There have been some concerns regarding voting on m.se and as has been pointed out there, this may have to do with increasing number of questions, e.g. some statistics here.

In this post I wish to discuss and ask for some suggestions which address these concerns (most answers on the question regarding voting do not address what can be done about it) and which I think may help maintain quality on this site. I have been using the site just for 3 months, and am not very experienced so I request all to excuse me if I am suggesting something silly.

There is a genuine concern that sometimes good questions (especially the ones that are relatively advanced) do not get enough attention and remain unanswered presumably because the number of questions on m.se is large and each question spends lesser time on the front page. There has been a suggestion to split m.se into more sites but it has its own problems as can be seen by looking at the answers there. While I would not advocate splitting the site, but I have to say (based on my personal subjective feelings, and I am curious whether more experienced users agree with this) that : questions which are relatively advanced but not research level (so that they are out of place at MO) do tend to get suffered more due to the traffic despite the fact that there are enough users who would have answered these questions if they were not buried inside larger pool of elementary questions.

I feel that in order to help this situation, would it not be nice to have a new tab (in addition to the usual 'newest, featured, ..., unanswered' tab) for advanced questions which just slightly fall short of being appropriate at MO ? I would suggest that users be given an option to indicate (by checking a box) whether their question falls in this category or not. Users with high enough reputation / moderators could make sure that only advanced questions are seen in this tab and may have the power to transfer more questions into this tab. Would not separating the more advanced questions be helpful, assuming that it can be done ? I feel that this feature would attract several MO users to visit and answer questions on m.se who would not do so otherwise because their primary interest may be in advanced questions which are buried in a pool of more elemetary questions on m.se.

Coming back to the voting issue I feel that the following may help :

1.) There could be messages requesting a user to consider voting more if the user views large number of questions (say 5 or 10) without voting any of them.
2.) How about having some reputation points for voting (say 1 point per 2 votes cast with some cap on the reputation that can be earned this way say 10 per week) ? What would be the advantages or disadvantages of this system ?

To summarize I would like to ask :

What can be done to maintain the quality of the m.se website in the view of the large number of questions per day so that the relatively advanced questions are able to get the attention they deserve? and What can be done regarding decreasing voting on m.se ?

