0

From time to time I see flawed answers in this forum which are not really suitable for accomplishing the task described in the question and which nonetheless get a lot of upvotes.

Therefore I suppose that there is a considerable number of people who - when they like/worship the author - click the "upvote"-button either without thinking about the quality of the answer and/or without grasping the answer.

In another thread someone asked for guidelines for clicking the "downvote"-button.

I hereby ask for serious guidelines for clicking the "upvote"-button.

Such guidelines could, e.g., explicitly demand that you only click the "upvote"-button when you understood the answer.

And how about this:

People could be allowed to ask questions on TeX-LaTeX Stack Exchange in any case. But answering/commenting could be restricted to accounts whose keepers/maintainers were able to pass a knowledge-test where they have to answer at least a few basic questions related to the ways in which (La)TeX works.

  • See here: tex.meta.stackexchange.com/questions/7627/…. – CarLaTeX Dec 25 '18 at 12:02
  • 2
    If everybody is allowed to ask, everybody is allowed to answer. If you want to ensure high quality answers, pay for it. All this here on TeX.SX is happening in peoples spare time. – Johannes_B Dec 25 '18 at 15:40
  • Out of interest: Which knowledge test did you do? How and by whom have you been graded? – Johannes_B Dec 25 '18 at 15:47
  • On the one hand - this addresses the comment of @Johannes_B - I think that thoroughness/carefulness should not just be a matter of pay. The circumstance of having paid for an answer alone does not guarantee that it is of good quality. Besides this: I usually get best-quality-answers at TeX-LaTeX-StackExchange for free. On the other hand - this addresses Ronald Klopp: Instead of agitating for prohibiting "ungraded" people from answering and upvoting, feel free to comment answers whenever you see room for improvement. (I might learn from the comments people give to my answers. ;-) ) – Ulrich Diez Dec 25 '18 at 17:09
  • 2
    Assume an answer does not provide exactly what the questioner asked for. But it is written in understandable language, does explain the underlying facts, and does point the questioner towards the right direction, and thus puts the questioner into the position of creating a solution her-/himself: Why not upvoting such an answer? – Ulrich Diez Dec 25 '18 at 17:27
  • @UlrichDiez academia.stackexchange.com/questions/54006/… – Johannes_B Dec 25 '18 at 17:31
  • 3
    Note the site mechanics are network wide so you should raise this at the network meta not here (but it would be rejected, it goes against the basic workings of the network) – David Carlisle Dec 26 '18 at 11:46
8

Central point

Like in any democratic model, it cannot be excluded that some votes are undeserved by certain standards. However, I would like to argue that the alternative of restricting the participation too much likely to make the situation worse.

What is the current situation? The moderators can, in principle, undo so-called voter fraud. Examples include serial upvoting or targeting a specific user. However, they use this very rarely, for very good reasons IMHO.

Is the current system "flawed"? I do not share the opinion of AndréC, who states that "At the same time, it is regrettable and difficult to avoid since the mathematicians, the scientists and the computer scientists behave like fans of the show "the Voice": we vote for love and not for reason!". In fact, I'd like to argue that one should be very careful with such statements. Who is "we" referring to? I could imagine that someone reading this is feeling wrongly criticized. Rather, I think that overall the system works OK. There is a correlation between the quality of the answer and the score. True, the system is not perfect, but I do believe changes along the lines you are suggesting are more likely to make it worse than better.

Why am I not supporting your proposal? The basic point is that if someone feels that a given vote is undeserved, it is still just one personal opinion. The idea of a democratic model is that each of us has a personal opinion. Each opinion has the same weight. So if a user thinks a given post does not deserve a vote and 100 think it does, the post gets 100 votes. IMHO it would be very unfortunate if a single user, regardless of whether or not their reputation score is enormous, could override the opinion of the majority. Your proposal may lead to this situation: there will be "better" users who have passed some test and "less important users" who did not.

