Hacker News Reader

Comment by nicolas_t | Hacker News Reader

I always feel a bit sad and wistful when thinking about Stack Overflow. It was a great idea and at the beginning it worked well but the moderation culture destroyed the site.

There's too much emphasis on closing questions that are duplicates (when in fact they are not, and technology changes so a duplicate question asked 2 years later is going to get different responses), closing questions because they are too open ended (but those tend to give the most interesting answers) and moderation is both too heavy handed and opaque (no records of deleted comments).

Also, often enough the accepted answer for a question is not the best answer and instead the best answer can be found by looking at other answers or the comments (which I agree with the author should not be seen as ephemeral but are instead critical)

So, all in all, Stack overflow was great when it came out, the concept is good and it could be great but falls well short of that.

View on HN

I agree with all of that, but the worst part to me(which I guess relates to your point about the accepted answers) is that there's no path to deprecating old answer and old questions.

I've been lambasted here in the past for saying this, but I'll say it again now and see if I get a different response; it's no longer helpful to search for "javascript" related questions and get mostly answers for jQuery. These days I'm able to solve a lot of problems just reading MDN, but sometimes I click on a Stack Overflow article and it turns out to be 12 answers that are all jQuery-specific even when the question itself doesn't call for it. jQuery isn't bad, but the world has changed around it and we can't keep pretending like everyone is using it for the same reasons we did back in 2011.

This not only does new developers a general disservice, but it can do a disservice to projects that have less-than-average momentum. For instance, I use Ember JS every day, and it's a framework that has really changed a lot since its inception, but you'd never know it if you looked up questions for it on Stack Overflow. The Ember questions on there are so woefully outdated that I believe it gives potential new users the complete wrong impression of it. There are outdated React questions on there as well, but at least there's enough inertia behind React that new activity helps overcome this to a degree. On the other hand, people almost certainly have given up on Ember because none of the Stack Overflow answers work at all. Sure, we could all submit updated answers, but they'll be below the fold and they will never be reselected as the accepted answer. There are other reasons why Ember has been in decline, but I do blame the mechanics of SO as a big detractor.

Preventing duplicate questions is fine, the overzealousness not withstanding, but if there's no mechanism to correct existing information as time goes on in a meaningful way, the only result can be rot.

> it's no longer helpful to search for "javascript" related questions and get mostly answers for jQuery

jQuery is used on 83% of all websites; React is used on 4% [1]. You’re right that people are probably using it differently than they were a decade ago, but even in 2021 your average jQuery question is probably useful to many more people than your average React one.

[1] https://css-tricks.com/how-the-web-is-really-built/

> jQuery is used on 83% of all websites;

There are a lot of websites I made in the past and they are still online and many of them use jQuery, but I don't need to check Stackoverflow in relation to those websites, because I'm no longer working on them (and in most cases no one else does either, they just happily keep running unmodified). All of those are part of that 83% that you mention.

Stackoverflow is a Q&A websites for developers so IMHO it should focus on tech that is currently in use on actively developed systems.

Even if only 10% of jQuery websites are maintained, and every single React website is, there are still almost twice as many actively developed websites using jQuery than React.

That's a fair point, I didn't think about it that way.

See my comment, and actually with that 10% number it doesn't hold true with only 50% of the logic applied.

50/10 vs 4*(unmaintained rate) + undercounting (possibly a large majority, possibly small) is probably React winning out

I still use jquery actively. Without substantial work and complexity, a typical react site's initial page load still requires an enormous javascript chunk for anything to show on the screen. When I notice a site I'm using is slow, unresponsive (interactivity wise), or buggy, more often than not the react dev extension is lit up. React (and ecosystem) may represent the future but they still have a lot of growing pains to overcome before totally replacing jquery.

I also use jQuery from time to time and I don't have anything against it, my point was not related to jQuery.

PS from your comment it sounds like you could be interested in Vue, specifically using Vue standalone without npm/webpack e.g. https://v3.vuejs.org/guide/installation.html#cdn

This doesn't make any sense. jQuery is a library for DOM manipulation. React is a view layer framework. They are totally different solutions for different problems. If jQuery suits your needs, you probably don't need jQuery at all because vanilla JS is quite capable these days.

They are both used to add client-side interactivity. I agree vanilla JS is often a capable alternative. jQuery offers a lot of convenience though.

This seems unreliable since it weights a website maintained by a single developer in their spare time the same as a website with a dedicated team of devs working full time.

Yeah, Stack Overflow is for humans and not for websites. A billion websites maintained by 1 person don't count more than 1 website maintained by 1 person. The question is how many web developers nowadays use these technologies during their day to day activities.