• If people agree that such a stratification is needed, it seems like the simplest way to implement it would be via tags (e.g. a tag “advanced-level”, or “undergraduate-level”, or similar). However, I have a vague recollection of a discussion of such a system in the past, rejecting the idea on grounds that it would be too subjective and difficult to give guidelines for. – Peter LeFanu Lumsdaine Dec 6 '13 at 22:45
• I once got a message that I was voting on answers but I should consider voting on more questions. (approximate wording) Perhaps there could be notices of people selecting favorite questions within our favorite tags. – Jay Dec 6 '13 at 23:20
• I am probably interested in 0.5% of all the questions on this site, but I really want to see these questions. I wish there was a way to filter out the seemingly innumerable amount of basic homework undergrad type questions, but I have yet to find a way to do this. I feel like even a crude subjective tagging system would be better than the current state of things :( – user2055 Dec 7 '13 at 0:11
• @JasonPolak: The following two setp method may improve your daily experience. 1) On the front page side bar you see headings like Favorite tags and Ignored tags. You can add all the tags you have zero interest in to the Ignored list. 2) Then you should go to your personal profile page. Click preferences. Scroll down, and check the box Hide ignored tags. Without step 2) the questions with ignored tags will simply be greyed out. With step 2) in place they get filtered out from the list of questions that you see altogether. – Jyrki Lahtonen Dec 7 '13 at 8:06
• I'm in favor of the sentiment of this suggestion. I'm not so sure about the suggested remedy though. Giving people points for voting may have the opposite from intended effect in that it will further widen the gap of points earned from advanced as opposed to elementary questions. A better remedy would be to try and recruit more advanced undergrads and beginning grad students. And also boot out those freshmen who don't take math seriously. – Jyrki Lahtonen Dec 7 '13 at 8:10
• @JyrkiLahtonen You are correct, I believe, about this A better remedy would be to try and recruit more advanced undergrads and beginning grad students. And also boot out those freshmen who don't take math seriously. That is the best solution, but it cannot be implemented programmatically within the StackExchange platform. It is the only approach that will effectively address the problem without unintended repercussions. To OP: You are correct too, in your observation. I am sorry that I cannot think of any answer that improves upon what Jyrki said. – Ellie Kesselman Dec 7 '13 at 8:48
• @Brian: I also feel that way about set theory ans general topology questions. But you have to remember that we have a handful of fanatic users (you and me amongst them) that really do their best to give good answers as soon as possible. It's not necessarily the case for most tags. – Asaf Karagila Dec 7 '13 at 9:19
• My personal experience and big sentiment is that yes, the level of Math.SE is falling. It would be great if "relatively advanced but not research level questions" get the same attention as it used to be when I started here and fell in love with Math.SE quickly. My impression is that currently, those advanced questions are competing with a flood of cheap questions dragging the attention of many active members to the much higher reputation gain they promise. In my opinion, to improve the situation, the reputation gain must somehow be reversed to give more advanced questions a higher expectation. – azimut Dec 7 '13 at 9:53
• Well, I think there might also be some concerns about the extensive attentions to questions completely irrelevant to math...... – Aiden Dec 7 '13 at 12:01
• @JyrkiLahtonen: Thank you for the suggestion. Unfortunately, I do this already. The current tag system is nearly useless for this because there is no way to filter levels in any given tag - for instance, abstract-algebra contains fairly obvious and cliche-type questions and very well-thought out questions and there is little ways to filter out between the two (for instance, #of votes might be one way but there is no way to hide all questions with less than X votes). I don't want to filter out by subject anyhow, only by some crude measure of level. – user2055 Dec 7 '13 at 21:22
• @JasonPolak You can bookmark a search query for that purpose. For example, this one shows the newest questions with the score of at least three and with no answers yet. Those may be more interesting to look at. One can tweak this in many ways, e.g., allowing for at most one answer (answers:0-1) but not an accepted answer (hasaccepted:no). – Post No Bulls Dec 7 '13 at 21:42
• @Jason: It’s worth experimenting, but I don’t know how useful the number of votes is: some of us upvote questions for displaying some reasonable (not necessarily correct) thought, independent of the nature or level of the mathematics involved. I’ve also seen interesting near-research-level questions downvoted, sometimes even below $0$, apparently because they were badly formatted and not accompanied by any indication of the asker’s thoughts. – Brian M. Scott Dec 8 '13 at 5:55
• I agree with your analysis. When I got on math.se I often check the front page, and perhaps one of the tags. If I see nothing of interest I often times log off. Even more aggressive tag searching seems futile since people often times (confusingly I may add) tag questions with seemingly random, obscure things. – Alex Youcis Dec 8 '13 at 9:12
• I'm glad someone put this discussion on the table. I have been worrying about this for a long time now. – Bruno Stonek Dec 10 '13 at 18:23
• I have since tried for a few days the method of @PostNoBills by filtering questions by votes. For the tags like abstract algebra and tags that are not specifically elementary, it seems to work pretty well and filtering out basic questions, whatever that means. So while it is not perfect, it is certainly an upgrade to looking at the unanswered tab, and gives me a fresh supply of nice questions to look at. – user2055 Dec 10 '13 at 21:24

My personal experience and big sentiment is that yes, the level of Math.SE is falling. It would be great if "relatively advanced but not research level questions" got the same attention as it used to be when I started here and fell in love with Math.SE quickly.

While it is true that they are drowned in the ever-growing flood of no-effort no-motivation solve-my-homework questions, I don't think this is the main reason.

The complete stackexchange system is based on the concept of reputation points. And it is working quite well as a motivation to make people answering questions. Certainly there are users which are not impressed by reputation points and only interested in challenging mathematical questions, but I think the majority is at least partially driven by the desire to earn many reputation points.

Cheap questions promise a significantly higher reputation gain than advanced questions. On the one hand, it is clear that people won't upvote answers they don't understand (so the number of potential upvoters is much higher), but on the other hand it's often a miracle to me where all those upvotes come from. Anyway, this is the current status quo. To see what I mean, just check the profiles of any very active user for the highest upvoted answers.

So the reputation-incentive for answering cheap questions is much higher than answering advanced questions. I have to confess that I'm not completely immune to this phenomenon myself (see these examples), and I'm convinced that most of us, being really sincere, will have to make the same confession. Hell, it just works too well being the first giving a one-sentence answer to some primitive question.

This is the main problem in my opinion. Reputation is dragging the interest away from the advanced questions. And as long as a an expert has a significantly higher income than a expert, as long as an an expert has a significantly higher income than an expert, nothing will change.

