jump to navigation

Math Overflow October 14, 2009

Posted by Scott Morrison in Math Overflow.

(written collaboratively by Anton Geraschenko and Scott Morrison, in google wave)

Math Overflow (MO) is a brand new mathematics questions and answers site. You should go give it a try! Several of the regular readers of this blog are already there. It’s much more fun if you actually ask a question, and the only way to get the full experience.

Math Overflow has lots of features that we think makes it really awesome:

  • Questions and answers get voted up and down, making it easy to find the gems in a large pool of content.
  • There is a reputation system. As your questions and answers get voted up, your reputation goes up. The more reputation you get, the more the system allows you to do. Above 10,000 reputation, there is little difference between between users and moderators.
  • It’s sorta wiki. Once you reach 2000 reputation, you can edit other peoples’ posts to correct typos and make the presentation clearer. Questions and answers can also be marked “community wiki”, in which case all established users (100+ reputation) can edit them. (By the way, when you post material on the site, it is under the Creative Commons Attribution Share-alike license.

Now MO isn’t intended to fill the same niche as math blogging. In fact, we’re not really sure what niche(s) it’s going to fill; it’s an experiment! It will be much more focused on “smaller” and “more technical” questions, especially ones that have definite answers. This is of course both good and bad. On the bad side, some really interesting questions will be less suitable for MO, simply because it’s harder to develop answers incrementally, conversationally, or collaboratively. In particular the “polymath” style of problem solving probably isn’t going to happen on MO to the extent that it happens on a math blog. On the other hand, MO has already proved itself surprisingly effective, especially if you really want an answer to a particular question. Hopefully after a while google will start effectively indexing MO, and we’ll even see incoming searches to previous questions.

At present the focus is largely in algebraic geometry, although the past week has also seen lots of growth in TQFTs and homological algebra. This is in part just because of the current user population, and it will hopefully broaden as people other than our friends start using it! Even if it remains concentrated in certain areas it may still be successful. If you’d like to influence the development, and encourage good coverage of your field, go ask and answer some questions, and invite your colleagues! One worry, of course, is that it will be overwhelmed by calculus students. Before you worry too much about this, go and browse through the current questions, observe the one or two inappropriate questions and how we’ve dealt with them. For now it really doesn’t seem to be a problem.

We’re hoping that this comment thread (and perhaps new ones as necessary) will become the central place for “meta” discussion of Math Overflow. In particular, if you have comments or criticisms about the site, or even the very idea, please tell us here. Complaints about bugs, the community, or the overbearing manner of the moderators are all okay too. We’re even happy to hear that you don’t like the color scheme. Some aspects of the site are beyond our control, but we’d like to do whatever we can to make Math Overflow a great place to do math.

Some known issues (and preemptive responses):

  • “MO is not a discussion site”. There was some talk in the comments here about whether MO is amenable to discussions. The Q&A format encourages people to ask particular questions which can be given definite answers. Some discussion will take place in the comments, in particular request for clarification or pointing out errors. To keep things on-topic, comments are limited to 600 characters. Also keep in mind that answers will often change order as later ones get voted up. All of this means that the framework forces any discussion to remain fairly focused. If you want to discuss something more broadly, well, come back the Secret Blogging Seminar. Remember that MO is not meant to fill the same niche a blog does.
  • There’s currently no LaTeX support. Obviously, this is the top-priority feature. This definitely won’t be implemented until the Stack Exchange software is out of beta, but it will probably be implemented shortly thereafter. In the meantime, do your best. You’re welcome to type raw LaTeX, use basic html (e.g. &Omega; and x<sub>1</sub>), or whatever you would normally write in email.
  • A few users have reported problems with their Google-issued OpenID. Occasionally, for reasons unknown, Google will issue an OpenID different from the one it issued before, so the site will not recognize you. If you have this problem, just create a new user and email Anton (geraschenko@mathoverflow.net). He’ll merge the two users and add the other OpenID as an alternate, so your account will respond to both. If it keeps being a problem, the easiest thing is to use a non-Google OpenID (like myopenid.com).

Finally, some statistics and a little background: Math Overflow is only 2 weeks old, it already has about 50 active users, 150 questions, and over 300 answers. It runs the same software (Stack Exchange) as the very successful stackoverflow.com for programmers. The hosting costs have been generously funded by Ravi Vakil at Stanford.