What can you do if you think an answer is off? It can happen that answers are incorrect. You can leave a comment. If there is a serious problem with the post, you can flag it in order to alert the moderators.

Bottom-line

Each of us has the right to vote. Use your voting power to upvote those posts which deserve it. If everyone does that, those posts which are considered more useful by more users will win. Of course, reminders like yours about what one should use as a basis for the decision to up- or down-vote are fine.

  • I don't remember saying that the system was not working well. – AndréC Dec 25 '18 at 19:28
  • @AndréC Well, you write: "we vote for love and not for reason!". How about we call it a language problem and you leave me alone? – user121799 Dec 25 '18 at 19:33
  • It's not about the voting system, it's about the people who vote, right? – AndréC Dec 25 '18 at 19:35
  • I'll leave you alone, but you talked about me. It is true that it is easier to talk in French, but well, nobody is perfect. – AndréC Dec 25 '18 at 19:40
7

Use the following semi-objective criteria:

  • Does the post show research effort? Is it useful and clear?
    This is literally the tooltip text associated with an upvote.

  • Does the post benefit the broader community?
    If it's a question, it could be simple, but still benefit the broader community. If it's an answer, does it explain what is being done to solve the question (perhaps through comments, or using a discussion)? Code-only answers don't entice my trigger finger.

  • Does the question make it easy to answer it?
    Don't tell me "it doesn't really warrant a minimal example"... I really don't want to copy your code and then wrap it into a \documentclass and document environment when you can so easily do it. Make the effort to support the folks who's trying to help you. Honestly, we'd much rather want to just copy-and-paste-and-compile and be on the same exact page as you. It's not that difficult, and it helps immensely.

  • 2
    +1 in particular for the 3rd point! Many time I see useless (for future users) just-do-it-for-me questions with a lot of upvotes! – CarLaTeX Dec 27 '18 at 8:32
  • @CarLaTeX It seems to me that as long as no one answers a just-do-it-for-me question, there are no votes, just comments. But as soon as one answer has been made, then the question is upvoted. – AndréC Dec 27 '18 at 10:13
5

Scope of the problem and appropriateness of measures It is not clear to me whether there is a significant number of flawed answers with (undeserved) high scores. I don't doubt that there are instances of this pattern, but I would like to see hard data that confirms that this is indeed an issue that warrants new 'upvote rules' or even a radical change to the workings of the site (it should be noted that 'we' can't unilaterally change the way this site works in a way to require a knowledge test before one can answer, comment or vote and that the powers would quite probably not look favourably upon such a proposal since Stack Exchange is already seen by some as an elitist place). I have probably backed myself into a corner demanding hard data on vote patterns (which are not openly available), but I feel that it is important to realise that a stark change in the site's dynamic needs a good reason: You don't want to shoot sparrows with cannonballs as they say in German.


What level of flaws are we talking about? It is probably true that some highly-voted answers have flaws by some standard or another. But keep in mind that one person's flaw might be another person's feature. Sometimes a balance needs to be struck between catering for every eventuality and keeping the code short and its explanation understandable. The concept of a flaw and the question whether or not a flaw is bad can therefore be quite subjective. At the same time the question whether or not an answer is suitable for the task described in the question is to a lesser degree subjective as well: Sometimes an answer gives the question a new framing or goes about resolving a XY problem. One may feel that the question has not properly been answered in that case, but one can also feel differently about that.

I doubt that there will be a lot of highly-voted answers that are outright factually incorrect. On the other hand I have no doubts that there are some answers that could be improved or for which more elegant alternatives exist.

Post score as a measure of quality It is also true that some (many?) people have had the experience that the answer score does not always reflect the post's quality or the time and effort spend on that post. One of my highest-voted answers is an answer to Force periods in "Ph.D. thesis" in biblatex that essentially just consisted of one redefinition of a string – I mean, the information is correct (in that the code will do what one would expect) and usable, but I would not rank this answer among the top ten of my answers on this site. On the other hand posts like my answer to How to make biblatex treat the prenote like an author list? only score 1 even though the question is much more intriguing, was more tricky and I spent a lot longer working out that idea.

