17

The TeX packages have developed a lot since the first questions were asked on sx. Many high voted questions from 2011 do not reflect the situation today. The users who asked and answered the question do not monitor this. It would be unfair to down vote the accepted answer.

Should we tag some of these historic questions as historic question?

Any ideas how to handle answers which were good in the past and wrong or bad in the present?

We should help the user who finds a 500 up question from 2011, which has to be rethought in 2019.

Somewhat related: Referring to new, similar questions in old posts

  • Also, there are related questions on other sites: meta.stackexchange.com/questions/131495/… – sheß Apr 2 at 15:00
  • 3
    Maybe it would be worth considering, as an alternative, adding a tag "obsolete". This does not necessarily imply that it should be deleted, only that the methods shown should not be used for new documents. – barbara beeton Apr 3 at 17:43
  • 1
    Out-dated questions are one thing. But are they really a big problem? IMHO out-dated answers are a more problematic, because readers use them, because they are high voted and perhaps accepted. Unfortunately users often do not read comments and often do not read more than one or two answers. – Schweinebacke Apr 5 at 12:06
  • 2
    @barbarabeeton: A tag may be great, but if the warning is only in a tag, I suspect most users will overlook it. Tags aren’t very visually prominent, and experienced users know to check them, but new users don’t. – Peter LeFanu Lumsdaine Apr 10 at 21:06
  • 2
    @PeterLeFanuLumsdaine -- That's precisely what I pointed out in another answer here. I'm strongly in favor of a very visible "banner" at the top of the entry, maybe even edited into the top of the question. Something that absolutely can't be overlooked. – barbara beeton Apr 11 at 0:17
  • Ahhhh the benefits of plain O:-) – morbusg Apr 15 at 22:37
12

Given that some old but no longer applicable answers really are great, it would be a shame to remove them. (They would be good grist for future TeX archaeologists.)

On the other hand, as has been pointed out in another answer here, adding a "warning" answer to a question with highly-upvoted answers would not necessarily be noticed readily by someone (in particular a newbie) as it will start off with zero votes, and the likelihood of it becoming the "top" answer is negligible.

What about trying to formulate a "stock" statement that can be added at the top of the question itself, that documents the fact that while the answers given were valid at the time, as of <date> they no longer apply for <reason>. Such a statement could be inserted with a colored background, and whoever adds it should sign it. Formulating such a "warning" statement would be a reasonable topic for a meta question.

I would be happy to do this, at least for questions for which I submitted an answer, if I happen to come across them, or if prompted by someone else who finds such a problem entry.

Edit: In a comment, @Skillmon points out that some old, highly upvoted answers are outdated, although the original question is still valid, so it would be misleading to mark the entire posting as obsolete. In such a case, it would be appropriate to mark the answer(s) only, leaving the question intact.

Even some "current answers" are bad, although they satisfy the OP and are accepted. I usually address those by adding a comment saying why they don't work (or don't work in a more general case). I suspect that such comments aren't often seen by a newbie looking for a quick fix, but I feel that editing the answer is bad manners. I don't feel that way about an answer that is several years old and outdated by an actual change in the core or a package.

