My suggestion: works on a frecency-based sort algorithm to display good recent answer above outdated one. StackOverflow is still good, just that you have to scroll a bit nowadays.
I think it's actually a really funny and naive assumption that had led them to having an "accepted answer" at all. For a q&a platform that focuses so much on avoiding duplicates, thinking that the first person to ask a particular type of question is an authority not just on accepting an answer at that point in time, but to have it be accepted forever - it just doesn't make any sense.
Some topics are definitely more susceptible to rot than others. Questions like “how do I do X on Ubuntu” tend to have a lot of outdated (yet accepted) answers from a decade ago. There have been a lot of Ubuntu releases since then, with a lot of cumulative changes, and tagging questions as release-specific isn’t universal or reliable.
If it solves the question, it is exactly the right answer, regardless how much time passed.
The problem is that people mark answers too fast as duplicates, so as time passes and tools develop, questions get closed as duplicates even when the question is slightly different or the answer won't work any longer.
> If it solves the question, it is exactly the right answer, regardless how much time passed.
But that's the crux of the matter - what exactly is this "question" entity?
I'm reminded of Heraclitus's "You cannot step into the same river twice" - if a question is an utterance made by an individual at a given point in time, then it might have "exactly the right answer" that they'd accept, but then it's extremely sensitive to those conditions, and even an identically worded question asked by the same individual a week afterwards might merit them to choose a different accepted answer, because the context for their question changed. For example, something like "What’s the proper way for me to deliver a signal to a different thread?" can have a myriad different contexts, and even if you give a wall of additional information and code, there probably would be something important omitted.
On the other side of this spectrum, we treat questions as a pointer to an underlying platonic idea of the question, which is exactly what StackOverflow say in their guidelines [0]:
> There are many ways to ask the same question, and a user might not be able to find the answer if they're asking it a different way.
Indeed, the closed-as-duplicate label uses the following text make it clear that they approach a question as an idea independent of the context in which it was asked:
> This question already has an answer
So if a question is independent of its questioner and can lead to other questioners being told that their utterance is a duplicate even a decade afterwards, then why should the original have any extra rights in deciding which one answer is the right one? Shouldn't this be owned by the community? At the very least, if a questioner writes a good question that is marked as a duplicate, then they should be given the same access to decide on the accepted answer to the merged question.
In some ways Stack Overflow feels dated, but in some ways I hope it will go on; I still often prefer responses found there from the AI's. It also brings the recurring theme of what the LLMs will be trained on when people create less and less content. But I guess some people will always, and it might be enough
I think if we started trying to track the version numbers of questions and answers, and attached them to existing answers and questions, things would sort themselves out rather quickly.
The real problem is asking a question about how to do something in Java 8, becoming the canonical answer, and now we're on Java 25 and your answer is still on top even though we've gone through two new strategies for idiomatically representing that slice of code in the API and there's a 3rd in beta testing.
I wouldn't say that, but it is a pretty annoying community, and one that I'm happy to leave behind, in favor of LLMs.
I think you may be right about the "doubling down." The Meta discussions seemed to get a lot nastier, as time went on. Might have something to do with centrists being driven out by zealots. Happens all the time, especially in communities in crisis.
My suggestion: works on a frecency-based sort algorithm to display good recent answer above outdated one. StackOverflow is still good, just that you have to scroll a bit nowadays.
I think it's actually a really funny and naive assumption that had led them to having an "accepted answer" at all. For a q&a platform that focuses so much on avoiding duplicates, thinking that the first person to ask a particular type of question is an authority not just on accepting an answer at that point in time, but to have it be accepted forever - it just doesn't make any sense.
Some topics are definitely more susceptible to rot than others. Questions like “how do I do X on Ubuntu” tend to have a lot of outdated (yet accepted) answers from a decade ago. There have been a lot of Ubuntu releases since then, with a lot of cumulative changes, and tagging questions as release-specific isn’t universal or reliable.