A somewhat accurate test would be to look at job listings. Last time I checked. Every front end job listing asked for react/vue/angular skills.

That 83% is very shaky at best.

First, number of websites != number of platforms. If you go to the source of the number, they even point out as much:

> WordPress, which is used in more than 30% of sites, includes jQuery by default.

So right off the bat, you're looking at 50% or so, maybe plus a few for people who actually make a point to write jQuery + for wordpress specifically.

Secondly, that study has detection issues [0] that seem to not pick up a good deal of bundlers/etc, which makes it really hard to say how many of React/Vue sites are missed.

Thirdly, jQuery being used as a dependency of a library will still count it in that 83%. They show a few combination options specifically, and show it's at least 1.7% [1]. With others, say 2.5% in total. Think about how many bundlers hide React/Vue and how many common dependencies could use jQuery, and you're probably much higher.

Given all that, I'd bet that number is closer to 17% than 83%, though I'd wager around 33%.

Even with that 33% being jQuery without a framework or not being wordpress, you have to remember this is published. How many of those have had code significantly changed in the past 5 years? This is what's relevant in this context. All of this long winded nuance to say that I don't think "your average jQuery question is probably useful to many more people than your average React one" holds up, at least not like you might think that cited number shows. That's not to say it's irrelevant, just less relevant.

[0] https://github.com/AliasIO/wappalyzer/issues/2450

[1] https://almanac.httparchive.org/en/2020/javascript#framework...

Agreed that it's all very directional, but I think you're discounting some of it a little too readily. Even if you're just using jQuery because it's a transitive dependency or bundled by default, you might still have issues that lead you to search Stack Overflow for help.

The broader point is, we should be cognizant of our selection bias here. SPA frameworks are overrepresented on HN because we tend to be programmers building applications in the tech industry. But there are a ton of people who don't consider themselves "in tech" that still need to make a portfolio for their photos/figure out a WordPress plugin for their e-commerce site/etc.

That's probably more of a measurement of popular frameworks or libraries pulling in jquery, rather than people wanting to use it for development. Your link suggests a third of the web is wordpress, for example, but wordpress seems to want to eventually remove it as a dependency.

That 4% constitutes many large shops.

You're missing the point I believe the OP was trying to make: ten+ years ago, jQuery was often by far the easiest way to implement most things. One line of jQuery could replace a dozen lines of plain JavaScript, even ignoring all the nonsense you'd have to learn about cross-browser support.

Today, so many things are possible and easier with plain JavaScript and/or CSS.

I'd even go further and say there's lots of new sites using jQuery unnecessarily because the accepted, highly-upvoted, but 10-year-old SO answer said that was the way.

To summarize: jquery used to be best answer to many JS questions. And now it never is unless the question specifies its use.

No one should be recommending the introduction of jquery (to solve some discrete issue) unless it’s already part of the stack.

But that’s exactly the opposite of how the current state of JS answers appears.

Stack Overflow's developer survey found that in 2020 43% of web developers were using jQuery and 36% were using React [0]. Still biased in favor of jQuery but not nearly as much as you say.

[0] https://insights.stackoverflow.com/survey/2020#most-popular-...

Based on nothing other than intuition, I assume the Stack Overflow audience still at least slightly selects for React developers over jQuery developers, compared to the general public.

Of course, that doesn't really matter — we're talking about Stack Overflow users, and you're right that this is a way better source of data on what Stack Overflow users care about. Not sure why I didn't think to check their developer survey.

Even React answers of the past are often class component based... and for my taste the react answers are either slapshot / they work but aren't best practices, or if they are good they rarely explain why you would do A instead of B.

> the moderation culture destroyed the site

There's a term in Russian "синдром вахтёра" which I think should be translated as "a doorkeeper/janitor syndrome". It's a phenomena when a not very smart person who doesn't understand the big picture of things gains a little bit of power against his fellow mates. While generally not intending to bring a harm, they begin to implement some weird but strict policies that reflects their very poor understanding of how things should work (metaphorically they think that every line should either be parallel or perpendicular to each other and no deviations should ever exist). Given that at least some of them might just enjoy having and exercising that little bit of power they have, the result could be just ugly and destructive.

Unfortunately, in many cases I've ever seen over the Internet communities (that's not to stretch the idea further), those who want to become a moderator are very prone to this syndrome.

I think to create a healthy community some extra special measures should be implemented to prevent this phenomena. Unfortunately StackOverflow is not doing it enough (if doing anything at all).

