Hacker News Reader

Comment by foxglacier | Hacker News Reader

I always thought StackOverflow was meant to fizzle out over time as more questions get answered and don't need to be asked again. Perhaps the decline is just a necessary part of their rule of having no duplicate questions - keeping it as a clean repository of knowledge rather than a messy forum.

Just the other day a question I asked about 10 years ago got flagged as a duplicate. It turns out somebody else had asked the same question several years later and got a better answer than my question got, so that other one is the canonical one and mine is pushed away. It feels kind of offensive but it makes complete sense if the goal is to provide useful answers to people searching.

View on HN

Unfortunately, the rule of no duplicate questions also destroyed lots of questions that weren't duplicates... because _someone_ couldn't be bothered to read them and realize it wasn't the same.

Plus, there were a lot of fun questions they were really interesting to start with; and they stopped allowing them.

Yes, this. I've asked a couple of questions where the only responses are from people saying "possible dupe of x" where x is something that has a couple of the same words but no relation to what I'm asking.

Turns out if you design a forum where a high effort, high quality post can be devalued by a low effort response, you discourage high effort, high quality posters.

This is where we need SO's "The Answer" feature, whatever they called it. Never seen it distilled so well.

The major trouble with StackOverflow is that nominally duplicate questions may have different answers if asked on 2011 vs 2026 - and the trouble is that answer rankings (the thing that determines what answers are in the top) don't decay over time. So if someone try to answer an old question with up to date info, they won't garner enough upvotes to overcome the old, previously correct but now outdated accepted answer at the top. (even with a ranking decay, there is little incentive to give a new up to date answer to a established thread - people are more likely to contribute to brand new threads)

It would be better to allow duplicates in this specific case, but mark the old thread as outdated and link the questions in such a way that one can see the old thread and compare it to the new thread.

This is something I saw all the time. I’d look something up, knowing that there was probably an easy way to do <basic programming task> in modern c++ with one function call.

Find the stack overflow thread, answer from 10+ years ago. Not modern C++. New questions on the topic closed as duplicate. Occasionally the correct answer would be further down, not yet upvoted.

“Best practice” changes over time. I frequently saw wrong answers with install instructions that were outdated, commands that don’t function on newer OS version, etc etc.

You raise an interesting point about decay. I have thought about similar systems myself. One flaw in a simple decay rule would be that some technologies are very stable, e.g., C & POSIX API programming. However, other tech is very fast moving like Python, Ruby, Java, C#, C++, Rust, etc. One idea to overcome this flaw, might be to have moderators (who are specialists on the subject matter) provide a per question decay rule. Example: Something like struct layout or pointer manipulation in C or the fopen() POSIX function might never decay. But something like a parsing JSON in any fast moving language might require annual updates. For example, a question about parsing JSON in Java might decay answers over a one year period to encourage people to revist the topic. I would like to hear Jeff Atwood and Joel Spolsky debate this topic with other "Internet points" experts for an hour-long podcast. They might brainstorm some very intersting ideas. I would also love to hear what they think about the "moderator problem". Some of the topics had incredibly toxic moderators who scared away newcomers and women. (Women are much less likely to participate in public software forums where public shaming is common.)

> One idea to overcome this flaw, might be to have moderators (...)

> I would also love to hear what they think about the "moderator problem". Some of the topics had incredibly toxic moderators (...)

Yeah having bad moderators and arguably a bad, dysfunctional community is perhaps a even worse handicap. If you go to threads on meta.SE (meta stack exchange, meta discussions on the whole ecosystem) you will see that people mostly believe the site policies are okay, and that's because everyone that didn't believe left years ago.

Maybe better ideas on how to evolve a Q&A site may evolve in a brand new site, unfortunately I think that SO and perhaps the wider Stack Exchange network is done.

Great point, because as the knowledge evolves might need to evolve ranking too by allowing some versioning and somehow ranking or marking the outdated ones

That's what the Bounty system was meant to handle. It could have been done better but it's not like they never considered it.

The problem with this, and why SO’s downfall was completely self-inflicted, is that the correct answer from 2013 is only occasionally still the correct answer in 2018. There are a lot of other issues with SO’s general moderation policy but well and truly it was as idiotic and myopic as it was toxic.

They treated subjective questions about programming methods as if they were universal constants. It was completely antithetical to the actual pursuit of applied knowledge, or collecting and discussing best practices and patterns of software design. And it was painfully obvious for years this was as a huge problem, well before LLMs.

That said, I will say after being traumatized by having my threads repeatedly closed, I got so good at boiling down my problem to minimal reproducible examples that I almost never needed to actually post, because I’d solve it myself along the way.

So I guess it was great for training me to be a good engineer in the abstract sense. but absolutely shit at fostering any community or knowledge base.

> that the correct answer from 2013 is only occasionally still the correct answer in 2018

Exactly! They should have added proper structuring to questions/replies so that it could specifically apply for Language/library version X. Later, such a question could be answered again (either by proving it's still correct for version X+1, or by giving a new answer) - that way people wouldn't have to look at a new reply with 2 votes vs an older, possibly outdated one with 100 and make a decision which to prefer.

Reply on news.ycombinator.com