  • 1
    That is pretty much what I had in mind when I first read the question and in the discussion under marmot's answer. I think this could be really helpful in some cases. But it is also true that there are a lot of questions and answers on this site and that it would be a lot of work to try and monitor them for outdatedness... Nevertheless, I would welcome it if the community agreed that such a warning message is a useful thing and endorsed adding it to outdated posts. – moewe Apr 2 at 20:08
  • 3
    I have edited updated a few highly-scoring posts by inactive users to make them compilable again before and I always felt a bit weird doing that, so it would be nice if there was a consensus to support such edits. – moewe Apr 2 at 20:09
  • @moewe I appreciate your willingness to do all this. (Personally I still do not see what is wrong if there are new questions which say: "I tried this solution proposed in that answer, but I am not satisfied with it for the following reasons:... ". And then a large number of users will see this and have a chance of saying "By now there is this and that new tool which allows you to ...". I am simply afraid that if single users go through these lists they may miss some things, which won't be missed if the post is seen by everyone under "New questions.") – marmot Apr 2 at 20:46
  • @marmot New questions are not wrong, but for highly viewed and voted old posts that are often among the first hits when you search for the issue it would be a shame to have them hanging around in an unusable state if something can be done about that. See for example my edit to tex.stackexchange.com/a/73246/35864, tex.stackexchange.com/a/70183/35864 or tex.stackexchange.com/a/46879/35864. How one gets alerted to the issue is a different matter... – moewe Apr 2 at 20:52
  • @moewe These are all valid points but imagine LaTeX3 becomes reality. Would you then rewrite all answers to say "Hey guys, you could do that with LaTeX3 in this and this way", or just wait till someone asks for this? – marmot Apr 2 at 20:56
  • 1
    @marmot Mhh, I'd say what I currently find acceptable is much narrower in scope: I edited answers that would otherwise have stopped working due to low-level code changes. Additionally, I would find it acceptable to edit answers that use deprecated commands that now issue a warning and encourage the use of their newer replacement. In some cases I would also find it acceptable to edit uses of deprecated macros that have only been clearly marked as deprecated in the docs but don't issue a warning in the .log, though that is a more delicate matter ... – moewe Apr 2 at 21:04
  • ... I would not, for example, replace all \bibliography{<bibfile>}s with \addbibresource{<bibfile>.bib}s in biblatex documents even though the biblatex docs have \bibliography down as deprecated. Re your LaTeX3 thought experiment I'm not quite sure what the exact parameters are there (has LaTeX3 replaced LaTeX2e or are we talking a situation that is more like the status quo?), so I'm not quite sure where I would go with it. W.r.t. the current LaTeX3 situation I would not edit answers to give a LaTeX3 solution, but if I thought one would be useful would write a new answer. – moewe Apr 2 at 21:10
  • @moewe You probably just got me in the wrong moment.;-) I am currently converting old transparencies, in which I used pstricks, to TikZ (and do other changes). The old transparencies were OK, but with my current knowledge I would write many things in a very different way. The same things can be said about older answers of mine (and also by others, I suppose), but going through all of them seems to me a bit tough. The more so since I expect that in a little time I would know even better ways. – marmot Apr 2 at 21:13
  • 1
    @user49915 -- I don't disagree with your analysis, but think that a very visible warning at the very top -- edited right into the question -- should be sufficient to make the point. As I proposed, such a warning should be signed, and should ideally also not be just boilerplate, but should contain useful information that leads to a current answer. Maybe this doesn't adhere to the letter of the site's rules, but it does adhere to the spirit, which is to provide a corpus of valid and useful information. – barbara beeton Apr 3 at 16:36
  • 2
    @user49915 -- I didn't mean to imply that only the original poster should be able to update with a warning, and not only the answers. Please read my proposition again. For this to be accepted would require consensus from a large segment of this group, and an agreed-on format. And I doubt that such a warning would help the Web engines, but if done as I propose, would surely catch the attention of a new visitor. – barbara beeton Apr 3 at 17:18
  • 1
    I would like to add to the discussion that there might be questions which are still valid, but the top answers are outdated. If we agree on adding a disclaimer for outdated information, I think it would be reasonable to add it to outdated answers, not to questions with outdated answers (provided the entire question isn't outdated, of course). – Skillmon Apr 5 at 18:04
  • 2
    @Skillmon -- Good point. I agree. Even some "current answers" are bad, although they satisfy the OP and are accepted. I usually address those by adding a comment saying why they don't work (or don't work in a more general case). I suspect that such comments aren't often seen by a newbie looking for a quick fix, but I feel that editing the answer is bad manners. I don't feel that way about an answer that is several years old and outdated by an actual change in the core or a package. – barbara beeton Apr 5 at 18:20
5

I would say that if someone downvotes an answer just because it made use of tools which were available at the time the question and answer were written, there is not much you can do. You can try to protect these answers by declaring them "historic" but, honestly, you will never prevent users of the above-mentioned type from doing harm. In addition, if an answer is tagged "historic" this may then lead to the impression that this answer is "not so good and requires to be protected", but many of these answers are great. For these reasons I am a bit skeptical about your proposal (but see of course the good intentions behind it).

After some chat with moewe, I believe to understand the purpose of your question/suggestion better: you seem to want a clean-up of old posts. I fully support this but still do not think a "historic" tag will be instrumental to achieve this. Rather, we need to have competent users who are able to judge what is outdated, and are passionate enough to go through the older posts to "clean up". Notice that this is a highly nontrivial task because it is often not black-white: method B is strictly superior to A. (To give you an example, the new command \tikzmarknode which comes with the newest version of the tikzmark library is clearly an improvement over some \tikznode commands that you can find on our main site since it detects the mode and color and so on. However, often you get to hear that the solution is not useful because the OP are using overleaf, so they do not have access to newer packages and libraries. SIGH.) Another complication is that those competent and passionate users who you expect to go through the older posts know all-too-well that what they are adding now is likely to be outdated in a few years.

