I feel that MathOverflow is unique in the StackExchange network. There are features which we might want which might be unique to this site, and which might not be immediate feature requests or requests that StackExchange would want to implement globally or even locally. There isn't currently a repository for such ideas.

I think that it would be useful to have a big list of feature requests and suggestions for some fantasy future MO 3.0, to serve as feedback data to MO and to SE, and to serve as an incubator for possible potential future feature suggestions. Having these voted up or down by users adds information, and comments on these are also useful.

Big list: Please suggest features, and make specific suggestions, for things you would want to see on a fantasy future MO 3.0. These don't have to be immediately implementable or likely to be implemented. One fantasy request per answer, please. Please try to avoid duplicate answers.

I'll start the ball rolling by inserting some sample answers. Please comment and vote up or down depending on whether these seem like good ideas to you, in the abstract.

I think the longer this list is, the more informative it is.


Note that posting an answer here is not the proper way to make a feature request. If the feature you want is possibly implementable now, post it as a separate meta-question with the tag .

  • Daniel, I added a notice in the hope that users would first try a feature-request before posting here. Maybe you can think of a better wording for your question. Ideally, this place should be a graveyard for well intended feature requests that get declined for feasibility reasons. Instead, it's attracting new feature requests that never get seen by the people who can determine whether they are feasible. – François G. Dorais Mar 14 '14 at 19:14

29 Answers 29

Restore +10 reputation for questions.

A good question might actually be more important than a good answer, in mathematics and in life. Having them count for less than answers decreases the relative motivation to write detailed and coherent questions, and I suspect that it lowers the overall quality of questions on MO. On Fantasy MO 3.0, I would return +10 reputation to questions.

  • 2
    I could not agree with this more. It is ridiculous when I get +10 each time someone votes for this mathoverflow.net/questions/141501/… but only +5 when someone votes for this: mathoverflow.net/questions/104059/… – Vidit Nanda Oct 18 '13 at 2:35
  • 5
    I would say: yes, but only together with some measures to avoid excessively rewarding people who only ask lots and lots of questions, but provide almost no answers to other people's questions. – Stefan Kohl Oct 18 '13 at 9:52
  • 10
    I am against changing back. An issue got raised by @StefanKohl. One might consider more subtle models of point distribution according to votes, which then might be better, but just as is I think the current way of distributing is relatively better than the one we had. Yes, really good questions are important, but more often than not detailed and coherent (mathematical) question do not get that many votes regardless, at least if they are somewhat specialised. By contrast, some general interest question that are not even that well-thought out get many votes. – user9072 Oct 18 '13 at 10:35
  • 15
    I agree with @quid. On MO 1.0, there were many users who got to 3000 rep by just asking semi-soft questions. This I think led to many open-close wars on soft questions. I think we see less now because of that. In fact the +10 encouraged people to ask very soft questions to try and get rep before somebody forced them to make the question CW. Of course rep is silly but the point is 3000 rep gives the power to open and close questions and if people get this power without every answering a question, this can lead to a change in the nature of the site. – Benjamin Steinberg Oct 18 '13 at 14:46
  • @BenjaminSteinberg under 2.0 the moderators are the only ones who can make a (semi)soft question CW, so the only way your fears can get realized is if there are lapses in moderation. I think this is unlikely since the mods are super-quick. – Vidit Nanda Oct 18 '13 at 17:13
  • 4
    @Vidit Nanda, I don't get your point. In a sense things are now worse. Flagging the mods for CW was possible in MO 1.0 too. But often people were pressured to add the CW themselves. – Benjamin Steinberg Oct 18 '13 at 17:48
  • @BenjaminSteinberg I suppose we things differently. In the earlier system the responsibility of making soft questions CW was shared between moderators (whose judgment I basically trusted) and askers (whom I didn't always trust to make question CW). Now, the power to do the right thing rests solely in the hands of people I trust, which seems like a definite improvement. – Vidit Nanda Oct 18 '13 at 19:26
  • 22
    While I agree that great questions are extremely valuable, they are also extremely rare, and changing to +10 for all questions would have much more impact on not-so-great questions. My impression is that, in most cases, much more effort went into answers than into questions; and it seems reasonable to make the rep awards reflect this. (I turned an earlier version of this comment into a separate answer to this question.) – Michael Zieve Oct 19 '13 at 0:24
  • 7
    I feel like this problem would be solved if each time the upvoter could choose from 2 types of upvote; one is worth +5 innernetz and the other +10. – Yuichiro Fujiwara Oct 19 '13 at 2:15
  • 2
    @MichaelZieve Do you have any evidence for the rarity of great questions or the claim that it takes less to ask a good question than to give a good answer? A huge percentage of the answers here, mine included, are "look at so-and-so's paper". – Vidit Nanda Oct 19 '13 at 20:52
  • 1
    @ViditNanda: I have no idea what you would consider to be "evidence for the rarity of great questions", since obviously what is a "great question" is subjective. Of course I am stating my observation. I have not seen many answers of the form "look at so-and-so's paper", although the percentage of answers like that could indeed be measured (and perhaps you could tell us what it is, since you say it is "huge", which sounds incorrect). Most answers I see are significantly more detailed, more original, and more thought-out than most questions I see. – Michael Zieve Oct 19 '13 at 23:46
  • @ViditNanda: I thought of a way to provide the evidence you requested. I checked the votes made by the 24 users who have earned the most points this year. All of them have voted up much more than they voted down. 18 of them have voted on answers more than on questions (usually by a huge margin). For instance, you have voted 338 times on answers and 206 times on questions. So apparently you also believe that great answers are more common than great questions. – Michael Zieve Oct 20 '13 at 0:05
  • 3
    @MichaelZieve That evidence would be perfect if there were as many questions as answers on this site. In particular, I'm sure you have often voted up multiple answers to the same question whereas it is not often that one encounters different questions to the same answer. – Vidit Nanda Oct 20 '13 at 0:41
  • @ViditNanda: First, although you are sure that I "have often voted up multiple answers to the same question", you are wrong. Second, is there any type of evidence which would satisfy you? Third, you made a testable (and probably false) claim about a "huge percentage of the answers here"; do I infer correctly that you are not willing or able to provide evidence for that claim, even though you made this claim in the same comment in which you requested evidence for my assertion of what "my impression" is? – Michael Zieve Oct 20 '13 at 11:38
  • 5
    The lazy question mathoverflow.net/posts/106560/revisions (please read the original version) has 156 votes. It is now CW but the OP amassed a lot of rep before that happened. The question was also rewritten stone soup style. The answers which are also highly upvotes required a lot of thought. – Benjamin Steinberg Oct 23 '13 at 15:34

Increase the allowed length of comments.