About these ads


1. Scott Morrison - October 14, 2009

Oh — something I thought to add: what’s the deal with not using your full real names as user names, folks? Personally, I find it annoying. But in your own enlightened self-interest, realise that participating in blogs, mathoverflow.net, the arxiv, and mathematical publishing are all forms of advertising for your “brand”, even if that’s not (and hopefully it’s not for almost everyone) your principal purpose. Since on job applications you need to write your real name, you might as well use it everywhere else, too.

2. Scott Carnahan - October 14, 2009

Color scheme complaint: it’s hard to distinguish questions with no accepted answers from questions with accepted answers. Both indicators are fairly similar shades of green (to my subjective view, when looking at my monitor).

3. geraschenko - October 14, 2009

@Scott C: I agree. I’ll try making the lighter green more light later today. Edit: Okay, I’ve lightened the light green. Is that better?

@Scott M: I’ve considered encouraging people to use their real names somewhere in the FAQ, but I’m not sure exactly where to do it. Perhaps in the “Be honest” section. Edit: I’ve basically added Scott’s comment to the “Be honest” section.

4. Andrew Stacey - October 14, 2009

I agree with the first Scott (I don’t disagree with the second, just don’t have much of an opinion there).

Apart from that, Three Cheers for Ravi!

5. Mr. X - October 14, 2009

@Scott M. — I agree with all your points, and I will add one more: the proliferation of short fake names causes problems for people like me who have genuine short names. I can’t tell you how many times I’ve been asked “OK, but what’s your _real_ name?”

6. Ben Webster - October 14, 2009

I blame the influence of xkcd.

7. Grétar Amazeen - October 14, 2009

I’m having trouble logging in. Today I haven’t been able to log in as Grétar Amazeen, as before. I always appear as unknown(google). Does anyone know how to fix this?

On a more positive note: great site!

8. Scott Carnahan - October 14, 2009

Anton: it’s a definite improvement; thanks!

I can imagine a universe in which the contrast was even greater, though. Maybe use more B in the RGB space? I don’t know how well that plays with accessibility issues for colorblindness.

9. Ben Webster - October 14, 2009


This is exactly what Anton was refering to above. The best thing to do is copy the OpenID URL you see when you go to the profile, and send that to one of the moderators so that they can add it to your profile.

Better yet, get an alternate OpenID (I just got one from myopenid.com. Takes about 30 seconds). I think you’ll also have to have a moderator stick that in your account too, but then at least you’ll never have the google problems again.

10. Grétar Amazeen - October 14, 2009


Thanks I missed that.

11. Andrew J. Leer - October 14, 2009

Is this site going to support LaTeX scripting to display math with eventually?

Like in tiddlywiki ascience pad

12. Math Overflow!! « Motivic stuff - October 14, 2009

[...] now there is a wonderful place for this (and other mathematical questions as well). Check out the SBS blog post and the site [...]

13. Math Overflow « Annoying Precision - October 14, 2009

[...] October 14, 2009 The news has already been spreading around the blathosphere, but recently a website called Math Overflow modeled on Stack Overflow has [...]

14. Urs Schreiber - October 15, 2009

I just had a look at Math Overflow for the first time.

I replied to three questions by pointing to nLab entries, that happened to exist. I think in each case the nLab entry contained all the information already given by other replies and more. And more structured.

I am saying this for the following reason: I am possibly the only one, but I do find it a pity that the MathOverflow design is subject to the same problem that also blogs have with accumulating information in a sustainable way: it’s gonna be all scattered.

All this effort people put into writing replies, and in the end the result is a page containing a list of contributions that are not harmonized, not integrated and not polished, not divided into subsections. With exactly the same and possibly even less effort each person contributing a reply could add that reply to a wiki page on the given topic. This way this page would incrementally improve and all the effort of those contributing to it would finally result in something nice and worthwhile. Of course the MathOverflow list of replies is better than nothing, but a structured wiki page on some topic seems to be much better even.

This doesn’t have to be the nLab of course. That just happens to be one attempt to do this. But it should be some wiki. Wikis are so much better suited for achieving what MathOverflow wants to or should achieve: incremental collaborative creation of useful knowledge databases.

What do you think?

15. Benoit Jubin - October 15, 2009

Your argument shows that mathoverflow is better suited to (rather) closed questions (I guess the replies you gave were to open ones), and actually to questions that are easy to answer for specialists of the field.

An advantage of mathoverflow over a wiki is that attention can be drawn to a specific question, and an advantage over a blog is that it can be drawn by anyone (even someone who does not have a blog, like me).

Maybe for some questions, mathoverflow could serve as an antechamber for a wiki: once a question is well settled and several answers have been given, one could structure them into a single wiki page.

16. Ben Webster - October 15, 2009


“Wikis are so much better suited for achieving what MathOverflow wants to”
There’s a good point in there, but I think maybe these sites are more complementary than you think.

Certainly, we should remind people to check nLab first. There are a number of questions that have just been referenced to Wikipedia.

But let me be honest here; writing on nLab doesn’t suit my psychology. It’s a fine project, and I’ve gotten some useful information out of it, but I’m just not motivated to do it. What should I be writing about? Who am I writing to? Do they actually care? I know that these are not real objections, but they are enough to essentially stop me from working on nLab. If someone were to write to me and say “hey, you should write/improve the nLab entry on such-and-such,” I would probably do it, but no one does.

So the important difference is that MathOverflow actually does have that person writing in and saying “you should tell me about such-and-such,” which adds a level of personal connection and fun to MathOverflow which I find lacking in the experience of nLab.

It is important to keep in mind that all this answers are under CC Share-Alike license, so as long as nLab is under a compatible license, any answers can be incorporated into nLab, as long as you attribute them to the author. I know that that’s not a trivial thing to do, but it does make these sites complementary.

17. astacey - October 15, 2009

Okay, now I’m scared. I was about to leave a comment when I discovered that the usual boxes for my information weren’t there anymore. I am apparently “logged in as astacey”. Who? What? Why? I never signed up for this!

Except that I probably did. Somewhere I must have signed up for something that said “We can use your information however we like.” in the EOL. In fact, the username I’m logged in as makes me pretty sure what that was.

Anyway, the point I was going to make was that I agree and disagree with Urs. I agree that some of the answers would be more suited to fleshing out on a wiki. But I think that that’s because we haven’t yet gotten a good idea of what’s a good question for mathoverflow. Or rather, it’s quite easy to state but quite hard to know if a given question fits the bill. Plus there’s a bit of “newthingitis” about it and hopefully it will settle down in a little while.

One easy advantage of the overflow system over a wiki is that on mathoverflow one is asking questions of other people. On a wiki, one has to know which page to go to to find the answer. On mathoverflow I can be a bit vague in my questions, knowing that they’ll be interpreted by people who can make a reasonable guess as to what I was talking about.

But these are all new toys in our playground and we’re not sure how to use them all correctly yet. Some may yet get buried in the sandpit (I have high hopes for the first mentioned piece of technology in this post), but there will probably be several that survive as fun things to play with.

18. Urs Schreiber - October 15, 2009

Yes, there is a good point of drawing attention to something. A wiki alone tends to just sit there like the Southern Oracle without advertizing what wisdom it might offer on which question.

But i think this means one needs a combination of both technologies:

For the nLab wiki, to name that again, we are currently trying to use it in combination with a forum software, the nForum. The idea is that a) everyone announes new additions to the wiki there and b) everyone can ask for additions or start related discussion.

But the point is that whenever we do obtain good material, good answers to something, we want to be sure to work them into the big wiki edifice and not have them lying around in the forum only to be forgotten if not the next day, then next year, only to be asked and answered again and again (or, worse entirely forgotten).

I do see the difference between “open” and “wider” questions and certainly a wiki would be suitable to grasp the wide and widest stories. But no open story proceeds without its closed subplots. I don’t think that there is a single non-trivial “closed” question on the MathOverflow side that wouldn’t like to sit in a corresponding page on the topic that it belongs to, together with its answer there.

So to make it a more positive statement: it’s not maybe that the hypothetical suggestion of the criticism here would be to close down MathOverflow and instead create some kind of wiki. It would already be quite useful — and maye THE useful thing to do — to have a wiki (maybe one of existing ones, like Wikipedia or nLab, or a new one) in the “background” (viewed from that perspective) into which answers would be filled in, after given.

There is copy and paste, so it’s easy to drop an answer at Overflow and paste it at the same time into the corresponding wiki entry.

I notice that many answers currently consist mostly of commented pointers to references. Just imagine what a highly valuabe collection of commented link list is eventually obtained by gathering together that information, already that bit of information currently present at MO, in a coherent list on some wiki.

Then, if I have a question of the kind “What’s a good reference for xyz?” I might first look at the wiki and see if the existing material helps me. If not, I’ll ask on MO, saying: “the wiki page on xyz so far sucks, does anyone have more information”. Then somebody says: “sure, that unpublished file of notes sitting on professor’s abc’s server is the source all insiders use.” Ah, great, so we enter the link into the wiki page with a comment “not published but widely circulated and definitely worth reading, just beware of the glitch on page 43.” Or the like.

This way, all the effort invested by all the contributors to MO would not just “flow through” for one-time usage but would eventually accumulate incrementally to something big and powerful. A database of worldwide shared insider knowledge.

And with the existing material on the wiki page existing, more and better questions are eventually generated also on the MO forum (or whatever it’s called). Because now we see at a glance the various pieces of wisdom accumulated over the months and years on a given topic, and then suddenly are able to draw cross connections. Like: oh, back then somebody said such and such, but now here is a piece that explains this from an entirely differentt angle. What’s the relation? And then somebody knowws what the relation is and there is actual progress.

This way the forum-scattered wisdom collected on the wiki would eventually, to overuse that phrase, be much more than the sum of its pieces.

That’s what I am imagining would be good. And anyone can start working accordingly. Me, in any case, whenver I come across anything at all noteworty these days, and be it a minor reference that I found on MO, instead of noting it down on a piece of scratch paper to lose it the next day, I type it into a wiki.

19. Urs Schreiber - October 15, 2009

Okay, great. I compiled my last comment above while Ben was posting his, so didn’t see that before. But I see we agree on the essential point, the complementarity.

If someone were to write to me and say “hey, you should write/improve the nLab entry on such-and-such,” I would probably do it, but no one does.

Oh, I see. Your day is not already filled with you chasing answers to knew question you come across yourself. That’s great, then I’ll start sending you questions from now on. The nLab needs more algebro-geometric stuf anyway.

20. Ben Webster - October 15, 2009

Well, you shouldn’t interpret that as a commitment. But sometimes it will work.

Of course, I could try to spend all my math time on my current research problems, but that would drive me insane; believe me, I came close over the past week banging my head against an apparent contradiction that seems to have been based on a bad calculation. Besides, explaining stuff on MathOverflow (and presumably adding it to nLab) is a very good excercise. It makes me go over important material in my head and remix it (in much the same way that writing a talk is a very valuable exercise for understanding your research). And of course, it’s useful to see other people doing this.

21. Andrew Stacey - October 15, 2009

By the way, in experimenting with all these technologies we do occasionally have to think of things like rights. There is almost no way that one could cut-and-paste an answer from mathoverflow into a wiki, since the formatting is all over the place (and even the markdown implementations differ between, say mathoverflow and the n-lab (incidentally, the mathoverflow markdown implementation is flawed as it looks inside tags and it shouldn’t)).

So any cut-and-pasting is going to be a look-and-rephrasing. At what point does one have to attribute this? After all, my answers so far on mathoverflow have depended on stuff I read elsewhere so I can’t really claim copyright on the actual content, merely on the exact expression.

(I hope I got my name right, this time)

22. Ben Webster - October 15, 2009


You have to attribute when you copy someone’s text. If you read their text, absorb the idea and put them in your own words, attribution is a courtesy, but not a legal necessity.

23. John Armstrong - October 15, 2009

Completely off-topic, but:

A wiki alone tends to just sit there like the Southern Oracle without advertizing what wisdom it might offer on which question.

Understanding full well that Die Unendliche Geschichte may be a more popular reference in Germany, you still win the Internet.

24. Anton Geraschenko - October 15, 2009

@Urs: Obviously, I don’t have any more experience with how Math Overflow will work than anybody else, but I have given it a lot of thought, and I think MO can be its own wiki partner.

First of all, I don’t think every question needs a wiki page. Sometimes the question and answer are just great as they are. But if a question resulted in a lot of half-answers or answers that should really be cleaned up, consider the following approach. Post a new answer to the question, starting with something like “This is a great question I’d really like a clean, organized answer that synthesizes all the things that people have said here.” Follow that by however much of the work you can bear to do. Before you post, make sure to check the “community wiki” box at the bottom right. That way, anybody with at least 100 reputation can edit the post. Also, the answer will not generate any reputation, so if people feel that such a synthesized answer is useful, they shouldn’t have any reservations about voting it up so that it appears right at the top of the list.

25. Anton Geraschenko - October 15, 2009


(incidentally, the mathoverflow markdown implementation is flawed as it looks inside tags and it shouldn’t)

I’m not sure exactly what this means, but you should post it as a bug at http://meta.stackexchange.com/. Or, if you explain it to me, I’ll post it.

26. Urs Schreiber - October 15, 2009

Well, you shouldn’t interpret that as a commitment.

Sure, I know. I was just kidding.

But let’s give it a try and see what happens. I just createdan nLab page with a question (of the “open” sort — maybe too open, but it happens to be bugging me anyway) that I would like to hear opinion on.

syntehtic differential geometry applied to algebraic geometry.

It ends with a standout box that lists more concrete questions. Well, for the moment there is just one. I am running out of time a bit,

I am gonna see that I post some summary version of that to MO now.

Incidentally, while cretaing that page I also felt like our nLab page on algebraic geometry could do with a more abstract nonsense perspective on the topic. But it turns out the minute I submitted that Zoran Skoda took issue with it and rearranged the material. So there is also some discussion going on there, sort of, concerned maybe implicitly with what in the light of recent developments the right higher cat-nonsense way of tallking and thinking about algebraic geometry is. Maybe somebody here is even interested in that.and feels like editing that entry a bit.

I’d be very much interested what the algo-geometric crowed here would do to the algo-geometric topic cluster at the nLab .

27. Urs Schreiber - October 15, 2009

think MO can be its own wiki partner.

I am not sure about that. The answer to a single question may nicely fit into a comment box, but that still doesn’t capture it in context then, doesn’t hyperlink it properly with entries on other keywords.

But the main point is: Few people will have time to play secretary and try to build cleaned up entries out of the bits other people dropped. It would be much more efficient if everybody who enters a new bit of information already tries to fit it into a grown entry, at the right place.

But of course I understand that my suggestions here might seem to undermine your efforts with setting up and running the MO site. I think if things work fine than Mo and some wiki would not only exist in parallel but be together more valuable than each by its own. That’s what I am feeling, anyway.

28. Carl Mautner - October 15, 2009

I find it annoying that one needs a `reputation’ to leave comments. I have comments and questions regarding other people’s answers, but don’t care enough to garner a reputation just to post them.

I find the idea of the site to be very nice, but the brownie point system to be repulsive.

29. Mikael Vejdemo Johansson - October 15, 2009

The brownie point system makes it difficult to join the site discourse with commentary that cannot be punished in the voting system – which I believe is by design: it relegates disruptive elements to asking questions and giving answers, both of which come equipped with a way to signal garbage.

And while it may feel insulting to have to qualify to participate fully in the discourse, it’s not really all that hard to get past the 15 points of reputation you need for comments to open up. (or was it 50? I forget)

There are many venues for mathematical discussion and for mathematical encyclopediae opening up right now. Sure, nLab is one of them, and an impressive effort. Wikipedia’s math pages is another. And I find BOTH valuable and use both, occasionally.

But there’s a wide spectrum of registers in the mathematical discourse online, and IMO, the MO fills a niche situated vaguely between Wikipedia and nLab – as well as features a more conversation-oriented tone: I can easily get intimidated by nLab; and writing for the Wikipedia requires a level of “everyone can read” that’s not always feasible. And in both cases, I’m not in a conversation – I’m writing a note for others to read, rather than responding to a question.

