effort by the posters. The easier it is to find the answer to a question yourself, the higher the probability that it will also get answered even though it is very likely a duplicate.

  And well, this problem has been debated endlessly over the years and a pretty solid, recurring argument from the answerer's perspective is that "it takes me less effort to answer the question than to find a duplicate". This is true, the lack of working on-site FAQ systems or other easy ways to find high quality duplicates remains an unsolved problem. Various sub communities tend to solve this by implementing their own tag FAQ, either on-site or off-site. I brought this problem up here as far back as 2016.

• The system is terribly inconsistent. Up-voting costs nothing and yields +5 or +10 rep. Down-voting costs 1 rep and yields -2 rep. This is ridiculous and leads to rep inflation. As noted earlier, down-votes are a much more efficient way to ensure quality than up-votes, so they should weigh as much and be encouraged.

  What about those users who do heavy moderation work in tracking down spam and close voting bad content? Why are those users penalized for making the sites a better place? They will no doubt down vote far more frequently than they up vote.

  A stray thought: maybe implement a system where a user must remain within a certain ratio between how many up and down votes they have cast, so that they'll eventually get blocked from up-voting until they start to cast down votes (and maybe vice versa).

• The rep system as well as the Q&A format in general is a terrible one for the meta sites. When these sites were designed, the devs were far too much in love with the Q&A format and thought it was the universal format for any kind of site. It is not. It is a system which encourages elitism, populism and discourages discussions. Instead everyone is supposed to be holding their own monologue and then people should silently vote on that monologue without explicitly arguing for/against. Don't you dare raise any unpopular arguments or counter arguments, or you may be down-voted into meta hell.

  And also the system kind of encourages having the same "beat the dead horse" discussions over and over, since it's far too hard to find older discussions, and even if you do find one and add an answer there, nothing will come of it since old posts don't get much attention, they aren't in the "feed".

  This is not a healthy system to use for meta at all. Good systems encourages diverse opinions and debates, bad systems put a lid on such things.

• What other systems work really well that we could learn from?

The badge system isn't ideal either, but it is the best working one in terms of measuring activity and technical knowledge both.

As for unlocking moderation tools and site privileges, it should probably be disconnected from the rep system entirely.

Also, while SE isn't social media, the Q&A format discourages users from hanging out with like-minded people with similar interests, or for discussing things in a less formal manner, without all the rules against subjective topics. There are the chats, but they are well-hidden and very inactive. A place to "hang out" in parallel with the Q&A (which is definitely not the place for such) would give people more reasons to stick around and also ease "on-boarding" of new users, if there was a place encouraging informal mentoring.

One example of such a "hang out place" could be blogs. There is the SO official blog, which rarely ever holds contains anything of interest to me, and on top of it the bloggers have various company associations and likely hidden agendas, making it even less appealing. I'd be much more interested in blogs posted by the site users, maybe attached to the tags. There used to be such a system in place even, though it was so well-hidden that I didn't even know of it until I read the message that it was closing down due to lack of interest.

The one thing I would absolutely change, is the rep cap. That should be per question instead of daily.

Take someone like Jon Skeet. He doesn't have to answer a single question anymore to still hit the rep cap every day for years to come.
I've met him once and he's a nice guy. I think he deserves the rep he earns. But I still think it's weird that one would earn rep for answers given long ago.

Or take my "home stack", SF&F. It has received an answer to a question about Groundhog Day by the screenwriter of the film. Great answer, don't get me wrong. I'm not linking it here, though. That would exacerbate the problem.
But he has earned enough rep from that single answer to cast close and reopen votes, without ever interacting with any other aspect of the site.

This would be remedied by having a rep cap per question. Jon Skeet wouldn't get "rep for jam", Danny Rubin would've gotten 315 rep (200 from votes + 15 for the accepted answer + 100 bounty) instead of 4000+.

In general, people would need to interact and keep interacting to gain rep and earn privileges.

Simple answers to common questions get lots of reputation.

If you are lucky or, old enough to answer a question that is easy to understand and commonly asked, you'll likely get lots of reputation. Even if your answer is simply correct rather than the best possible. This can be true if the question is a (near) duplicate.

To game this system, there is a race to get a valid answer up quickly and improve the answer while the votes accrue.

Difficult answers to well written but specific questions get almost no reputation.

If you read a question that others can't quickly understand and provide an expertly written answer that covers all the angles and makes world leading suggestions that answers the question perfectly, you'll probably get 25 points of reputation.

Once these questions are answered, others will rarely invest the effort to understand them and consequently, appreciate your answer.

Giving a better answer to an established question gives unreliable, slow rewards.

If an existing common question already has a few well "up-voted" answers, there is a disincentive to answer, you feel like it is too late to get reputation. If the world has moved on and there is now a better answer, is it worth adding it to the question?

If the answer is good, it will attract votes, slowly. So slowly that it is unlikely to usurp the outdated previous leader. That old answer got all its rep in flash when the question was new, but now, floats at the top misdirecting the impatient.

I see several long term users award bounties for good answers, nice in itself, but they award 500 rep bounties to new users, often on first answers, sometimes repeatedly.
This creates a user who has access to tools and actions they have not had time to get to know. And I have seen some that did act with that access in a way that I did not see as good. (Not bad enough to hand on to staff but it took a lot of comments, and even some e-mails to the user, to teach them how not to use those options.)

I would suggest that no bounties can be awarded to a user who has fewer rep than the bounty is. (Maybe put it in hold somewhere, to award when they reached a next level of reputation.)

This is a very isolated example and not a full-blown "what to improve" post. But it should be a guide when you improve things.

When you think about changes, please always make sure that the user receives constructive criticism.

Okay, a downvote is not constructive per se. But it is a pretty clear signal that this content as is is not wanted on this site. And as bad as it sounds, you could still "constructively" delete your own post, if it gets too many downvotes. Not great... not terrible. At least it stops the bleeding and improves the site as a whole.

However, once your question received an upvoted answer, you cannot. The system does not allow you to delete your own question.

So now there you are, with a question that is genuinely bad and maybe cannot be improved (lets say it was lacking research, you should not retroactively add the research that would invalidate the answer).

A downvote comes in. Nothing you can do. Another. Nothing you can do. Another. Nothing you can do. Just sit there and take the "punishment" with no way out. This is no longer about curating the site, since the obvious option to delete the bad content is banned by the system itself. It is only about revenge. Revenge for having done the bad thing in the first place.

Whatever you change, make sure that there always is a way forward for users, other than sitting there and "take it". There should be no traps, where users cannot react to negative feedback in a way to make it stop.

What is really great about our reputation & privilege system? What do you think makes it great, and why is it worth keeping?

It allows people who have authored very good answers or questions to edit other questions to improve them, close bad questions, etc. I think that stuff stuff should be kept.

What is broken about it, and why? Are there any solutions to it?

A lot. Just because somebody knows how to make good questions or answers doesn't mean they know what should be deleted. Make it so rep is not attached to privileges and instead actions. For example, Codidact (another Q/A site) gives privileges based on relevant actions. For example, editing freely once getting a lot of suggested edits approved. You can also take a look at roles the community would like.

What other systems work really well that we could learn from?

Like I said before, you can learn from Codidact and Discourse does with trust levels. You can lose TL3 (trust level 3) (which lets you edit categories and tags) if you are not active enough in the past 100 day by default.