I am running of space more often than not and have to either castrate both my thought and my language in such cases or make a chain of comments instead of a single one. Posting an answer instead is often not an option because the comments are tied to the post or even to the previous comments and may be quite perplexing if displayed as separate entities.

  • 7
    I think this is particularly important on meta. A length limit of a reasonable degree may make some sense on the main site if it's to deflect undesirably lengthy discussion, although, as fedja says, the current character limit seems too severe to me. But it makes no sense on meta when it's the place for discussing important issues about the main site. – Yuichiro Fujiwara Oct 19 '13 at 14:27
  • 3
    I certainly agree about meta. I like even more Scott's suggestion: meta.mathoverflow.net/a/1059/2926 – Todd Trimble Oct 20 '13 at 15:30
  • 2
    I agree on meta, but not on the main site. Long comments should be answers. – Daniel Miller Oct 24 '13 at 13:46
  • 3
    @Daniel Miler The difference between the comment and the answer is not in the number of symbols but in the general logic and the relevance to the previous discussion. Sometimes I can give a full answer in 3 words and need a page to make a comment. However, as usual, this suggestion is open for discussion :-). – fedja Oct 24 '13 at 14:45
  • 2
    On the main site, I actually find that the effort needed to simplify a technical comment so that it fits within the 500 character constraint is often helpful for the clarity of the post. – Emil Jeřábek Oct 24 '13 at 15:03
  • @Emil: Minor correction: the limit is actually 600 characters. – Stefan Kohl Oct 24 '13 at 19:52
  • I think Brian consistently shows us how much squeezing can be done to fit lengthy comments in a short space (mathoverflow.net/users/3927/…). :-) – LSpice Aug 29 '14 at 16:10

Minor edits

See this post of Scott Morrison :

<< Can we have a "minor edit" checkbox in the edit interface, along with the parenthetical text "minor edits do not bump posts on the list of active questions, but are subject to review but another user"? >>

Change the term "reputation" to something more neutral, e.g. "points" or "score"

I can't help myself, but I feel that having a numerical "reputation" value attached to oneself is in a certain sense disreputable in itself, enough to possibly scare off potential users. -- Maybe this is a subtle issue of use of language, or reflects some cultural difference (Europe vs. US?), but I don't know. In any case, renaming the thing to something more neutral like e.g. just "points" or "score" would remove this concern.

I wonder whether I am the only one who feels this way, and what other people think about this.

  • 9
    I like "karma" best. I think it expresses the purpose of the system most accurately. – Tom Leinster Oct 18 '13 at 11:51
  • 6
    On meta we have sorting of users by "participation"; I think this term could just be adopted for main, too. – user9072 Oct 18 '13 at 12:55
  • 7
    @Tom: "karma" is not a neutral term. -- I would avoid using terms related to religion or spirituality, for the obvious reasons. – Stefan Kohl Oct 18 '13 at 12:59
  • 1
    @quid: I think "participation" is also possible -- though using that term might have the drawback that one might argue that disruptive participation is also "participation" in a sense ... . – Stefan Kohl Oct 18 '13 at 13:12
  • 4
    Yes, but positive participation gives positive participation points and negative participations gives negative participation points. But perhaps "contribution" would have a more positive connotation than "participation". – user9072 Oct 18 '13 at 13:18
  • 15
    @quid: I think "contribution" sounds fine -- it describes what is meant quite well, and it is neutral in that it seems unlikely that it causes strong objections. – Stefan Kohl Oct 18 '13 at 13:23
  • 2
    @Stefan, I see your point, though I think it's one of those words (like "heaven") that's acquired a life of its own quite removed from its religious origins. Anyway, "contribution" seems a good substitute - I'd vote for that. – Tom Leinster Oct 18 '13 at 13:47
  • 4
    I see no reason why we need an extra word like "reputation" or "participation" attached to "points" or "score" at all. We can be free to attach whatever interpretation we want to the numerical value. – S. Carnahan Oct 19 '13 at 0:22
  • 2
    @S.Carnahan Perhaps this is anyway what you mean, but, at the moment the tab there at mathoverflow.net/users is labeled "reputation" so an interpretation is given and there is thus (in some peoples opinion) need for changing this. – user9072 Oct 19 '13 at 9:49
  • 13
    I hate to say it, but this reputation does translate to some extent to a real world reputation. I went to three conferences by now, and despite the fact that I have never proved anything significant so far (my masters, while interesting is not even close to being really difficult), still people knew my name and had questions for me. If that's not reputation, then I don't know what is. :-) – Asaf Karagila Oct 19 '13 at 16:20
  • 2
    @quid: It's not only the tab you mention -- more importantly, on one's user profile page below one's name and photo, it says "xyz reputation". I wonder what people who do not use MO think about this when they google one's user profile page. – Stefan Kohl Oct 19 '13 at 21:16
  • 1
    @Asaf: Of course as everything you do, also what you post on MathOverflow contributes to how people see you. But what the points on this site measure is just a rough expression of how valuable people find your contributions to this site, and not more. – Stefan Kohl Oct 24 '13 at 9:26

Allow reopen votes and close votes to cancel each other.

If the difference between close votes and keep open votes does not reach 5 the question remains open. Likewise a closed question will remain so as long as the difference between reopen and keep closed votes does not reach 5.

I think this might help with reducing the amount closing/reopening/reclosing/... of controversial questions.