Never mind that "how do I do X on Ubuntu" was never a programming question in the first place.
Stackexchange != Stackoverflow
Maybe GP meant ServerFault
You can change the accepted answer after the fact.
I've done that. I like to avoid it, if I can, because the original accepted submitter gets a demerit.
If it solves the question, it is exactly the right answer, regardless how much time passed.
The problem is that people mark answers too fast as duplicates, so as time passes and tools develop, questions get closed as duplicates even when the question is slightly different or the answer won't work any longer.
> If it solves the question, it is exactly the right answer, regardless how much time passed.
But that's the crux of the matter - what exactly is this "question" entity?
I'm reminded of Heraclitus's "You cannot step into the same river twice" - if a question is an utterance made by an individual at a given point in time, then it might have "exactly the right answer" that they'd accept, but then it's extremely sensitive to those conditions, and even an identically worded question asked by the same individual a week afterwards might merit them to choose a different accepted answer, because the context for their question changed. For example, something like "What’s the proper way for me to deliver a signal to a different thread?" can have a myriad different contexts, and even if you give a wall of additional information and code, there probably would be something important omitted.
On the other side of this spectrum, we treat questions as a pointer to an underlying platonic idea of the question, which is exactly what StackOverflow say in their guidelines [0]:
> There are many ways to ask the same question, and a user might not be able to find the answer if they're asking it a different way.
Indeed, the closed-as-duplicate label uses the following text make it clear that they approach a question as an idea independent of the context in which it was asked:
> This question already has an answer
So if a question is independent of its questioner and can lead to other questioners being told that their utterance is a duplicate even a decade afterwards, then why should the original have any extra rights in deciding which one answer is the right one? Shouldn't this be owned by the community? At the very least, if a questioner writes a good question that is marked as a duplicate, then they should be given the same access to decide on the accepted answer to the merged question.
[0] https://stackoverflow.com/help/duplicates
In some ways Stack Overflow feels dated, but in some ways I hope it will go on; I still often prefer responses found there from the AI's. It also brings the recurring theme of what the LLMs will be trained on when people create less and less content. But I guess some people will always, and it might be enough
I think if we started trying to track the version numbers of questions and answers, and attached them to existing answers and questions, things would sort themselves out rather quickly.
The real problem is asking a question about how to do something in Java 8, becoming the canonical answer, and now we're on Java 25 and your answer is still on top even though we've gone through two new strategies for idiomatically representing that slice of code in the API and there's a 3rd in beta testing.
This feels hopelessly dated and given the reception it doesn't seem like it works that well
Isn't this a standard feature these days?
> This feels hopelessly dated and given the reception it doesn't seem like it works that well
For some reason, the copy button won't even include a valid attribution source URL at all:
I just tested it. Yes it does?
Oops wait no URL blindness just kicked in. You're right, that absolutely is not valid :)
Isn't AI doing this efficiently already in gemini?
Reads like from another time, a better time when not everything was ruled by tech and ai.
2024?
[flagged]
Genocide support? That must be an exaggeration.
Nope, they were banning accounts network-wide for calling the Gaza situation a genocide on the politics site.
Lame.
I can't wait 'til the site is dead, they have the worst community on the planet, even worse than Reddit and [REDACTED].
Worst thing is they saw this coming and doubled down on what everyone was telling them was the cause of trouble. There were memes out of it.
Classic example of product people leaving and marketing ones taking the helm.
It was a lifesaver back in the day, but struggling for an answer and desperately looking for a blue link, I don’t want to go back.
> worst community on the planet
I wouldn't say that, but it is a pretty annoying community, and one that I'm happy to leave behind, in favor of LLMs.
I think you may be right about the "doubling down." The Meta discussions seemed to get a lot nastier, as time went on. Might have something to do with centrists being driven out by zealots. Happens all the time, especially in communities in crisis.
There are definitely a lot of zealots and fanatics