I think the rules about self-promotion are well-intended. If you ever get quora results to common programming or technology searches, you'll see what I mean; people responding with only loosely related comments which link to their own product, which "solves the problem for you". I get it, without any moderation, SO would be just as useless.

I think there should be some nuance in how those rules are applied. The author of this post appears to be one of the authorities on webgl. If they've written helpful posts on the topic, and those can be referenced from SO answers, they should be able to do that. Heck, even if they sell something on their blog, it's still highly relevant. It's same the problem Google, Youtube, Apple, and basically every other service right now is having with attempting to moderate what can be shared on their platforms, without losing the quality content. Incredibly often, they end up throwing the baby out with the bathwater.

I don't have a solution; the problem is in promoting quality content while preventing spam. It's not all-or-nothing, we need to strike the right balance with moderation.

I wonder if something more like a wiki format would be better long term. Say you take a question like "How to loop through a list of key,value pairs in JavaScript. Through the last 10 years, this has changed a lot through vanilla to libraries like jquery to new iterations like ES5/6. What I would love to see is just a page with this question asked, maintained over time with an emphasis on when the previous solution was relevant, and highlighting the most recent, preferred options for solving the problem.

So, rather than a score based social network style system, turn the questions into living, and evolving documents. Also, sometimes a person gets stuck having to work on code that's a decade old, so it would be nice to have a history as well, and not just the most recent solution that may not work in their environment.

This doesn't sound bad on the surface, but my worry would be that we would wind up with Wiki-style moderation instead of SO style moderation. i.e. 'guarded articles', edit wars and power-mods overriding outside consensus.

Instead of SO style moderation, with closed questions being treated as canonical, corrected information buried in comments, and power-mods overriding outside consensus?

A wiki curated from stack overflow answers would be amazing!

StackOverflow powered GPT-3 haha.

Have you done much editing on Wikipedia? That seems like an optimistic view of what the experience would be.

Wikipedia isn't perfect, but they get one thing right that Stack Overflow gets horribly wrong: they make it obvious who the audience is.

A lot of people visiting Stack Overflow think that the purpose of a Stack Overflow question is to solve the OP's problem. They write "Dear Libby" style letter templates into the question box, and often provide very opinionated answers that try to give way more advice than was strictly asked for by trying to infer what's "really going on" based on the subtext of the question. And there's a lot of UI in there that encourages seeing it as an advice column.

But the Stack Overflow people themselves don't want it to be a place to seek general advice. They want it to form a general resource, with Stack Overflow being less of a bulletin board and more of a reference guide. The problem is that their UI doesn't telegraph this very well (I blame the way they leave the original asker's name and avatar visible on the question), and neither do their marketing materials. They're kinda trying to have it both ways, baiting people with the chance to get their own question getting answered and then switching to editing their question into something more generally useful than something just for them.

It is not a surprise that people feel cheated.

The irony is that that is actually what StackOverflow is: you can edit any question to make it clearer... and you end up leaving the original person's name on it (which I frankly find kind of evil).

The irony of all of this is that stackoverflow works just fine 95% of the time.

This is my take, too. I really like answering questions; it's nice to know that the thing I'm writing will be immediately useful to somebody, and maybe a bunch of somebodies. I'm a little uneasy doing free work for for-profit companies, but I set that aside for both StackOverflow (~3k points) and Quora (awarded Top Writer two years running).

In both cases, it was bad moderation that pushed me into quitting. A trick I learned from working on Wikipedia is that you can think of these systems as MMORPGs where you get a useful product out the end. I think all of these platforms have game-design problems, where people with power can find satisfaction in moves that end up harming the reader experience and frustrating contributors.

Both companies seem to be doing fine without my help, so I guess they can afford to burn contributors out after a while. But like you, I think it's sad how much these things fall short of their potential. I'll never do anything for SO or Quora again, but I do miss the chance to help people.

> I really like answering questions; it's nice to know that the thing I'm writing will be immediately useful to somebody, and maybe a bunch of somebodies

Totally agree. Explaining something to somebody else is a great way to learn, too.

That was my experience on early SO, and it was great. I'd just browse a tag I knew, and pick questions where I knew most of the answer, and the act of explaining it would be useful to me, too.

Now I find doing so neither fun nor fulfilling. First there's lots of garbage questions that are impossible to answer. When you do finally find something good, it's hard to get it before it gets closed as dupe or someone else answers. And if it's from a new user, there's still a good chance they'll never come back to accept the answer and awards your fake internet points.

I'd love to see a SO that encourages more new users to actually answer those "dupe" questions (instead: hide them from search and older users)

>"Explaining something to somebody else is a great way to learn, too."