  • 1
    An issue with this is that the decision then can depend purely on timing. Open/close of a question could depend on whether, say, the fifth close vote arives (slightly) before the first reopen vote or (slightly) after. An additional detail would also be if one could cast, say, a reopen vote before a close vote was even cast or a second reopen when there is only one close. (Also, I got briefly confused why you say for reopening the difference should reach 5, but then figured that likely you mean that upon closure the counters are reset to 0.) – user9072 Oct 18 '13 at 23:00
  • 3
    @quid, yes, that is possible, we can not wait for everyone who wants to cast a vote to cast a vote. But it won't be worst that the current one and in many cases it can be better as those who are against closing or reopening do not need to wait till the question is closed/reopened to cast their votes. – Kaveh Oct 18 '13 at 23:41
  • I still find it a bit strange tu have this depend on timing, but granted in practice this would not be that significant. However, what about the other points can one pre-emptively vote or not? That one cannot would go a bit against the spirit of the propsal (of no need to wait to cast the votes) yet if one could I am worried this will blur the distinction between up/down and open/close voting and in the end cause people to cast open/close votes with less consideration. – user9072 Oct 19 '13 at 7:03
  • 4
    Perhaps there should be more weight for reopen votes than for closed votes. For example, each reopen vote cancelling 2 close votes. – Daniel Moskovich Oct 19 '13 at 13:09
  • 5
    @Daniel, not sure about the exact numbers but I think the idea (if I understand it) makes sense: if a significant portion (not necessarily majority) of voters think a question should be open then I think it makes sense to keep it open. – Kaveh Oct 19 '13 at 14:09
  • 1
    @DanielMoskovich (and Kaveh): but please also note that a bias towards open is already present (and this proposoal will make this still somewhat stronger), under the proposoal one needs five votes to close more than there are votes to reopen to get it closed, and one only needs to get back to the same number to get it reopened (And under the current system it is like that, up to "rounding errors".) And, since OPs can vote on their own question (and with a significantly lower points-limit) there is still some more bias towards open. – user9072 Oct 20 '13 at 11:59

Restore "No longer relevant" as a reason for question closure

It happens sometimes that a question turns out to be naïve and is immediately answered in comments, or the OP notices a mistake in the question. Alternatively, the question is asking for a reference to XYZ, and in the interim a paper on XYZ is posted to ArXiv or a difficult-to-find reference is posted online.

In all of the above situations and more, I would like to be able to vote to close a question as "no longer relevant", but this no longer exists as a default closure reason in MO 2.0. I would propose to restore in in Fantasy MO 3.0.

  • This could (at least in theory) be set up as one of the custom off-topic close reasons. – Adam Lear Oct 21 '13 at 3:53
  • It often is! But it's different from being "off topic", and I think it's common enough to request that it be brought back... I wonder if there are any statistics on this? – Daniel Moskovich Oct 21 '13 at 5:37
  • What sort of statistics are you looking for? – Adam Lear Oct 21 '13 at 5:58
  • Statistics to test the hypothesis that "no longer relevant" is a desirable standard closure reason: How common it is for people to list "no longer relevant" or a variant thereof as a custom off-topic close reason (what I mean by "variant" is imprecise- things like "has been answered in the comments" would be included, for example). Actually, an interesting statistic would be a list of all custom off-topic close reasons which have been used, ranked by number of times used! – Daniel Moskovich Oct 21 '13 at 6:21
  • Sorry it took me a few days to put this together... Here are all the written-in close reasons that were supplied for closed questions since this new system was rolled out: pastebin.com/8QZ2Xi9c. And the overall off-topic usage breakdown: pastebin.com/aEgRa7kH. – Adam Lear Oct 29 '13 at 23:08

When moderators make a question CW, by default they remove all rep earned from the question

There are many soft questions which quickly earn enormous rep for both the question and its answers before the moderators mark the question CW. In most of these cases, if the moderators had noticed the question earlier then it would have been CW'ed earlier, and no rep would have been earned. So why not set things up so that when a moderator CW's a question, he/she has the option to remove all rep earned so far by both the question and its answers? I think this should be the default setting. There might be situations when a moderator CW's a question and for whatever reason does not want to affect rep earned so far, so that option should exist as well.

ADDED LATER: In response to Todd Trimble's comment that users would find it annoying to lose rep: one could avoid giving them this rep in the first place. Have a button so that a user can flag a post for moderator attention while explicitly giving the reason that the post appears to merit the Community Wiki tag (maybe only allow users having at least a certain rep to flag posts in this way). Once a post has been flagged in this way, have some notice appear below the post to say that the post is awaiting moderator decision on whether it should be Community Wiki. After a post has been flagged in this way, record all subsequent votes for the question or its answers, but do not apply these votes to change anyone's reputation until the moderators have decided whether or not to make the post CW. Keep track of the votes, and if the moderators decide the post is not CW then assign the corresponding rep.

I think this would work because it often only takes a minute or two (or less!) for someone to post a comment saying a post should be CW.

  • Yeah, that would generally be great. – Asaf Karagila Oct 19 '13 at 16:18
  • 9
    Generally speaking, I think policies where people suddenly lose rep through no fault of their own could prove very annoying to some. Better I think would be a mechanism where a questioner who intends to have a question be CW have that question sitting in a holding pen until a moderator can convert to CW, where it would then be opened. – Todd Trimble Oct 20 '13 at 15:26
  • 1
    But people would only be losing rep which never should have been earned in the first place, so mods could give a very reasonable explanation if any user later complained about losing such rep. I have seen several soft questions which are asked by people unfamiliar with the site who don't know about CW, but which people familiar with the site immediately recognize as meriting the CW tag. When moderators switch these to CW, assuming that the moderators agree that these are questions which should not have earned any rep, then it seems reasonable to remove rep which shouldn't have been earned. – Michael Zieve Oct 20 '13 at 23:27
  • 7
    If a call for CW could trigger a (possibly temporary) hold on rep, then that would make me feel a lot better about the proposal. It's the subtraction of rep that I don't sit easy with. I don't know why exactly; something psychological about it. – Todd Trimble Oct 21 '13 at 0:56
  • @Todd Trimble: is your objection to the fact of users losing rep, or to the fact that it might not be clear to them why they lost rep? I think the latter happens already when one has gotten votes from an account which is later closed. In both cases, rather than simply having the awarded points disappear from a user's account, one could keep them in the user's detailed list of rep changes, but add a corresponding negative amount with the explanation that the points from a specific CW'ed question or from a closed account have been removed. – Michael Zieve Oct 21 '13 at 6:39
  • 1
    @MichaelZieve Probably more the first. I might explain it by pointing to an analogous phenomenon whose usual name I don't recall, but I'll call it "irrational risk aversion": the sort of situation where if someone either offers you 20 dollars, or offers you a chance to bet 100 dollars with a 65% chance of winning in your favor, you'd still go with the sure deal. By that analogy, losing hard-earned points after having won them is felt more acutely than never having gained them in the first place. It might be irrational, but then you could say that the whole allure of the point system is too! – Todd Trimble Oct 21 '13 at 7:39
  • @ToddTrimble Loss aversion could be the term you are looking for. Risk aversion, as given in your example can be a consequence of it. But the former seems more pertinent as the issue is here is really mainly about loosing something not so much about deciding to take a risk. // Generally, I upvoted the proposal since I think it would be good (in an abstarct sense) and also fairer in some way. However, it could likely be the case that it creates many issues/conflicts/hard feelings and is better not done. – user9072 Oct 21 '13 at 11:04
  • @quid Thanks -- I was too tired to hunt down the exact term when I last wrote. – Todd Trimble Oct 21 '13 at 18:10
  • I would support letting people gain reputation points also on CW answers and questions. In any case, if we want to adopt the Michael's proposal we should, in my opinion, make the rules for CW questions stricter. It will be unfortunate if questions regarding application of mathematics, mathematical modeling, mathematical education, or relations with other areas, will automatically become CW. – Gil Kalai Oct 26 '13 at 18:45
  • @Gil: I think you're addressing a different issue. My objection is that, currently, earning points for eventually-CW'ed questions and answers isn't based entirely on votes, instead it is also based on how long it takes the moderators to notice the question and mark it CW. It doesn't make sense to me that the moderators' sleep schedules (for instance) should play a role in determining how many points someone gets for a question or answer. – Michael Zieve Oct 26 '13 at 22:53
  • Michael, in this case the fair solution would be that answers given to a non CW question continue gain reputation also after the question has become a CW. Your earning-points expectations is based on the status of the question when you answer it. – Gil Kalai Oct 29 '13 at 5:41

