10

This post today got me thinking about how to deal with obsolete answers. I feel that the solutions suggested to Mehrdad's question are not fully satisfactory:

  • IMHO it is totally unfeasible to expect anyone having hundreds or thousands of answers to manually browse through past answers to periodically detect obsolete ones
  • editing an answer long after it has been upvoted / accepted may introduce a risk of confusion and uncertainty, as it becomes blurry what the upvotes / acceptance were actually awarded for
  • also, old answers may still continue to be useful, even if they are not the state of the art anymore, for those who for some reason need to work with old technology
  • relying on new users finding and downvoting old obsolete answers is IMHO very inefficient with SO having well over a million posts
  • IMHO it is a bit unfair to downvote an old answer which was good and proper at its time of creation, just because the public opinion about the state of the art has shifted
  • merging fresh duplicates with their old original may easily hide the fact that highly upvoted / accepted answers to the old question have become obsolete in the meantime - even if a new answer is better / more up to date, it gets behind the older ones if it has less upvotes

So my suggestion is for cases when it is visible from comments to the old and/or the new question that the new question obsoletes the old one, to introduce a special two-way link between the two instead of merging them: "this question obsoletes this earlier question / this question is obsoleted by this newer question". Possibly together with locking the old question to stop further edits / votes, as there is no point modifying an obsolete post.

IMHO this would solve the above issues:

  • no need for anyone to manually update or delete thousands of his/her old answers - we can instead rely on the whole community to determine the state of the art
  • old answers are still kept for reference, but clearly marked as "archive"
  • new and fresher answers are not shadowed by obsolete but more popular ones

Please feel free to post your comments or thoughts on this. This is still a half baked idea; let's try to improve it - and SO - together :-)

1
  • Considering that people who can edit tag wikis are more or less 'stewards' of their respective tags, this might work. I'm just at a bit of a loss on how the system could help guide them to such reviews.
    – user50049
    Commented Feb 27, 2011 at 1:20

1 Answer 1

4

I have a slightly different view on this.

Each question represents one particular problem. And when version of the tool isn't specified, it's implied that the current version at the time of posting is used.
So, if existing answers solving some issue for .net 1.0 aren't applicable to your shiny new .net 4, it means you have a different question.

Although it's tempting to create a full list of solutions for single type of problem for all points in time, I feel like it's slightly out of scope of QA site.
I do support linking old questions to the new ones, yet comments seem sufficient for that.

And as for downvotes... Well, downvoters will downvote, they'll always find a reason, so I wouldn't worry much. Also, it might be a sign for you to refactor an answer or add 'obsolete' warning to it.

12
  • 1
    I am not after creating a full list of solutions, only managing the full list of solutions which seems to be created inevitably over time, just possibly on a haphazard and inconsistent way. Comments are versatile, but a "this post is obsolete" comment may easily go unnoticed, especially at the tail of a long list of existing comments where it isn't even visible by default. Commented Feb 26, 2011 at 22:21
  • The problem is, will your .net 4 question not get closed as a duplicate of the previous .net 1 question?
    – badp
    Commented Feb 27, 2011 at 1:23
  • @Péter From my experience, it'll flow up eventually (and more up-to-date answer to a 'popular' question too). I'm not opposed to other ways of linking, though. In particular, 'linked' section could be reused and improved to include 'label' and score from the comment. Commented Feb 27, 2011 at 1:40
  • @Péter Instead, I'm opposed to any 'hard' measures. Like locking a question or declaring one question 'less relevant' than another just because a random 20k clicked 'obsolete' link. Commented Feb 27, 2011 at 1:43
  • 1
    @Péter: If the situation has changed and the set of answers is now obsolete, edit the question to say “Note that this question was posted before feature X appeared. If you have Foo 4.2, see <new question>”. Lock the old question if there's really too much confusion. Commented Feb 27, 2011 at 1:44
  • @Péter Developing the comment idea. What if a comment "@obsolete! try this for jquery 10.3 http://stackoverflow.com/questions/12345" resulted in additional notice after the question text, together with comment score? Commented Feb 27, 2011 at 1:45
  • @badp If you've reviewed linked question and found that it's not actually a duplicate (for whatever reason, e.g. incompatible framework versions), edit it with corresponding notice and nominate question for reopen. Optionally, go to chat and ask folks to help with reopening. Commented Feb 27, 2011 at 1:48
  • @Gilles Yes, editing is also good idea. Commented Feb 27, 2011 at 1:50
  • @Nikita: Why go through the complication of a special comment, especially now that anyone can suggest question edits? Commented Feb 27, 2011 at 1:51
  • @Gilles I prefer comments to editing in general, since everybody has their own truth and voting allows to determine whose truth is truer. So, on a second thought, I'm not sure if I want people mass editing questions because they thought something. Especially, since we already have a problem with stupid edits being regularly approved. Commented Feb 27, 2011 at 2:00
  • 1
    @Nikita, I agree it shouldn't be the decision of a single user whether the new question obsoletes an older one or not. It could be resolved by votes though, similar to the question of duplicates. Possibly with an additional limit to voters, so that only those users are allowed to vote who are otherwise allowed to edit tag wikis for (some, or the most widely used of) the relevant tags. Commented Feb 27, 2011 at 13:38
  • Regarding the downvotes, I am not personally affected (as yet - I have been around only for a year). And of course I can put up with losing a few reputation points. Let's give an example to illustrate my point: Kernighan and Ritchie created C, which gives them great reputation. Should this reputation be lessened by the fact that since then OO languages have become more popular and many consider C as an obsolete, niche language? Commented Feb 27, 2011 at 13:43

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .