News: 0177899101

  ARM Give a man a fire and he's warm for a day, but set fire to him and he's warm for the rest of his life (Terry Pratchett, Jingo)

How Stack Overflow's Reputation System Led To Its Own Downfall (infoworld.com)

(Monday June 02, 2025 @11:36AM (msmash) from the self-inflicted-wounds dept.)


A new analysis argues that Stack Overflow's decline began years before AI tools delivered the "final blow" to the once-dominant programming forum. The site's monthly questions dropped from a peak of 200,000 to a steep collapse that began in earnest after ChatGPT's 2023 launch, but [1]usage had been declining since 2014 , according to data cited in the InfoWorld analysis.

The platform's remarkable reputation system initially elevated it above competitors by allowing users to earn points and badges for helpful contributions, but that same system eventually became its downfall, the piece argues. As Stack Overflow evolved into a self-governing platform where high-reputation users gained moderation powers, the community transformed from a welcoming space for developer interaction into what the author compares to a "Stanford Prison Experiment" where moderators systematically culled interactions they deemed irrelevant.



[1] https://www.infoworld.com/article/3993482/ai-didnt-kill-stack-overflow.html



Apt comparison (Score:4, Insightful)

by RobinH ( 124750 )

As someone who experienced this decline, I think the comparison to the Stanford Prison Experiment is very apt. It started out amazing and then became a private club for cruel people to claim their turf. But Stackoverflow is complicit in this. They actually encouraged it, not through the reputation system, but by specifically handing them the baseball bats they used to beat the newcomers with. Moderators were using any excuse to close new questions or shit on new users, even when just letting someone answer the question would have been both helpful and easy.

Re:Apt comparison (Score:4, Insightful)

by nevermindme ( 912672 )

The mods were attempting badly to state the question / answer is in the search results, somehow they stopped saying that before deploying the bats. The learning model gives the example first, then answers the questions second, the way if Stack Overflow had ever closed a question with an acceptable answer tag would have with human moderation. Unlike slashdot, some threads never stopped changing the answer. The answer in 201x with version x of python is a perfectly good ending to a question, unless there is a discovery in calculation of length of a string.

Re:Apt comparison (Score:4, Insightful)

by cfalcon ( 779563 )

I think part of the problem is that external search engines would simply put worse interactions to the top based on some metric known only to them, so there was a desire to eliminate things that were superfluous, to channel searchers into the places that have their answers.

But of course it ran into the exact problem you described- generally a class of moderators wants that position for some reason. You're looking for the moderators who share a vision of a really useful place where everything works great, but many moderators will just be there to enforce some value (sometimes political) or because they have a keyboard sadism streak and that's that. Basically when you take volunteers for "who wants to have power", only some of the people coming forward want to use that power for thing you want, the others will do that to fulfill the job role but they'll REALLY use it for $LAME_THING.

Re: (Score:2)

by phantomfive ( 622387 )

I don't understand why StackOverflow was so obsessed with people not asking the "wrong" kind of question. Storage is cheap, it doesn't hurt to have the same question asked more than once.

Re: Apt comparison (Score:2)

by topham ( 32406 )

On another forum a user recently complained because someone had edited their question on Stack Overflow. The final question no longer represented what they were asking about and expressed distain for the edit; someone else on the forums who has enough editing powers in Stack Overflow to do such then justified it. Or tried to.

Editing a users question as if it was the original question they asked is NOT ok.

Refreshing the question and answering that version of it may be useful, but sometimes obliterates the su

Re: (Score:2)

by Racemaniac ( 1099281 )

Some of the issues with asking the same questions over and over are:

- More interesting topics getting drowned out by the same beginner questions repeating over and over

- The questions same questions not always getting the same answers. In the end the experts won't revisit the same question for the 50th time, and beginners will start answering beginners.

- No knowledge aggregation doesn't really happen. Rather than the topic of that question getting dug up from time to time when someone asks for more clarific

Re: Apt comparison (Score:3)

by shm ( 235766 )

Have never understood this behaviour. One can see it also on Wikipedia and Reddit.

Translation: (Score:5, Insightful)

by Sebby ( 238625 )

> the community transformed from a welcoming space for developer interaction into what the author compares to a "Stanford Prison Experiment" where moderators systematically culled interactions they deemed irrelevant

Translation: they started power trippin'. Enshittification ensued.