Separate points into Question score and Answer score

Why not have two reputation scores in Fantasy MO 3.0, a "question reputation score" and an "answer reputation score"? Then the rep levels required for various privileges can be in terms of some linear combination of the two rep scores.

The two ways of earning rep seem very different from each other, so why not separate them?

  • 1
    I took the liberty of changing 'rep' to 'score' as per another request. (Another option is 'points', if you prefer.) – François G. Dorais Oct 19 '13 at 0:31
  • It would be interesting to call them "theory-builder score" and "problem-solver score". – Federico Poloni Jan 9 '16 at 6:26

This is a trivial request but it can be helpful: allow to order answers to questions by "newest" and not only by "oldest" (maybe it is possible now?)

  • 2
    If anyone has missed the relevant thread, here's the same request on MO's meta: meta.mathoverflow.net/questions/140/sort-by-newest-option – Yuichiro Fujiwara Oct 20 '13 at 6:17
  • I'd have nothing against a fourth tab 'newest', but I'd not want to replace the 'active' tab (nor the 'oldest'), as I find 'active' more useful for (old) questions with many answers and it is quite close to 'newest' in typical circumstance. Or put differently, the main (only?, but perhaps I miss something) usecase for 'newest' I can see is to find new ans easily in a q I already read before, but then typically I am also interested in modifications to old ans so that 'active' is better, and even if not there are often not that many edits that one cannot easily ignore those ans when browsing. – user9072 Oct 20 '13 at 12:07
  • hmm, yes, actually "active" has a rather similar function. (Anyway if at all then to add not replace.) – Gil Kalai Oct 20 '13 at 15:54

Make it possible to flag errors

Presently the only ways to point out an error in a post are adding a comment and editing the post. However comments are not very visible in particular if a post has already many of them. Editing the post does not have this drawback, but it still may not be what one really wants, in particular if one does not know how to correct the error.

I think being able to set an "error flag" instead would often be a better alternative. Such an error flag may show up as a warning sign on the left margin below the voting arrows. A description of the error could be displayed when clicking on or moving the mouse above that sign. I suggest to make an error flag bump the post like an edit, and also generate notifications to the same people as an edit (plus the OP of the question, if an answer is flagged).

This proposal is motivated by the thread Dealing with possible incorrectness.

  • 1
    Interesting idea. As a side-effect it could also help to focus/simplify discussion of what (if any) is the use/meaning of a down-vote. – user9072 Jan 2 '14 at 23:03
  • 1
    Something like this partially exists, see meta.mathoverflow.net/questions/189 and meta.stackexchange.com/a/191984/147496 – François G. Dorais Jan 3 '14 at 0:41
  • Also, Stefan, is there are reason you don't believe this could be done now? – François G. Dorais Jan 3 '14 at 20:08
  • @Francois: I don't know. -- Implementing such error flags would need some programming -- not much, if the SE software is implemented in a reasonable way -- but it would likely be some more work than realizing the few presently highest-voted proposals in this thread. – Stefan Kohl Jan 3 '14 at 21:15
  • Maybe a partial implementation would be more doable? – François G. Dorais Jan 3 '14 at 21:28
  • @Francois: I think all this depends on what SE people can be convinced to implement -- and with that you surely have more experience than me. – Stefan Kohl Jan 3 '14 at 21:31
  • I think you should also post a formal feature request (perhaps with only some parts) so that SE staff have the opportunity to notice this. – François G. Dorais Jan 3 '14 at 21:37
  • 1
    @Francois: O.k. -- I have posted this as a feature request here: meta.mathoverflow.net/questions/1356. -- Feel free to adjust the proposal if you think you can make it more likely to be implemented if you remove or change certain parts. – Stefan Kohl Jan 3 '14 at 22:35
  • 1
    Thanks Stefan. I just upvoted your request and I will try to make sure it gets the attention it deserves. It might not get implemented exactly as you ask but it is only by asking that progress can be made. – François G. Dorais Jan 3 '14 at 23:16

Let users opt out of reputation

This was the topic of a thread by Francois Dorais, and is discussed there: Opting out of reputation

To quote:

Indeed, some users have resorted to various schemes to avoid accruing reputation on MO such as answering in comments, answering in community wiki mode, discarding points through bounties, etc.

There are a few obvious caveats that pop to mind immediately:

•Since user reputation is tied to their ability to participate in the site, this option mostly makes sense after a certain minimum threshold is reached.

•This should not stop others from voting on posts from users who have opted out since that serves other purposes.

•Since it is undesirable to have users opt out and back in, the choice to opt out would probably have to be irreversible.