Indeed - I've become an expert at several systems by hanging out in support forums and answering other peoples questions. the harder, the better! Who needs video games when you can still solve problems, help others and increase your value/skill set too?

> There's too much emphasis on closing questions that are duplicates (when in fact they are not)

This is a great point. And it also relates to abuse of power.

Recently I raised a question without precedent on Stack Overflow. A person had given a detailed answer, which pointed to issues not available in documentation. His answer was in fact an expanded analysis of the overall language specs - without it, I had no idea how to debug my code.

One megalomaniacal moderator marked my question as 'duplicate', and then pointed to another question -- where he claimed to have answered the question 'definitively' (he had not) and wanted me to upvote his answer.

There was no way for me to challenge his 'duplicate' mark, which is a real loss for anyone else encountering the same issue as I did.

I stopped because... and I wrote it down...

"This question is not a good fit to our Q&A format. We expect answers to generally involve facts, references, or specific expertise; this question will likely solicit opinion, debate, arguments, polling, or extended discussion"

I think there should be a place for this kind of stuff, some place hackers can do these "prohibited" sorts of things...

> “... or extended discussion“

Oh my, not the dreaded extended discussion!! When will people learn that our tech overlords have found the most optimized means of communicating? We mustn’t revert to the dark times when conversing was openly practiced by the barbarians.

Comments are not for extended discussion; this conversation has been moved to chat.

But does this place have to be StackOverflow? I think it has space for discussion in comments, chat rooms, etc., but the main artifact is intended to be an easily searchable Q&A archive.

I think they need to work on communicating that much more clearly, but ultimately they don't want to (primarily) be a place of extended discussion, debate, and so on, and I'm not sure it's fair to say "no, they're wrong" to that.

> But does this place have to be StackOverflow?

I was kind of referring to the very nice platform I'm typing on right now... :)

Disagree, stack overflow is still the best place to get answers to common programming questions. If you need to sort a list in Python it’s much better to find the answer on a short stackoverflow post than to sift through dozens of blogs, each with their own style, or to try to parse pages of documentation that goes into way more detail than you need. No other site has as much content as easy to find as stackoverflow, even if it’s not perfect.

It is much better to read the official documentation.

People want quick answers, not browsing documentation. SO is useful in the sense that it provides answers to specific questions which a google search finds instantly.

True, but I've seen lots of answers that are "hacks" when there's usually a better, supported way that was spelled out in the docs, e.g. won't fail when you upgrade the library say.

Or they don't want to understand the issue at hand, so they put in a hack, when obtaining a complete understanding of the issue which might take 5 minutes of their time reading the docs, would provide a better solution.

If the docs have examples that cover your use case, which is far from certain.

Exactly. I code in half a dozen different languages that all have their own ways to split a string. When I can't remember the syntax to do this in, say, Python, being able to just say "python split string" and it goes to a reasonable SO answer is a big win for me.

I love reading docs too, so I can get a deeper understanding. But sometimes I just need an answer in a hurry.

Yes, people are lazy, most do not even learn from SO answers. They copy paste, move on, no greater understanding achieved.

I have never enjoyed python’s official documentation. It isn’t easy to find answers. Sometimes the value of an SO answer is that you are sent to exactly the right part of the python docs.

> There's too much emphasis on closing questions that are duplicates (when in fact they are not, and technology changes so a duplicate question asked 2 years later is going to get different responses)

I've always been a little surprised that that they don't have a formal system of 'recycling' popular questions where they have a new 'edition' every 1/2/3/whatever years and they just prune the ones that don't get new answers.

It's valuable to get the latest answer to questions, but lots of us operate in past technological climates, so it's good to keep old answers accessible.

It’s slowly dying, everyone knows that in technology standing still means you’re moving backwards and StackOverflow trends towards standing still.

Once every question has been asked and answered every new question is a duplicate and no new information can appear in StackOverflow.

I am seeing more and more often that you look for a solution to an issue in StackOverflow and all you get is fixes that only work on obsolete software. For instance for operating systems a lot is for Windows 7 and before and doesn’t work on Windows 10 anymore.

Note you can still post answers to older questions; this is actually encouraged!

Closing of duplicates is the least of Stack Overflow's problems IMHO; it's done for a very real reason: if I search "how to do foo in bar?" then it's really great to have one question with all answers, instead of having to trawl through several questions. Also, as someone who spent quite a lot of time writing in-depth answers over the years, I'd rather not repeat that every month.

I have a lot of criticism of Stack Overflow, and I mostly stopped answering because I got fed up as well, but this isn't a problem IMO.

