> They may be useful, but none of them is objectively right and can cause religious wars with people downvoting answers about libraries they don't like and massively upvoting their favorite library.
Like many, I find the answers to these "religious" (in fact, they are not, because done properly the answers are supported by arguments) questions to be the most useful content.
In many ways, I can answer the most objective questions myself from the documentation or source code. But what I cannot do this with is distillation of expert knowledge that requires time, and the more distilled it is, the more subjective it is. Subjective doesn't mean unsupportable by facts (experience).
Letting more opinion based question in stack overflow sounds like a nice idea but in the end it doesnt work out because of the conflicting goals problem john skeet talked about. Answers that just recommend a 3rd party library optimize for solving the immediate problem an answerer is having asap but they don't work very well at creating a long lasting resource and they also give answerers a hard time. Opinion questions can get outdated very quickly and it is hard to tell if you come from a google search if the answers are still applicable. They also don't work well with the site format because you cant summarize the solution into one accepted answer. Instead, you get hundreds of tiny small disconnected answers, which cant really be edited to consolidate or keep them up to date, and no room for discussions (SO comments don't work well for long discussions)
That's just repeating the SO dogma, which I think is suspect in most cases.
> Answers that just recommend a 3rd party library optimize for solving the immediate problem an answerer is having asap but they don't work very well at creating a long lasting resource and they also give answerers a hard time.
I see no problem with this. Furthermore, if you can ask a question about using a particular library, you should be able to give that library as an answer.
IMHO, any problems around "3rd party libraries" could be solved from a combination of improved culture (e.g. explain the solution with a library reference) and technical tools (e.g. tagging with library version, result/answer ranking tweaks, etc.)
> Opinion questions can get outdated very quickly and it is hard to tell if you come from a google search if the answers are still applicable.
SO could solve that problem rather than giving up. For better or worse, SO is the watering hole for programming questions. As long as dups are kept under control, competent people can find the old questions and provide new, updated answers that will converge on completeness. Also, "outdated" answers can still have value, since not everyone is using the latest stuff.
What SO needs is features for fixing the practical problems you outlined:
> They also don't work well with the site format because you cant summarize the solution into one accepted answer. Instead, you get hundreds of tiny small disconnected answers, which cant really be edited to consolidate or keep them up to date, and no room for discussions (SO comments don't work well for long discussions)
I would say some form of extra curation, answer clustering, and old answer decay would go a long way towards addressing this, and it would be a great improvement on the existing situation of digging through old forum posts or making new ones (to an audience that may or may not have good opinions).
Except, it does work. I've found many closed opinion questions to be just the info I was looking for. Until they effectively stamped that behavior out, it was one of the best ways to get a request-for-recs in front of a massive audience of informed professionals. It was messy, but very useful.
"Except, it does work. I've found many closed opinion questions to be just the info I was looking for."
Same for me. I think it is not wise to expect your users to be that fanatics, that they can't recommend any tool, without starting a flame war all the time.
I mean there certainly are fanatics on the internet who fight with religious passion about the weirdest things, but I believe they are not the majority.
So just banning something that might cause trouble, but loosing all the benefits, I think was not a good decision.
But apart from that, I found lots of help on SO. But like others, not so much anymore, but mostly simply because I rarely need to look up things and the problems I do have are too obscure, for the main users, which are the target group, so I solve them by other means.
It also took a significant amount of community curation to maintain the utility of that. As the communities that tried to maintain the posts have found that it didn't work well, those posts became worse and worse content that wasn't up to date and would have people adding a post for no other reason than to add a post (not even checking for if the answer was there before).
The utility became less and less over time. It can be done with a small enough and dedicated community behind it (the C++ book list), but without that amount of attention and dedication its not that good anymore.
There's also the number of questions. Stack Overflow back then got maybe a few hundred questions a day. You could glance over all of them in a tag in a page. Now it gets 8000 questions per day. Asking a recommendation question doesn't get your post there for a large community of experts. You're lucky if you've got a few minutes of visibility now.
It's easy for you to say they're not useful. But it's harder to ignore the myriad up votes. Even objective technical questions suffer from the "same answer multiple times" problem you explain. IMO repeat answers act to provided consensus in a more finessed way beyond just added votes. I don't think it's inherently "not useful".
In many cases, up votes on Stack Overflow have no more significance than a like on facebook.
The problem of "same answer repeated" becomes contrary to the "find the answer rather than scrolling through pages" when posts get sufficiently popular. Go through all the pages of https://stackoverflow.com/q/1711 and count how many times the mythical man month is mentioned. Unless you read through all 214 visible answers, do you know how much consensus exists from material on the last page?
Is that really so bad of a problem it requires sacrificing the utility of similar questions? It really feels like a few bad apples poisoned the lot. Maybe opinion based questions could have a TTJ (time to judgement) on them. If they result in useful info, keep them, if not, punt. IDK I'm just spit-balling now. I really don't see why SO has to take such strong ideological stances toward community moderation--which is a human issue not a mechanical one.
There's a place for such questions, it just isn't Stack Overflow. It isn't the trash either... but its that the structure and the framework of Stack Overflow doesn't work for that sort of question.
The main thing is that it takes some degree of moderation to keep the people who have the answers that people want. This means matching the site to the vision of the site that the people who have answers want to visit... and that's a lot of clauses. Without that moderation striving for what the site could be, you get https://answers.yahoo.com/dir/index?sid=396545663&link=list and there's a reason that experts tend not to go there and try to answer those questions.
Thanks for that last link. Really interesting read.
I've read Jeffs essay. I've used programmers.se and basically vowed never to go back after a few years because the moderation itself is piss poor (there one person gnat who's sole mission is to close every new question unless it's asked by one of their inner circle of buddies). It's really a joke. There's literally no reason for programmers to exist if not for expert opinioned discussions but they started cracking down on that and rendered the site useless. It's not just me either. Other good contributors have posted the same reasons for leaving in meta and in other forums.
I think you're conflating useful questions and good moderation with opinionated Q/A. I don't think anyone wants SE to be yahoo answers. But a question like what is the correct way to implement IO multiplexing is really useful. I think in SOs crusade to rid the site of opinion-based questions that are hard to handle, they threw the baby out with the bath water and got rid of some of the most useful discussions as a byproduct. Is SO dead? No. But I certainly get less utility out of it these days because I'm less willing to contribute when 90% of the time questions are met with instant downvotes and flags because someone didn't understand the finesse in a question and/or didn't assume positive intent so they could justify being pedantic assholes. I'm pedantic too. I draw the line at using it as a reason to be an asshole though.
I'm not saying SO didn't do its homework. But I am giving one data point that feels SO's usefulness has been marginalized because of the way the community has interpreted "no opinion based questions". I feel that zealotry has spilled out into the general air at stack overflow and it's no longer engaging or fun for many people to contribute these days. Take that for what it is, I guess.
Like many, I find the answers to these "religious" (in fact, they are not, because done properly the answers are supported by arguments) questions to be the most useful content.
In many ways, I can answer the most objective questions myself from the documentation or source code. But what I cannot do this with is distillation of expert knowledge that requires time, and the more distilled it is, the more subjective it is. Subjective doesn't mean unsupportable by facts (experience).