So my proposal is that, in order to "clean up",

  • we may instruct newcomers that it can help to also look at the time stamp of a post, not only at the votes, and
  • we agree that if we add a new answer we explain what is better or new compared to existing posts (and give proper credit to them, of course).
  • 2
    My main motivation to mark answers or maybe entire questions as historic would not be to prevent unfair downvotes, but rather to warn present and future visitors that there are better ways to tools available now. Especially older, highly-voted answers to common questions risk spreading outdated information and the high vote score gives them visibility and legitimacy. – moewe Mar 31 at 16:30
  • @moewe But "better" is an opinion. What is wrong with adding a new answer to a question saying by now there is an alternative available? I would not think that referring to the alternative as "better" is necessarily a good move. – marmot Mar 31 at 16:33
  • Sure, take the "better" with a grain of salt. I'm mainly concerned with answers that contain outdated information (e.g. mention a limitation/bug in a package that has been removed, mention other states of affairs that are different now ...) or questions with an outdated premise. With highly scoring answers a new answer or a comment that comes in much later may never reach the level of visibility of the older answers. This is especially problematic for users who want a quick answer and gravitate towards the highly-voted ones and exacerbated by closing new questions as duplicates of the old one. – moewe Mar 31 at 16:41
  • As an aside I do believe that sometimes it can be useful to label new alternatives as "preferable" (or maybe even "better") especially if packages evolve and officially deprecate the older method. This is not about belittling or criticising earlier answers it is about recognising the fact that sometimes things evolve. – moewe Mar 31 at 16:46
  • @moewe But my concern is that "historic" will precisely do that: tell the user that this answer is outdated, which many may take as "not good any more". I would like much better if we could instruct users to also look at the time stamp, not only at the votes (and if we could manage to only add a new answer only if it has real advantages, and the new answer explains what is new and gives proper credit to the existing post). – marmot Mar 31 at 16:50
  • In any case the historic label is something that would have to be handled very carefully so that there is no suggestion that it is just used to promote certain new "better" answers when the old answers are not factually out of date. – moewe Mar 31 at 16:50
  • Looking at time stamps alone is often not enough. Many answers from more than ten years ago are still perfectly valid, while some answers as young as only a few months could potentially be outdated and not applicable any more. People who want to find an answer usually won't know if an answer is outdated or not. Of course the "historic" label will warn people that the information is outdated and that relying on it is "not good (enough) any more". That is the point, but I don't see why that would be inherently bad. – moewe Mar 31 at 16:58
  • Suppose back in 2011 someone asked about packages A and B. Those packages are very widely known and many people want to know which one to use. At the time package A was not compatible with package C and so people recommend to use package B, which was at the time seen as a successor/alternative to A, instead. Fast forward eight years, development of package B has stalled, development of package A has taken up speed again. A is compatible with C and works overall very well. Package B still works well enough, but rough edges are starting to show. .... – moewe Mar 31 at 17:04
  • ... The answer saying C and A are incompatible is outdated and not accurate any more. If one were to ask people now, most people would recommend A over B. Still the answer has accumulated (deservedly so) a very impressive score. What should be done in this situation? – moewe Mar 31 at 17:07
  • @moewe These are all very interesting thoughts and problems, but I still believe that a "historic" tag will not solve them. The crucial information here is that some competent user has the knowledge to judge what is going on. If that competent user has time and passion, they can add this information to the respective post. This is independent of a "historic" tag. – marmot Mar 31 at 17:14
  • 1
    You do have a point that a "historic" marker would have to be applied by someone and that that someone should probably be competent enough to just update the answer to reflect the status quo (if the author of the answer is not around, if they are around they might as well be nudged to update the answer themselves). – moewe Mar 31 at 17:35
  • 1
    @moewe Yes, that is what I want to say. We would need some competent volunteers to go through the older posts and to add some information, and an agreement by the community that we need a clean-up. Just introducing the "historic" tag won't buy us much. – marmot Mar 31 at 17:37
  • 1
    @user49915 "No more compilable" is a relative term. If you ever wrote an answer for an overleaf usable, you may have produced a code that is "not yet compilable" for them. – marmot Mar 31 at 17:57
  • @user49915 Overleaf currently uses TeX Live 2016, whereas the most updated distribution is TeX Live 2018 and 2019 is coming soon. For example, with Overleaf you can't use a new package created in 2017, unless you manually add it to your project. – CarLaTeX Apr 3 at 6:15
  • You can instruct newcomers a lot, but who does read such instructions? – Schweinebacke Apr 5 at 12:08
4