I think I said something like this last week (Score:2)

by laughingskeptic ( 1004414 )

... [1]https://slashdot.org/comments.... [slashdot.org]

[1] https://slashdot.org/comments.pl?sid=23705427&cid=65418703

SucksOverflow (Score:2)

by devslash0 ( 4203435 )

Most of the time what's on StackOverflow is completely useless to the problem you yourself are trying to solve.

Firstly, because the problem that is described in the post is very specific to the author's context.

On top of that, 99% of responses to the original post are irrelevant. People clutching at the straws when having no clue what the root cause is themselves. No intention to find out what the root cause is either. Just throw a random thought at the post and hope for the best.

If you're lucky, you someti

Re: (Score:2)

by twms2h ( 473383 )

Actually no. I have found many very useful answers on StackOverflow. If you don't, maybe its you?

That's not to say that the growing enshitification hasn't been obvious for years, basically since the original founders sold it.

never (Score:2)

by Tom ( 822 )

> self-governing platform where high-reputation users gained moderation powers

Yeah. Never, ever, do that. I've run a few online communities. Back when your own forum was still a thing and you could survive without being a group on Facebook, a subreddit or a Stackoverflow.

Your most active users aren't always your best users, and they almost always are NOT the ones you want to have as moderators.

If I could do all that again, I would give mod rights to the people who contribute just a bit, but consistently over a long time, and who read more than they write.

Re: never (Score:2)

by devslash0 ( 4203435 )

Correct. If you want moderators - employ them; make them work for you.

But that's what you get if you don't want to pay, if you just exploit people.

Same thing happening with Wikipedia (Score:4, Insightful)

by xack ( 5304745 )

A lot of Wikipedia is being progressively hidden behind "blue locks" or "extended confirmed" restrictions, meaning only editors with 500 edits can edit them. In practice this locks out casual editors and only people with entrenched interests, mostly ulterior, are incentivised to edit them. This will ultimately be the downfall of Wikipedia as many of the top admins are retiring because they didn't recruit fresh talent. Wikipedia is down to 839 administrators from their peak of around 2000 in 2007. Same effects will happen to companies that replace entry level workers without degrees or experience with AI.

Like money, political power, etc. (Score:2)

by medusa-v2 ( 3669719 )

> As Stack Overflow evolved into a self-governing platform where high-reputation users gained moderation powers, the community transformed from a welcoming space for developer interaction into what the author compares to a "Stanford Prison Experiment" where moderators systematically culled interactions they deemed irrelevant.

Start on an even keel, with a community that rewards relevance (new, competitive economy), some people gain status (wealth) and use that to reshape the environment to their liking, until (billionaires) they get to the point where the can effectively run the whole show in such a way that punishes people for having needs rather than helping them succeed (medical debt, college costs, low wage jobs, etc).

You don't have to feel the same way I do, but I stand by the claim that the analogy is 1:1 on the details I

RTFM (Score:2)

by dark.nebulae ( 3950923 )

RTFM has always been a problem in our industry. I can remember in Usenet days asking questions and getting responses like RTFM or "I found your answer over here on [1]http://goatse.cx/ [goatse.cx] ..."

SO works until everyone gets tired of answering the same questions over and over because the posters don't want to RTFM, they just want a quick answer.

And so AI is killing them now because AI doesn't care how many times someone asks a previously asked/answered question, it will happily reply to every question as best it can

[1] http://goatse.cx/

Re: (Score:2)

by drillbug ( 126567 )

And where exactly *is* the manual these days? Most of what I see has been mechanistic doxygen style nonsense that isn't crafted to explain why and how, only to restate basic facts about what.

Most of the questions that I find myself ending up at SO with are regarding how to achieve a particular effect, then once a human being actually explains the key points, I can find the rest. Google has become a crutch that has let software developers ignore good docs for the past twenty years, and AI summaries are

My 2 cents (Score:2)

by fluffernutter ( 1411889 )

I tried to ask questions on SO a couple times but I never really had the time to "work on my score" to answer questions and I'm not sure if I consider myself enough of an expert in anything to provide the detail some of those people provided in their answers. They were totally ignored and I never knew if I had done something wrong or if just no one knew the answer. When I have found the answers on SO they are usually pretty good. A lot of the answers were even updated every few years over many years as t

Re:My 2 cents (Score:5, Informative)

by CubicleZombie ( 2590497 )

I created an account, posted one correct answer to a relatively simple question, and was downvoted. So I logged out. That was many years ago.

I still find it useful when it comes up in a search.

And it's better than Reddit. I don't even bother clicking reddit links that come up in a search.

The Beatles:
Paul McCartney's old back-up band.