Could such an opt-out work on MO?
  • 8
    You can "opt out" from reputation by making your answers CW, so the only trouble we'd ever see is when asking questions (and still wanting to reward good answers). While I'm all in favor of +10 instead of +5 for upvoting questions, anyone who has considerably more questions than answers probably should be subject to the reputation system. – Vidit Nanda Oct 18 '13 at 2:50
  • Still, it would be nice to have a feature where any user could opt out of reputation, and that user would cease to have reputation displayed on their user page or beside their avatar. – Jason Polak Oct 20 '13 at 21:23

Commutative diagrams

It would be wonderful if there was some way of adding support for xypic to mathoverflow.

  • 3
    There is a support for AMScd, though. – Asaf Karagila Oct 24 '13 at 14:54
  • 5
    One way to do that is to pressure the XyJax developers to propose a reliable support plan for their product. meta.mathoverflow.net/q/988 – François G. Dorais Oct 24 '13 at 15:03
  • @AsafKaragila: I hadn't known that.. but the flexibility of xypic would be even better. – Daniel Miller Oct 24 '13 at 22:26
  • Of course. You should read the link in the first comment [written by me,] to the question linked by @François. – Asaf Karagila Oct 24 '13 at 22:27
  • 7
    Even better- tikz-cd ! – Daniel Moskovich Oct 25 '13 at 4:35
  • @AsafKaragila: I just read François' answer. That is really too bad. – Daniel Miller Oct 25 '13 at 14:13

Editable comments

Since comments are used a bit differently on MO I think it is reasonable to allow a bit more editing.

  • 13
    Maybe edible comments too? Many grad students might be using the site, and as grad students they are always on the look for free food. If the comments are edible, it means that they can easily find food! – Asaf Karagila Oct 23 '13 at 10:46
  • 8
    Please make a somewhat precise suggestion. What is a bit more? Just increase the 5 minutes to something longer or indefinitely? Editing also by others (as for posts)? Revision history? – user9072 Oct 23 '13 at 11:10
  • 1
    @quid, I think 5min limit is too short. How much more freedom should we have in editing our own comments needs to be discussed, but if you ask me I think there shouldn't be a time limit. Since these are suggestions for fantasy MO3.0 I don't think they needs to be that precise. This suggested feature is about people being about to edit their own comments. – Kaveh Oct 23 '13 at 18:55
  • 1
    Thanks for the reply. I agree they need not be "that precise" this is why I only asked for "somewhat precise" ;-) – user9072 Oct 23 '13 at 21:05

Having an incubator thread on Meta for soft questions

This was suggested by Noah Snyder and is discussed there: Crazy idea about soft questions

To quote:

I was thinking the other day about big list and other soft questions and had the following crazy idea.

It seems to me that what we want is to occasionally have good soft questions. The main thing that causes the close fights is the feeling that if we don't close soft questions there will be too many of them. We could have a thread on meta where people propose soft questions, and every week or two the one that's highest voted gets posted to main.

This would result in higher quality soft questions, fewer close wars, a consistent stream of soft questions for people who like them, and a cap on soft questions for people who don't.
  • 16
    This isn't so much a wish list thing, we can just do it. – Noah Snyder Oct 18 '13 at 3:18

Watching questions and answers for comments.

In StackExchange, you can mark questions you are interested in as "favourite", and later find new answers for those questions by checking the list of your favourites ordered by activity. This way, you can learn if someone posts a reply to a question you are interested in, even if you haven't asked that question.

However, there seems to be no way to watch a quesion or answer you haven't posted for new comments posted for them. I'd like to see a mechanism for that. One reason why this might be useful is Dealing with possible incorrectness : if you read an answer to a question, you'll want to find out later if that answer is found to be incorrect and someone comments on this.

Possibility to categorize favourites.

I am a hoarder, every question which I would like to return to, I favourite. The problem is that it is a bit hard to manage. I would like to be able to create custom question categories. E.g. Math competition inspiration , Sequences, Unsolved, Research-related, etc. would be some of my categories, where I sort my favourite questions.

This is similar to how I use Youtube.

Return to a "discussion" style forum for meta, like http://tea.mathoverflow.net/, but properly integrated (e.g. with uniform logins) with the main site.

  • 2
    I couldn't agree with this more! Maybe this time with threading enabled (^_^) – Daniel Moskovich Oct 18 '13 at 7:23
  • 9
    I was very much against adopting this new meta before we did, but now I think it works quite well and I feel there is more participation on meta now. (This might also be do to proper integration, but I do not think it is only due to this.) Thus, this is certainly not a wish of mine, and perhaps I am now even against it (not completely sure). – user9072 Oct 18 '13 at 9:35
  • 4
    Scott, I was thinking about posting this. One candidate might be Jeff Atwood's new system called discourse. @quid: I feel that the integration has a lot more to do with the participation. – Asaf Karagila Oct 18 '13 at 11:47
  • @AsafKaragila Maybe, but hard to know. I still feel that it can be easier, in particular when one is not very involved with the site, to ask a question on something, then to start a discussion thread about something. There are some things I do not like that much about the new meta, for example, I would prefer the community-moderation tools like closing and spam-flagging where only available to moderators (as sometimes they get used in ways I do not approve of). But that voting is possible is somehow good in the end, also but not only as it sets a low entrance barrier for some involvement. – user9072 Oct 18 '13 at 12:41
  • 1
    @quid: There is a voting module for the software used on tea. I recall that people said that we can have voting if we really want to. I too agree that voting is a good thing, and also that it is easier to ask a question than to start a discussion, but I still feel that a meta platform which is actually a combination of the two approaches will be in fact superior (as long as it is well integrated into the main site, of course). – Asaf Karagila Oct 18 '13 at 12:44
  • @AsafKaragila I do not say the current system is perfect, and maybe what you propose would be an improvement, and I am confident it would be one for you and me, that is to say for those that are willing to get familiar with it. But I also think that it is an actual advantage that meta is near-identical to main and so infrequent users of meta are in no way scared away by (unfamiliar/new) technicalities. – user9072 Oct 18 '13 at 12:50
  • 3
    If we continue using the current meta, I think we might want to consider the following passage on FAQ: MathOverflow is not a discussion forum. As a side-effect of being very good for to-the-point questions and answers, the Stack Exchange software is bad for discussions and designed to minimize them. (emphasis by me) To me, it seems either that the reason given there for banning discussion is wrong (i.e., it's actually good for discussion) or that somehow we're using the software that is terrible for (and designed to minimize) discussion for important discussions about the main. – Yuichiro Fujiwara Oct 19 '13 at 1:39
  • To be clear, I'm not taking either side. If the Stack Exchange software is good for discussion (so you can have good discussion threads once you get the hand of it as quid claims), then the current FAQ is wrong or even a lie; the software mustn't be the reason for banning discussion on the main site. If it's not that good or actually bad as the current FAQ claims, then it'd be a very convincing reason for switching meta's software to a more discussion-friendly system. In any case, I think we might want to either change meta's software or revise the passage on FAQ. – Yuichiro Fujiwara Oct 19 '13 at 2:01
  • @quid: I'm more than fine with SE communities being aristocracies, ran by those who are brave enough to dwell in the meta sites. :-) – Asaf Karagila Oct 19 '13 at 11:31
  • 1
    @AsafKaragila well, I guess I am in some way part of it, so yeah, no problem :-) More seriously, I prefer a meritocracy, which is not that far from what we have in practice (I think). Yet, then it is also necessary that the system allows for new ones to enter and "to prove their abilities." And the entrance bar should not be high. – user9072 Oct 19 '13 at 11:39
  • @quid: I think that any user that has been "around" for a while will be taken seriously on meta. One just has to want to use it right. I mean, the intelligence required for abstract mathematics is far far greater than the one required to understand how a website works. I think. Well, I hope, anyway. :-) – Asaf Karagila Oct 19 '13 at 12:04

Users with some number of points should be able to view a post without answers that was deleted by the OP. Today there was a question from a new user about injective Banach spaces that drew substantive comments from Yemon Choi and Taka Ozawa. It has vanished. Since I cannot find it among deleted posts, that means (I think) that the OP deleted it.

It is bad enough that a new user asks a question without giving any identifying information. If the OP then deletes the question after receiving substantive remarks the OP is IMO being quite rude.

  • 3
    10k+ users can view selfdeleted posts just like any other deleted post. The problem is to find them. Do you mean to ask for their inclusion in the recently deleted list or something else? In the specific case your browser history might help you to find it again. – user9072 Jul 4 '14 at 18:23
  • Really? It is not listed under recent deletions, nor does it show when I do a search for deletions. How do you find it? – Bill Johnson Jul 4 '14 at 19:07
  • I also went to Ozawa's recent actions (which includes comments) and his comments on the thread are not listed. – Bill Johnson Jul 4 '14 at 19:18
  • Sorry if I was not clear; I think we are talking sideways. No, they are not listed under recent deletions. This is why I asked if you mean to ask for their inclusion in this list. Yes they are hard to find, as I said "The problem is to find them." But one can (with 10k+) still view a self-deleted post (if one knows the URL). So what you explcitly ask for is possible as you aren't blocked from viewing such a post (by contrast a 2518 points user is blocked from viewing deleted content even if they know the URL, and one could imagine that for selfdeleted content everybody is blocked). – user9072 Jul 4 '14 at 19:35
  • So, it is not about "to view a post" but about something else. Now I infer you would be happy if such a post was included in the list of recently deleted posts. But one could also envision something else, for example, a dedicated list (exists for moderators) or that recently asked yet already deleted questions stay visible in the general question list for highrep users, or that related activity (comments) stay for a while in the respective users activity tab. Or still something else. So I wanted to know what precisely it is that you would want. – user9072 Jul 4 '14 at 19:36
  • Finally for finding such posts: If you never saw it it would be quite hard to do so. One could step through the posts by number. But in the particular case as you saw the post earlier you could have a good chance if you are using the same browser where you saw it and do not have "agressive" deletion settings for your browsers history. By browsing the list of URLs you visited in your browser, you should be able to find the URL of the question in question and then you can still view it. – user9072 Jul 4 '14 at 19:38
  • 4
    Thanks, quid; I understand now. My recommendation would be for the OP deleted posts to be listed under recent deletions. – Bill Johnson Jul 4 '14 at 19:56
  • I cannot use the history feature to find the post because I am working on a different computer. I saw the post yesterday while I was in my office. – Bill Johnson Jul 4 '14 at 19:58
  • Bill, luckily I had not got round to clearing the cache on this device: try mathoverflow.net/questions/173233/… – Yemon Choi Jul 4 '14 at 22:57
  • 1
    And you can vote to undelete (as I just did). – Gerry Myerson Jul 5 '14 at 10:20
  • 1
    You are welcome. I agree it can be irritating if posts just are gone, and it would be better if self-deleted posts could also be traced a bit better like you propse. (The handling of deletions and deleted content IMO was really better on MO 1.0.) // On the specific case and a bit more generally: I would say if one does not want ones comments answering a question deleted then one should give them as answer (in CW if one wants). Possibly OP thought in good faith it is appropriate to delete the question since it seemed it is was not considered as deserving an answer proper or an upvote. – user9072 Jul 5 '14 at 12:31
  • This is a bit off-topic here but, since @GerryMyerson mentions it: Generally I am almost always against undoing self deletions. What would be the point in this case? Does somebody plan to give a proper/detailed answer now? If yes, then maybe I could see the point. But perhaps OP is working on a revised version of the question. – user9072 Jul 5 '14 at 12:42
  • @quid, I am in favor of undoing deletions if significant mathematical contant has been deleted. – Gerry Myerson Jul 5 '14 at 12:57
  • @GerryMyerson I know there are different opinions. But I guess this is not a good place to discuss them in general. (A tangential point to consider: eventually autodeletion would almost delete the question as it is now, and chances are the autodeletion criteria will be change so that it actually will get deleted in the future, if noone gives an answer or votes possitively on it. So if nothing else it feels a bit pointless to undelete it without clear perspective of what will happen next, so that it will quite likely be deleted again down the road.) – user9072 Jul 5 '14 at 13:20
  • Thanks, Yemon. The OP also deleted a comment the OP made that was substantive (that an injective $C(K)$ space, $K$ compact Hausdorff, has a dense open extremally disconnected subset). The OP did not give a reference but said who had proved it. Apparently deleted comments really are deleted; at least, this 10K user cannot find it. – Bill Johnson Jul 5 '14 at 13:37

Customisable question lists

Allow the user to choose an arbitrary function of creation and modification dates, score, tags, answers, bounties (and maybe others) as an alternative to the existing tabs. For example, to remove questions with negative score, and/or to softly prioritise more relevant tags.

Edit - Summary of the comments: Many such queries are available using by the search box, for which there is good documentation and which generates re-usable URLs. More general queries may be attempted using the stack exchange data explorer, which does not appear to have user-friendly documentation.