Voting baselines can be different Different topics and kind of questions generate different levels of interest and so comparing scores across questions is probably not always a good idea: It would be more relevant to compare scores of answers to the same question. But even then the baselines can be seriously skewed by answers coming in much later than existing answers. The effect is amplified by the usual sorting order of highest ranking first, which means that people who are just looking for a working solution and are not shopping around for the 'most elegant' or 'conceptually nicest' answer vote for the first solution. Remember that for many people LaTeX is just a tool to get the real work done. The TeX code itself is almost never of interest, the result matters and so often one can get away with subpar or less elegant code.

If you want votes to truly and comparably reflect the quality of an answer (leaving aside subjective criteria of which there are still going to be a lot), you need a significant overhaul of the voting system. It would not only be about who is allowed to vote, it would also have to be about everyone voting on all posts to ensure a fair baseline. If that brings down the number of votes significantly below the current level (there are some highly voted an answers, but the median score of answers is around three), a simple up-down system might not give enough nuance any more, a more complicated voting/scoring system might be needed.

Guidance on voting behaviour In general there seems to be very little official guidance or rules about voting. This makes sense since an idividual's voting behaviour is not part of the public record and so there can be no enforcement of any sort of rules except for extremely abusive behaviour of the voting mechanism (serial voting, 'downvote targetting', ...) carried out by Stack Exchange staff. I personally think that it (only) makes sense to vote if one (firmly) believes that the post deserves that vote and that usually that belief should come from an understanding of the contents of the answer. But there are different levels of understanding an answer and probably only the highest levels of understanding allow for the judgement you seem to hope for. In particular then, only people who already know the answer should vote. People who by having read the answer have gained an understanding of the issue at hand and believe that the answer did well in enabling that would have to be barred from voting because their basis for voting would be the presentation of the answer and not the underlying facts themselves. I can't speak for anybody else, but a not insignificant portion of my votes were spent on answers that explained things that I did not know before well.


As for restricting answering (and commenting!) to people who have passed a knowledge test, that would go completely against the spirit of this network. The idea is that anyone here can ask and answer questions and share knowledge.

Level of the admission test I'm also not convinced that this is going to help much. Depending on your interpretation of 'flawed answer' you might have to set the bar pretty high to avoid small flaws slipping into answers (for highly voted answers serious flaws should be quickly identified if many people are allowed to vote and leave comments, but small corner cases may need very specialised knowledge that is not found in many). That may just turn the entire site into the seven gurus of TeX answer your questions – an interesting idea maybe, but certainly not in the spirit of Stack Exchange. Setting a low bar will mean that those who are able to find small flaws and never post answers with flaws themselves will be in the minority again. (Especially if one accepts that there are few answers with serious flaws that makes the restriction to people who pass a test not very attractive.)

A general test or tests for tags? Even though the TeX world may be small by some standards it is still so large that some people have specialised set of skills. You would probably be fine reading my biblatex answers, but whether a tex-core answer of mine is going to be good is a different question. Some people are great with TikZ, other with picture mode. A generalised knowledge test is not going to cut it if you want to cover more than the pure basics. Admitting users for different tags separately is going to be a huge ask. You need people setting the questions, you need a committee approving them, ...


I'll close this ramble by echoing marmot's advice (from their answer which I also otherwise agree with) on what you can and should do if you notice something is wrong with an answer: Leave a comment. Explain what is wrong. If you think you have a better solution, post an answer yourself (though you shouldn't expect it to rise very high if you are late to the party, even if it is superior).

  • I do not agree with your statement "If you think you have a better solution, post an answer yourself". It just needs to be different. It is the diversity of approaches that allows us to evolve. – AndréC Dec 28 '18 at 19:48
  • 1
    @AndréC But if surely your maxim is "If you have a different approach, post an answer yourself" you must in particular agree with my statement "If you think you have a better solution, post an answer yourself" (at least if the "if ... then" is the usual material implication and if we assume that having a better answer implies having a different approach). That said I don't disagree with what you said, though I would be hesitant to add something that I myself regard as an inferior suggestion when there already is an answer. – moewe Dec 28 '18 at 22:04
  • My understanding is as deep as out burrow. ;-) (No, more seriously some of your, but not only your, posts were the reason I joined this site. They helped me to solve some problem with the bibliography. Sadly I used them before I joined, so no upvotes for those.) And I also would say it makes more sense to add an answer if it has advantages compared to existing answers. At least with TikZ one could draw the same thing sometimes in 20 different ways, and I would like to argue that it does not make sense to write 20 "different" answers then. – user121799 Dec 28 '18 at 23:26
  • @moewe When you say better, you classify the answers according to criteria. What are these criteria? – AndréC Dec 29 '18 at 6:39
  • @AndréC I used 'better' in the context 'if you think you have a better solution', so I was referring to a subjective idea of 'better' or 'good' here. Same with 'inferior' in my follow-up comment above. Personally, for me correctness plays a role (obviously; by correct I here mean that the answer does what it was intended to do), edge cases play are also interesting, but the rest is mostly aesthetics. – moewe Dec 29 '18 at 6:51
  • @moewe It is this subjectivity that is the problem. By saying different, subjectivity is avoided. – AndréC Dec 29 '18 at 6:55
  • @AndréC I'm not sure why subjectivity would be a problem here: After all, coming up with and writing down an answer are all subjective processes. There are probably many ways you could write an answer, but you settled on a particular one because you believed it would be the best way. You might have used objective criteria for that assessment, but in the end it is subjective how much weight you place on which bit. Avoiding subjectivity for the sake of avoiding subjectivity seems pointless. – moewe Dec 29 '18 at 7:06
  • @moewe If we are in subjectivity, all points of view are equivalent. – AndréC Dec 29 '18 at 7:22
  • @AndréC Mhhhh, in a very black-and-white world where the only choice is absolute subjectivity or total objectivity maybe. But our world is probably more nuanced and there are shades of all sorts of -ivities in between. There are objective criteria to judging an answer (does it give the expected result in the relevant test cases? is the information factually correct), but then there are components that are up for discussion (aesthetics: shorter vs longer code, coding conventions, ..., edge cases, extensions of the idea ...) and so more subjective. – moewe Dec 29 '18 at 7:33
  • @moewe That is why I asked you what the criteria are for saying that one answer is better than another. You started the list. Keep going. – AndréC Dec 29 '18 at 7:38
  • @AndréC These are the main criteria that I could come up with on the spot. I'm sure I could pluck a few more - less impressive or important ones - out of thin air if pressed for it and shown examples. But I'm not sure this leads anywhere. In the end you have to coerce all these criteria into one decision (how do I vote: up, down, not at all? how do I rank answers A, B, ... X against each other?). – moewe Dec 29 '18 at 8:12
  • Let us continue this discussion in chat. – AndréC Dec 29 '18 at 9:34
2

It is true that there are a significant number of people who vote for the answers or for questions by pure sympathy. There are several reasons for this:

  • these people have already made great answers and benefit from a vote of presumption: yesterday's correct answers allow us to assume that this one is also a great answer.
  • Voting for people and not for answers is economical: it avoids having to read and understand the code of the solution, which saves time.

At the same time, it is regrettable and difficult to avoid since the mathematicians, the scientists and the computer scientists behave like fans of the show "the Voice": we vote for love and not for reason!

As for the remedy that you propose: it is impossible to implement because it would then be necessary to ask someone (and who?) To create knowledge test MCQs for each package. It is a gas plant.