I'm convinced that the answering situation would be completely different if answering advanced questions would be rewarded significanly higher than answering cheap questions. If we want to raise the level of Math.SE, we have to find some reliable way to achieve this.

I want to address the comments of Brian M. Scott and Jyrki Lahtonen below which somewhat doubt the connection "cheap question -- higher expectation in reputation". First of all, everything I'm saying is just my (though quite strong) personal impression.

So I'm not telling that every answer to a cheap question guarantees you a boatload of reputation. But, again, I'm convinced that the expectation is significantly higher than answering an advanced question.

As a side note, I rarely answer questions on calculus. I'm more on the algebra side.

• I have to agree that you have very valid points, and that the factor you mention may be playing a major role (in addition to the factor that 'advanced questions are buried in a large pool'). What you suggest could be implemented by attaching weights to tags and calculate the reputation for answering on a question taking into account the weights associated to the tags of the question ? I have not thought much about it though.And would having guidelines about 'Which questions/answers should be voted up' help ? – user90041 Dec 7 '13 at 14:00
• The actual reputation gain from ‘cheap’ questions is often $15$, because the OP hasn’t enough reputation to upvote, or even $0$, because the OP doesn’t know about accepting. – Brian M. Scott Dec 7 '13 at 19:41
• Azimut, you have valid points. I used to think a lot more like you, but then I did some digging. Largely because I felt that calculus is the worst offending tag in this respect. But, if you take a look at @Brian's profile (as an example), you will see that he has 800 zero score answers (many of them accepted though). He earns an average of 3 upvotes per calculus or topology answer, less than that in combinatorics. I know, I know. For Brian anything less than 300 points is a bad day, and for me anything above 80 is a great day. But the difference comes from somewhere else :-) – Jyrki Lahtonen Dec 8 '13 at 8:16
• The thing is that the blockbuster questions where answers gain dozens if not hundreds of upvotes occur only at the level of calculus/elementaryNT and below. And the vote counts on those make me shake my head in disbelief. But those are rare, and it is easy to overestimate their significance due to their high visibility. Somebody suggested that those upvotes come from bored SO-users, who click the "hot list". Bored SO-users cannot appreciate the finer points of math, but seem to be able to appreciate somebody cranking out a tough integral or calculate the remainder of a 6000 digit integer. – Jyrki Lahtonen Dec 8 '13 at 8:28
• @BrianM.Scott: That's certainly true. But it doesn't contradict my impression that the expected reptuation for for answering cheap questions is much higher. I've added a few lines to my answer on this. – azimut Dec 8 '13 at 14:03
• @JyrkiLahtonen: Many thanks for your comments! Your comment on "blockbuster" questions is certainly true, but I think the problem already starts on lower levels of upvotes (in each single case to a lesser extend, but much more frequently) I've added a few lines to my answer to explain how I came to that opinion. – azimut Dec 8 '13 at 14:08
• Now that answer has seven (LOL). The length of the answer plays a huge role. Anything longer than three lines apparently triggers some people's TL;DR; gland. – Jyrki Lahtonen Dec 8 '13 at 14:56
• @JyrkiLahtonen: So this is again evidence of: Early and pregnant answer to an easy question, a bit advertising on meta -> quick reputation :-) – azimut Dec 8 '13 at 16:16
• As opposed to making it to the hot list frequented by bored SOers?. Don't click if you want to hang on to the illusion that this site fairly values sage and useful answers. – Jyrki Lahtonen Dec 8 '13 at 20:21
• @JyrkiLahtonen: Ouch! – azimut Dec 8 '13 at 20:38
• @JyrkiLahtonen That's awful but this may be even worse. OP answered his own question. Maybe you should have migrated it over to StackOverflow when there was still time.... – Ellie Kesselman Dec 9 '13 at 20:19

The current active front page is probably not attractive enough for someone visiting from MO to stay. Showing them the interesting tab by default would be an improvement, but SE declined to do so. I do not believe for a moment that SE will want to develop a new tab for questions that are user-rated as "advanced".

I do not agree with the suggestion by azimut that the reputation system is at fault, i.e., that expert users would answer more advanced questions if they were better paid in fake internet points. By and large, such users do not care about fake internet points. They may care a bit about votes, which indicate that someone read the post and benefited from it. If many of a user's answers remain without votes or comments, they may feel like their time is being wasted, much like preparing a seminar talk to which nobody shows up. Giving additional fake points per vote does not help here: they still come from the same number of voters.