Remaining suggestions: Add a relative times feature and form interface to search, add documentation to the data explorer.

  • 1
    At least in principle, this should all be possible using the Stack Exchange Data Explorer. – user9072 Oct 21 '13 at 11:37
  • 1
    I think this feature would require adding functionality to the search function. The search function already allows filtering questions by score and tags, so some of this is already possible. – Ricardo Andrade Oct 21 '13 at 12:21
  • 2
    I have to confess I was not aware before @RicardoAndrade's comment of quite a bit of the functionality even (normal) search allows already, filtering by dates seems also possible. It is now not quite clear to me what precisely should be added to search or what room still exists to be filled between somewhat common ideas for searching and highly complex ones that then however could be done with the data explorer. Could the request be made more specific for example adding an example of a use case? (See mathoverflow.net/help/searching for search doc) – user9072 Oct 21 '13 at 12:33
  • @quid I guessed that some of this functionality may exist, and find it somewhat comforting that such an experienced user was not completely aware of it. Can the search facility save searches for later use? Is there a user-friendly guide for the data explorer? – user25199 Oct 21 '13 at 12:55
  • 2
    @Carl: The search function cannot save searches by itself. However, it gives a url which, whenever visited, repeats the exact same search. One can always bookmark that url, for example. – Ricardo Andrade Oct 21 '13 at 13:18
  • @Ricardo Thanks. Relative dates (eg last n days) do not appear to be a current option for search. – user25199 Oct 21 '13 at 13:47
  • @Carl I do not know much about the data explorer (thus I only wrote 'should' and the 'in principle' as I from a rough look I find it not very user friendly) so I do not really know; somebody (François G. Dorais?) recommended to start with existing queries that are available (follow the link and explore 'queries') also likely many things were already done by someone, though one would have to find it. – user9072 Oct 21 '13 at 17:12
  • On a general note for the 'search' I could see some advanatge in having (also) a more intuitive interface (with several lines, and drop-down menus and so on). I could however imagine that having such functionality exposed too much, in particular as something that is done all the time 'by default,' say every user can create costumized tabs, might run the risl of being 'too expensive' regarding load on the servers (but this is a pure guess, I do not know much about such things). – user9072 Oct 21 '13 at 17:15
  • @quid Thanks, these comments are really helpful. I didn't reply yet about the use. Basically when checking the site I want to see all recent questions that are non-negative in my preferred tags, good questions in other tags, and questions with bounties, without having to do multiple searches. – user25199 Oct 21 '13 at 20:56
  • You are welcome! One thing that occured to me (and was confirmed reading around the docs of data explorer data.stackexchange.com/help) is that for regularly checking the site the data explorer is not really an option as it is only updated once a week. And, relative dates for the search sound like a really good addition. – user9072 Oct 21 '13 at 21:14
  • 2
    Some Stack Apps could be helpful? stackapps.com – François G. Dorais Oct 21 '13 at 21:33

When someone edits a post, it currently appears in the main list of questions with the editor's name and reputation. I propose that edits appear there with the original poster's or last answerer's name and reputation instead.

The current system overemphasizes the importance of edits. This proposal would provide a better view of the relative contributions.

  • What exactly do you propose? ATM in the "active" view (to which you link basically) the user displayed is the last to edit (or create) a post in the package of posts (the question or an answer) and it says "modified". By contrast in the "new" view it says "asked [OP of question]" Especially, if the modification is that a new answer was added do you still propose that OP of question is displayed? (Personally I am against any change here, it is useful and natural to see last that modified it in a list created on last modification.) – user9072 Mar 10 '14 at 18:00
  • @quid I did not and do not propose any changes to what happens when questions are answered. I propose only a change to what happens in the active view when questions are edited. – Matt F. Mar 10 '14 at 19:02
  • Sorry it is still not clear to me what you propose exactly. Say, Q asks a question, which is answered by A, so then according to your proposal it would say A when it reappears (as it is now). Now E edits the question (not the answer). Should it then keep saying A (as editing is not significant enough to warrant a change of the displayed name, which would be in spirit of what you propose I think) or change to Q as it is the original poster's name of the post that got edited (which is what you wrote which however feels at odds with your justification of 'relative contributions'). – user9072 Mar 10 '14 at 19:41
  • 1
    @quid Point well taken. You were correct about my intentions and I have edited the proposal accordingly. – Matt F. Mar 10 '14 at 20:48
  • @MattF. Why is this here? Is there any reason to think this couldn't be done now? – François G. Dorais Mar 14 '14 at 5:26
  • @Francois G. Dorais: I am not an expert on StackExchange functionality, so this is in the category of fantasy for me. Perhaps you know better: how would we do it? – Matt F. Mar 14 '14 at 14:46
  • @MattF. If you post a question tagged support or feature-request, you will get responses. You won't get much responses by posting here since answers here don't ping MO mods nor SE staff. – François G. Dorais Mar 14 '14 at 19:05

Latex preview

For long posts, the automatic Latex preview slows down a lot the writing.
So I suggest to add the box "Stop automatic Latex preview" that we can check when it's too slow.

Improved search feature.

The current search feature lets you search questions and answers using terms occurring in their text and using categories. There are at least three ways how I'd like to see this improved.

  • Search whole text of threads taken as a unit, so that if you search for two words, you find questions that contain one word if an answer to that question contains the other word. Such hits may of course be ranked below hits that match a single question or answer.
  • Search in text of comments as well. This is useful because then if someone doesn't know some terminology for some mathematical phenomenon, then someone can mention it in a comment and people can find the question using that later.
  • (Restrict search to posts by a particular user. The most important use case is searching among your own posts.) Update: François G. Dorais tells in the comment that this third one is already possible. Great!
  • 1
    Using Google with site:mathoverflow.net will do item 1 and a bit of item 2. – François G. Dorais Jan 3 '14 at 19:07
  • 2
    Search already does item 3: use user:me or user:2000, for example. mathoverflow.net/help/searching – François G. Dorais Jan 3 '14 at 19:25
  • 1
    I think the number one feature for an improved search should be: searching TeX content. But then you need things that look alike should match somehow, even if the TeX is different. – Gerald Edgar Mar 7 '14 at 14:22

Restore the + 10 reputation for questions but make each negative vote also count -10 reputation.