  • 2
    I propose egreg and David Carlisle for creating the knowledge test. – CarLaTeX Dec 25 '18 at 12:06
  • @Johannes_B I really like the answer from egreg tex.meta.stackexchange.com/a/6692/138900 – AndréC Dec 25 '18 at 16:19
  • 1
    @CarLaTeX In case this was not just irony: I am against such a knowledge test as it might cause people new to (La)TeX to shy away from participating. I think no one depends on the reputation-system in a way which justifies a significant increase of the probability of keeping off potential participants. – Ulrich Diez Dec 25 '18 at 17:19
  • @UlrichDiez I was ironic, of course :) – CarLaTeX Dec 25 '18 at 17:38
  • 1
    @CarLaTeX I am not capable of irony. ;-) Nevertheless I enjoy ducks. :-) That's why I firmly insist on answers provided by our esteemed ducks a priori getting a sympathy-based reputation-bonus of at least 10000... ;-) – Ulrich Diez Dec 25 '18 at 18:01
  • Since voting records are not public and I'm not a mind reader anyway, I don't think we can say conclusively that it is true that a significant number of people only vote out of pure sympathy. I also think that we need to define exactly what we mean here: Are we talking about people trusting high-rep users when they write an answer (I guess that is implicit in the system) or about people targetting high-rep users with additional upvotes (why would they do that?). ... – moewe Dec 28 '18 at 11:21
  • ... The second bullet point somehow assumes that people vote for the sake of having voted (there are only a few incentives to vote a lot). Given the general voting level on this site I doubt that this is the case. It is much more likely in my view that people vote on answers (to questions) they find interesting. – moewe Dec 28 '18 at 11:25
  • @moewe Voting makes it possible to obtain badges. – AndréC Dec 28 '18 at 12:11
  • @AndréC Yes, sure. That's why I wrote 'a few incentives'. If I counted correctly there are four bronze badges (suffrage, supporter, critic, vox populi), two-ish silver badges (civic duty and sportsmanship) and one gold badge (electorate) related to voting, each awarded once per site. Two of the bronze badges (the ones related to one vote) are unlikely to skew the scores a lot: People who ask questions usually upvote good answers they get, people answering upvote competing answers if they are good or sources for inspiration or just posts they learn from. ... – moewe Dec 28 '18 at 12:38
  • ... Granted, the multi-vote badges (>=30 or 40 votes in a day) could make a difference, considering that this site does not receive as many questions and answers as other sites and so to get to the thirties one may have to bring one's standards down a bit or vote on stuff one does not fully grasp. – moewe Dec 28 '18 at 12:42
  • @moewe Voting is also an encouragement. Receiving a vote is a pleasure and makes you want to continue to participate in the life of this site. – AndréC Dec 28 '18 at 13:18
  • Mhhh yes, but I'm not sure what you're getting at. Do you want to argue that people vote for people (and not their answers) because they want to make them feel good and keep them here. Or do you want to argue that people vote for other people because they hope that they will reciprocate. Or do you want to say that people vote for other people because they want others to experience the warm fuzzy feeling that an upvote gives? Anyway, it is probably futile to discuss the intentions of voters since we just don't know their motives. – moewe Dec 28 '18 at 13:45
  • @moewe I think there's a mixture of all this in undefinable proportions. – AndréC Dec 28 '18 at 15:12
  • As I said: I doubt this is the main motivation why people vote, but I can't prove that. What I'm taking issue with is the claim at the beginning of this answer that a significant portion of people vote purely based on sympathy. If that is true or there are good reasons to believe that is true I would be interested to hear about it, but at the moment I'm assuming this is speculation. I also don't think it is fair to single out mathematicians and scientists as a fanboying/fangirling crowd who completely abandon reason (is there any evidence for that generalisation), but then I'm a mathematician. – moewe Dec 28 '18 at 15:23

You must log in to answer this question.

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