We could add the year when the question was asked to the title like "How to ... in 2011"

  • this leaves the old question in a good state
  • we can open a new question "How to ... in 2019" on demand and link the questions
  • a good question or answer will not see down votes if it was good in the past
  • transparent for the reader
  • +1, best idea till now, in my opinion! – CarLaTeX Apr 5 at 8:29
  • But what about outdated answers to still valid questions? And would we need a vote mechanism to decide whether or not a question or an answer is outdated? – Schweinebacke Apr 5 at 12:01
  • @Schweinebacke The answers are already linked to the question, so it is no problem. – Jonas Stein Apr 5 at 15:29
  • @Schweinebacke We need no additional vote system. On other sx. they ahve to ask the same question for every version again. TeX is more static, but the packages develop too. For duplicates we can either close as duplicate, or answer it is still the same as in 2011 (with link). – Jonas Stein Apr 5 at 15:40
  • I think we are talking at cross-purposes. But doesn't matter. – Schweinebacke Apr 5 at 17:38
4

The problem of high-voted question that do not reflect the situation today is a problem, but is it really a big one? If the question does not reflect the situation, does anybody searches for such problems? And couldn't such question not simply be closed with a new close reason »historic only«? Closing is not a vote statement.

The problem of high-voted answers that do not reflect the situation today is a related problem and I think the bigger one. Some of them are really good answers with useful information about how things work and exciting code examples, but nevertheless not longer recommended or even can result in errors. So for one purpose of a question-answer platform: “giving solutions for problems”, such answers are not longer useful. But for another purpose of such a platform: “providing know-how to find solutions for problems”, they are still useful.

So for newcomers it would be useful not to be diverted by such answers that do not longer provide the solution or are not longer recommended to be used as a solution. If I find such an answer I usually write a comment in the hope, that the author of the answer finds a solution to improve the answers. But sometimes even the author is not able to do so. And sometimes removing such an answer indeed has been happened. Removing an answer does not mean, that the platform deletes it. It is just marked as removed and users with high enough score still can see them. So even removed answers are not lost. However, you cannot longer vote for them nor can you comment them nor can the average user see them. And you must read the answer and/or the comments to find out whether or it is still useful.

Maybe a second kind of removed answers could be useful: “Not longer an answer to the question but still useful.” We could also name them “closed answers”. The name does not really matter. Such answers should not be shown for newcomers by default but maybe even newcomers should be able to display them. And it would be useful to be able to comment such answers and vote for them.

However this would be a completely new platform feature. So currently a stock statement indeed could be a workaround.

But IMHO there is another problem: Who should decide whether or not an answer is still valid or not? Do we need to vote about? Or should everyone who has the privilege to edit answers be able to add the stock statement?

  • There are many users (high rep or not) who have the requisite knowledge to judge answers in a specific area. My opinion is that someone coming across a probably outdated answer should assess their knowledge honestly, and seek confirmation from other knowledgeable users in the chat if there is any hesitation, before taking the leap. – barbara beeton Apr 5 at 18:29
2

I don't know if that's possible with stackexchange, but couldn't this be added as a flag/review-queue?

Suggestiosn: Once you discover an outdated question, you flag it to be closed for being too localized. The same should be possible for answers, i.e. users can raise an "outdated" or "too localized" flag on answers and then more experienced users can review those and either modify the answer or "close" them or mark them as "historic". Closed answers are kept but visually set apart (e.g. by showing a remark), this should deter people from downvoting it, but encourage new answers to be given.

1

This is the purpose of post notices.

Skeptics SE uses post notices extensively; see discussion on their meta.

  • 2
    Interesting concept. It would be nice to see it in action; none of the questions listed in the cited posting actually shows it. However, this question points out that the notice appears below the question and thus isn't immediately visible. If an "obsolete" notice is to be implemented, it really needs to be really obvious. – barbara beeton Apr 10 at 14:33

You must log in to answer this question.

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