It may be reasonable to stratify some tags, which was done for number theory and set theory (which have the corresponding elementary- counterparts). E.g., has a lot of undergraduate computational questions among which the advanced ones easily get lost. Maybe something like complex-analysis-theory can be used for more conceptual questions. But I'm not even 50% sure that stratification would do more good than harm.

• There is a complex-numbers tag, isn't there, for questions like "express $1/(2+3i)$ in the form $a+bi$". I'm pretty sure I've taken more than a few question tagged complex-analysis and retagged them complex-numbers. – Gerry Myerson Dec 8 '13 at 3:05
• @GerryMyerson I appreciate your effort. But this is not what I had in mind. I meant the difference between "calculate the integral $\int_{|z|=3} \frac{1}{z^2+z}\,dz$ using residues" and "is there a holomorphic map from disk to polydisk whose boundary cluster set contains the distinguished boundary?". – Post No Bulls Dec 8 '13 at 3:17
• Ah. OK, we're on the same page now. Anyway, I'm not sure how effective the stratification has been in, say, number-theory, where many elementary questions still have the number-theory tag, and many questions have both tags. – Gerry Myerson Dec 8 '13 at 3:21

To add my two cents: Why are there fewer questions and answers at the advanced level? Because yes, there are more beginner users coming in, hoping to get homework insight/answers. Is this a bad thing? Not really. The added traffic makes the site more valuable, thus sustaining its existence for those who want to do more cool stuff on it.

Seriously, if you want to halt the number of repeated, beginner questions that do not add the the information-value of the site, then do a little due diligence in response to an easy question and find a page within the site where the answer inevitably must lie, link to it, and then close the question. Yes, we have rep hounds who will answer the question anyway (guilty as charged). But still, perhaps an anti-rep-hound will be equally charged to find such links and close out such questions. When we do this, then perhaps we increase the advanced-to-beginner question ratio, while maintaining the traffic level. Everyone wins.

By the way: as one who attempts to answer only where I bring value, I do not feel that all computational problems should be thrown into the same useless-undergrad vat that was offered as an example by @PostNoBills. Some of the examples I have worked out involve intricate tinkering that would never be asked of a standard undergrad. Rarely are they even asked of a grad for that matter. Maybe the folks at MO would not appreciate them, but it doesn't mean they do not add novelty to the site.

• I offer two thoughts against your call for anti-rep hounds. First, there is not nearly the same incentive for rep hounds and anti-rep hounds. Second, I've found myself to be terrible at using the search function on MSE to find a post I know exists. I'm sure googling well would help, but for me the effort involved in this is also very far out-of-balance with what it takes to answer said question. – Antonio Vargas Dec 8 '13 at 16:57
• @AntonioVargas: maybe, but how hard can it be? I have done this with a few egregious examples. The ones where you know they have to be here somewhere. A link to that answer usually deflates the quick rep-hound, that's all. I know because I have been one and have been so deflated. Of course, then, as Jyriki notes, there are some users who simply do not care and answer anyway. But at least we have tried. – Ron Gordon Dec 8 '13 at 17:03
• By the way, I do not mean that out-of-site links (e.g., Wikipedia) are an acceptable way to answer the question. (Many rep hounds use these in order to put a placeholder in a question that is perceived to be an attention-getter, but where they do not have the skills to provide a thorough answer.) I mean, exact question, in-site, with good answers. – Ron Gordon Dec 8 '13 at 17:06
• Probably my last sentence was an exaggeration. At least it requires more "effort" because it is not explicitly fun like doing math. – Antonio Vargas Dec 8 '13 at 17:09
• @AntonioVargas: and I maybe placed too much faith in my feeling that it would be fun to deflate a rep hound. Sometimes I can be a little cruel. – Ron Gordon Dec 8 '13 at 17:11
• I disagree, strongly, with the premise that "more traffic" is equivalent to "better". – user14972 Dec 8 '13 at 20:10
• @Hurkyl: I agree that some traffic is preferable to others. In an ideal world, we could close off the site to those who hate math and are only here to cheat. But I think that the people who financially back this site do want more traffic and I imagine measure how the site is growing. I am also sure they do care about the quality of the site content, but getting substantial clicks matter for advertising and for keeping the operation of our beloved site financially viable. Thus, we take the good with the bad, so we can enjoy the good. – Ron Gordon Dec 8 '13 at 20:27
• @Ron: We can't enjoy the good if we're filled up on bad. – user14972 Dec 8 '13 at 22:14

Why not go ahead and add meta-tags for the question level? E.g.