To maintain MO alive or well, we need a regular flow of good questions, and not too many bad questions. It seems to me that one has given more weight to the latter, forgetting the former which in my opinion is more important. To the limited extent that the reputation system can do anything about the situation of question, we should have our incentives right. A good question is very important, perhaps even more than a good answer. But we're all afraid of bad questions getting voted up by too many times. Those questions often has a large number of positive vote, and an almost as large number of negative votes, hence in the current system give the OP a large total reputation. The -10 reputation should in a large part solve this problem.

  • I decided to post this as a separate question and feature-request – Joël Jul 4 '14 at 13:07

Standing Bounties

Similar to monetary rewards for unsolved conjectures, it would be nice to have the option for non-expiring bounties (possibly also showing on a tab, other than the recent bounties).

This is a feature that seems to be popular on MathOverflow but less so on Stack Exchange meta.

  1. Reduce the list on the main-window.
    Currently the list exists from every question that comes in and I can only manage it a very tiny bit if I use tags-deselection. But the success of this procedure depends not only on the sophistication of the set of tags (which of course ;-) is always too low...), but much more on the the ability of the poster, his/her question-focus (which might not meet my understanding) and lastly on his/her sheer willingness to tag meaningfully at all (a smaller problem than in MSE, but well...).
    Besides of this the main-window -which is in fact a classical "to-do"-list- cannot be nicely cleaned and/or managed.
    For instance, a simple single right-click on the question should allow to take the question away from my list.
    (Private ranting): for whatever reason we seem to have generally in current user-interfaces, that it is not easy to "put the dust away" and have it clean for the day, as it is otherwise common use in any well managed household.
    Such a concept has of course some impact, for instance a manageability of some "new/unseen", "seen/still interesting"- and "uninteresting/seen/done"-lists.

  2. Collaboration/"peer groups" of MO-users
    If that would be possibly implemented - that would already be a great relief for me.
    After that, one more science-fiction comes to mind: let us introduce the possibility to make such lists sharable between users of similar interest - so if I have found some people here, who have similar interests like me, and we "trust each other" - why cannot the "uninteresting"-right-click be made by some of them and be effective for me when I login later (and vice versa - of course). That would really keep the space on the main-window clean and focused on the questions which are interesting for me (and the collaborators) and help me concentrate on that interesting discussions. (I leave here the consequence, that such shared lists require then managing tools - because that's no completely unknown concept in the 2014-software-community and thus not really worth a discussion at this place of formulating a wishlist)

1 or 2 rep. pts / upvote for a comment

Many comments are very relevant and would deserve to be rewarded.
Of course, some rules could be applied for limiting the abuses.

Example of usefulness of the comments:
- for improving a post
- as a sketch of answer
- a link to a paper or an other post
- many others things

  • 5
    For approved suggested edits there is already a reward of 2 points (with some global upper limit); covering the improving posts. I rather think that comments are overused so I am against anything that further encourages their use. There is nothing wrong with giving a sketch of an answer as an "answer" in the technical sense. There is also nothing wrong with editing in (rather then commenting) a link in the orignal post. – user9072 Mar 6 '14 at 15:45
  • 1
    @quid: if people do like that, a priori I'm agree, else... – Sebastien Palcoux Mar 6 '14 at 15:52

OP consent required for migration

This regards concerns raised by Mark Sapir in this closed question, in which the OP was annoyed that his question had been migrated to math.SE: Migrating a question which is already posted on the other site

The proposal is:

1. When there has been a vote to migrate, the OP is informed of the decision, and is asked to give consent. If consent is not given, the question automatically goes "on hold" as "off topic".

2. Once the OP has given consent, the question is moved to the review queue of the site being migrated to. If users there agree for the question to be migrated, it is migrated. Otherwise, it remains in MO as "off topic".
  • 3
    I think 1. could be alright, 2. feels unnecessary and has the same issues to it that pre-moderation has in general as recetnly dicussed meta.mathoverflow.net/questions/1009/… Also I do not see much reason to have more (pre)moderation on the target site in a case where several people thought it could be alright there (in comparison to a situation where just somebody decided to ask there right away). – user9072 Oct 18 '13 at 10:02

The following is a feature request by Ali Sadegh Daghighi, here: https://meta.mathoverflow.net/questions/1032/how-can-i-keep-from-asking

Problem: The problem is about frequent asking by some users in MathOverflow and Meta. It seems this kind of asking is not pleasant for some other users even if the questions be on-topic, appropriate by content and popular by up votes.

Question: Is there a way to restrict the number of questions which a particular user can ask?

Possible Answers: It seems we need some new laws as terms of use like these: .

(a) Every user is allowed to ask a limited number of questions per week (or month) up to his/her reputation rank.

(b) The time interval between two questions by a same user should be at least a week or a certain number of questions in the forum array (for example $10$ questions).
  • There was a stackexchange restriction on the number of questions one could post in any 30-day period --- is that still in effect? The restriction was a lot less restrictive than one-per-week suggested in this answer. – Gerry Myerson Nov 3 '13 at 5:18
  • 4
    @GerryMyerson on MSE it is 6 per day and 50 per month (as far as I recall). It is not enabled on most stackexchange sites, but can be enabled on request (though I don't think the numbers are negotiable). Personally, I don't see any reason to have such quotas on MO. – Tobias Kildetoft Nov 3 '13 at 9:58
  • 2
    A list of all rate limits on StackExchange sites can be found at meta.stackexchange.com/questions/164899/… -- though I don't know which of these apply to MO as well. – Stefan Kohl Nov 4 '13 at 9:36
  • 1
    There are certain types of questions that can become annoying/problematic if they are posted with high frequency, but I think it is better to have such situations handled manually by the community than hard-coded some rule into the software (I don't know if that is possible, even experienced users face false positives with the limited existing simple restrictions). – Kaveh Nov 5 '13 at 9:02
  • 2
    Also I don't think such cases are frequent enough to require a rule and it seems to me that the request is mainly to provide a justification for the user to post similar type of questions that he was posting (and I think hey did become annoying after the first few). – Kaveh Nov 5 '13 at 9:09

Categorizing comments/answers like slashdot by attaching particular descriptors

Slashdot categorizes comments into different types like normal, offtopic, troll, redundant, insightful, interesting, funny, informative, etc... One can also sort comments by these categories. http://en.wikipedia.org/wiki/Slashdot#Slash_and_peer_moderation

You must log in to answer this question.

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