5 ms·

> 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 mass

by js8 9y ago

> 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).

mlvljr 9y ago
Exactly, the answers like "been using library X for three years but finally gave up due to Y" is the kind of personal advice-style answers that usually are the most helpful. PC mods crowd, you're missing that! :)
ufo 9y ago
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 Opinion questions can get outdated very quickly and it is hard to tell if you come from a google search if the answers are still applicable.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)
- acjohnson55 9y ago
  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.
  - shagie 9y ago
    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. Look back at one of those https://stackoverflow.com/q/22697 https://stackoverflow.com/q/22697 - "only" 14 answers to it... and look how many of them are saying the same thing. That's not useful. https://www.slant.co/topics/259/~best-mock-frameworks-for-java https://www.slant.co/topics/259/~best-mock-frameworks-for-ja... Try https://www.slant.co https://www.slant.co for the recommendations - its designed for that.
    - dcow 9y ago
      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".
      - shagie 9y ago
        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 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?
- Chaebixi 9y ago
  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).

by js8 9y ago

mlvljr 9y ago

Exactly, the answers like "been using library X for three years but finally gave up due to Y" is the kind of personal advice-style answers that usually are the most helpful. PC mods crowd, you're missing that! :)

ufo 9y ago

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)

acjohnson55 9y ago
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.
- shagie 9y ago
  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. Look back at one of those https://stackoverflow.com/q/22697 https://stackoverflow.com/q/22697 - "only" 14 answers to it... and look how many of them are saying the same thing. That's not useful. https://www.slant.co/topics/259/~best-mock-frameworks-for-java https://www.slant.co/topics/259/~best-mock-frameworks-for-ja... Try https://www.slant.co https://www.slant.co for the recommendations - its designed for that.
  - dcow 9y ago
    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".
    - shagie 9y ago
      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 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?
Chaebixi 9y ago
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).