> Also, often enough the accepted answer for a question is not the best answer and instead the best answer can be found by looking at other answers or the comments (which I agree with the author should not be seen as ephemeral but are instead critical)

Generally those comments should be integrated in the answer; usually if I come upon an answer with helpful comments I will edit the answer to include that information.

I do wish the entire "accepted answer" feature would be removed though, or at the very least stop pinning them to the top.

Shouldn’t a proper sorting (based on activity, karma, number of answers) take care of this (= duplicate questions)?

Not necessarily; and Stack Overflow doesn't control Google's sorting. It would also mean people would have to keep answering the same questions over and over again, which would be a waste of everyone's time.

> Note you can still post answers to older questions; this is actually encouraged!

Although, this can lead to its own weird problems.

I've been learning Objective C, and I've come across a lot of Stack Overflow questions which ask how to do a thing in Objective C, but where all the top answers are in Swift. Sometimes an older Objective C answer is just buried; sometimes it isn't there at all.

I realize that Swift is the newer language, but it's not the one I'm using, and the question was about Objective C.

Objective-C/Swift is a bit of a special case; but asking "how to do X in Objective C" is probably fine, even if there is already a Swift question. And a lot of questions tend to be about libraries/frameworks anyway, and the Swift answer should be usable in Objective C too.

> Objective-C/Swift is a bit of a special case; but asking "how to do X in Objective C" is probably fine, even if there is already a Swift question.

To be clear, it's the other way around—someone asked a question about Objective C, and people answered how to do it in Swift.

From the perspective of "it's good to post updated answers to older questions" this makes sense, because there's a decent chance the asker would have requested Swift if the question was asked today. But as it stands, the question lists Objective C.

Which is to say, a lot of questions are "products of their time", and posting modern answers to older questions sometimes requires assumptions about what the asker would have wanted. I'm not sure what the best solution is.

> someone asked a question about Objective C, and people answered how to do it in Swift.

Well, then this is a wrong answer haha.

> Which is to say, a lot of questions are "products of their time", and posting modern answers to older questions sometimes requires assumptions about what the asker would have wanted. I'm not sure what the best solution is.

I think it probably depends on the exact question and details; there isn't really a good one-size-fits-all solution.

A question is considered a duplicate only when it's an exact duplicate. Sometimes question get closed as duplicates erroneously as they're not exact duplicates, just very similar, and this can be an issue, but I don't think that's the biggest problem on Stack Overflow, and a lot of questions really are exact duplicates.

A lot of these issues are kind of scaling issues: with about 7,000 questions every day, many of whom are not exactly stellar, it's pretty hard to deal with them all. For year I've argued that improvements in messaging and such would make a big improvement, and Stack Overflow is finally doing some work on that now. For example the comment was changed from "Possible duplicate of [other question]" to "Does this answer your question: [other question]". The latter is much more friendly and helpful. People don't mark questions as dupes to be mean, but just as the easiest way to guide you to an answer (hopefully...)

I bet this is a result of the dichotomy of purpose that this comment by notriddle is describing[1]:

> They want it to form a general resource, with Stack Overflow being less of a bulletin board and more of a reference guide. The problem is that their UI doesn't telegraph this very well (I blame the way they leave the original asker's name and avatar visible on the question), and neither do their marketing materials. They're kinda trying to have it both ways, baiting people with the chance to get their own question getting answered and then switching to editing their question into something more generally useful than something just for them.

Easy to look at an old question and think "There's no point posting another answer here, the asker is long gone." If we thought about it more like a wiki page and less like Yahoo Answers that wouldn't be the case.

[1] https://news.ycombinator.com/item?id=26163113

Yeah, the UI/communication could improve here. Stack Overflow suffers from a lot of mismatched expectations.

Some of my most successful answers have been to old questions by the way: I try to solve a problem, the Stack Overflow question doesn't contain a solution I like, I figure it out, and I post it as an answer.

I often have questions closed down and then, later get a badge from stack overflow for having a popular question.

Content needs to be clustered by technology. Python 2 answers are different from Python 3. One way would be to archive older definitively answered questions and allow new questions to be flagged for the newer release, such as Python 3. Wherever there were breaking changes, answers become invalid for the newer technology. Flagging every question by year would be the easy way. Any question prior to ES6 release would be legacy JavaScript, for example.

same goes for all question-answer type platforms. they all fall into the traps you describe above. unfortunately, i don't see any of them doing anything about these issues.

I scroll directly past the accepted answer on many SO questions as I often find that the second or third highest ranked answer is often actually the best answer.

Reply on news.ycombinator.com