• soft-question
• recreational-math
• high-school
• undergraduate
• graduate
• research-applied
• research-math

Then people could not only filter out questions at levels they are not interested in, but it also gives answerers an indication of how much familiarity they should assume from the OP.

I think bounties can be one answer. Many bounties are for crazy questions, but users interested in increasing visibility of hard questions can place bounties on all interesting questions. Most of the users complaining about question quality also claim they don't care about reputation, so this solves both problems.

• One problem with bounties is that they tend to attract half-baked answers from professional bounty hunters. Besides being slightly annoying, this reduces the question's visibility in the long run (since there is an answer posted). – Post No Bulls Dec 8 '13 at 3:29
• I haven't studied bounties closely, but a better use of bounties may be to offer them afterwards to good answers. Jonas Meyer is approaching the 10000 point figure in offered bounties. Byron Schmuland is over 5000. Some others are around 2000. I am planning on increasing this activity myself, but have a lot of catching up to do. Of course, there are some high rep users who only answer and, judging from their voting profile, hardly read answers by others (or at least don't upvote them). I don't think this says anything bad really. Just a very different approach to the site I suppose. – Jyrki Lahtonen Dec 8 '13 at 7:58
• @Jyrki: I admit that my focus is on providing answers, though I do try to remember to upvote things that I think deserve it. I do most of my upvoting of answers when I’ve just come back online and am catching up, or when I come back to a question that I parked and find that someone’s answered it well; when I’m in answer mode I mostly don’t see other answers. I pay no attention to bounties (in fact I barely notice them), but I gather that many do, and I agree with you that ex post facto bounties might well be better. – Brian M. Scott Dec 9 '13 at 4:13

What about having votes not having weight simply 1 but some kind of strictly increasing function of the voter's reputation? Then those with higher reputation will have a greater say in which questions are interesting or which answers are good. It will of course be biased towards their subjective opinion of what is interesting or good, but at the least it will result in homework questions and answers to them appropriately (lower) valued.

• Hmm... To the downvoter, you should provide a reason why this is not good, because I can't see anything really wrong with my suggestion, except for what I stated... Unless of course this site is meant for getting homework done, which would be quite unfortunate. – user21820 Dec 8 '13 at 9:33
• You should know that votes have slightly different meaning on meta, they simply express agreement/disagreement: math.stackexchange.com/help/whats-meta – Martin Sleziak Dec 8 '13 at 12:26
• Some feature requests on meta.SO, which are similar to your suggestion: meta.stackexchange.com/questions/54494/…, meta.stackexchange.com/questions/75378/…, meta.stackexchange.com/questions/57278/… – Martin Sleziak Dec 8 '13 at 12:27
• I am not sure whether your suggestion would move bias away from homework question. I personally vote for homework question, if the OP has shown some effort, and I also vote for answers on such question, if the answerers explained things clearly and the answer seems to be helpful. – Martin Sleziak Dec 8 '13 at 12:28
• @Martin Oh I see, yes I forgot about the meaning of the votes so thanks for reminding me. I also agree that votes should be given to homework questions that are illuminating, interesting or have some value to future readers, but this would at least remove the (in my opinion) undue weight given to routine homework problems and their solutions. As others mentioned, bounties do not solve the problem but in fact create problems of their own, since it makes answering very incentive-driven instead of interest-driven. That is the main reason for my suggestion. – user21820 Dec 9 '13 at 9:09
• And it occurs to me that those who use math.stackexchange often to get homework done will of course downvote (disagree) with my suggestion, and if so I won't be surprised if it should go well below 0. =) – user21820 Dec 9 '13 at 9:11
• I doubt that people who use math.SE just to have their homework solved by others would come to meta (and downvote your post). My guess is that typical visitors of meta are users who spend lot of time on MSE. – Martin Sleziak Dec 9 '13 at 13:45
• In that case, what makes people prefer bounties than weighted votes? – user21820 Dec 11 '13 at 2:51
• In fact, this suggestion is intended to deal with the problems that Azimut has described in detail, so it doesn't make sense to me for people to agree with him but disagree with this or other proposals aimed at that, unless the problem is actually what the majority desires to remain. – user21820 Dec 11 '13 at 3:00
• @user21820: It makes perfectly good sense if someone thinks that the proposed remedy is either unlikely to be effective or worse than the disease. – Brian M. Scott Dec 12 '13 at 1:01