30. Noah Snyder - October 15, 2009

I think MO and a wiki have a very different feel. MO is very easy (in fact too easy) to get involved in, whereas a wiki is much less involving. There’s a trade-off between level of polish and ease of involvement. When someone asks a question on MO I can dash off an answer. It’s fun, it keeps me fresh at answering simple problems, and it doesn’t feel like writing a math paper. If I notice a problem on nlab (say I’m unhappy with the treatment of “simple objects” and the definition of “semisimple”) I have to track things down, come up with the right precise definition, etc. At that point I might as well be writing my paper where these issues come up.

(Now in the long run writing a good wiki for mathematics is better for mathematics than doing research. But in the short run I want a school to hire me, and that means when I’m doing serious math writing with correct details then I want it to be in a paper. There’s a limit to how much of that sort of writing I can do in a day, and it’s already much slower than my speed of research.)

On the other hand, it is the case that people should check nlab before posting questions to MO that are likely to be answered in nlab. The easiest way to do that is to have a few people on MO who know nlab and link to it when it’s needed.

31. Andy P. - October 15, 2009

A huge difference between MO and a wiki is the audience. nCat has a very distinctive point of view, and this strongly limits its audience. This is probably a feature, not a bug, and I would guess that it will hold for most analogous types of math wikis. MO, however, is much broader, more analogous to the sort of informal chats that happen at departmental teas. The question and answer format is perfect for this.

32. Noah Snyder - October 15, 2009

@Carl: Unfortunately the internet being the way it is, you need to do *something* the discourage comment spam. The 50 threshold is pretty low, a single good question or answer will get you past it. Also if a comment is substantive enough you can always post it as its own answer.

33. David Speyer - October 15, 2009

Ben writes: “You have to attribute when you copy someone’s text. If you read their text, absorb the idea and put them in your own words, attribution is a courtesy, but not a legal necessity.”

I would put that a bit stronger than “a courtesy”, since most of us are or want to be professional academics. In academic writing, it is expected that you credit your sources and it will reflect poorly on your reputation if you do not. I think that norm should apply to blog posts as well as articles; I am less clear about how it should apply to wikis.

34. Urs Schreiber - October 15, 2009

Now in the long run writing a good wiki for mathematics is better for mathematics than doing research.

The difference between how I work internally and much of the rest of the world seems to be that I work on the wiki as part of my research and teaching, not as something besides.

I never do much besides research. Don’t have time for that. I never blog just for the heck of it, either, but because it somehow concerns my research.


Much of the sheaf and category theoretic stuff on the nLab was created while I prepared a course on sheaves and stacks.

Much of the (oo,1)-category theoretic stuff on the nLab was created while I took care of a journal club on (oo,1)-categories.

Much of the rest that I put on the nLab is background material that I need for my research on differential nonabelian cohomology.

You see the pattern: I develop my original material in my “personal” wiki section, and all stuff that is standard enough or stabilized enough flows to the main part.

And don’t tell me you don’t all have this, too: to develop new ideas in the part of math that we are doing it is constantly necessary to absorb existing ideas and concepts by others. What should you do once you have absorbed a new piece of structure? Write it up in an nLab entry. Not as a courtesy to the rest of the world that just steals your time. Just for your own benefit. For nobody else.

But if you do it on the wiki instead of on your personal private piece of scratch paper, you have the additional advantage that the material sits there and is being eyeballed by others. I got so much useful feedback on nLab entries I had writted, which I would never ever have gotten otherwise.

It could be even better. If not everyone spent the day with thinking up 101 reasons for not to write something into the wiki, but just did it. Okay, I stop now. :-)

35. Ben Webster - October 15, 2009


I was referring to copyright law. There are ways of recasting other peoples ideas without using their words that are intellectually dishonest, though usually we do this in a way that’s fine. I don’t feel like it was intellectually dishonest of me to write my entry on Koszul duality without explicitly citing BGS and Keller and all the other papers I learned that stuff from. It was clear from context that I was not claiming that stuff as my research, just my understanding of the facts. But actually copying words from MathOverflow and posting them somewhere else without attribution is illegal, and you could theoretically be sued for it.

36. Ben Webster - October 15, 2009

Similarly, I don’t regard someone reading my MathOverflow answer, learning something, and then using that new understanding to rewrite the nLab entry on Koszul duality without attributing me as problematic or intellectually dishonest.

37. Mikael Vejdemo Johansson - October 15, 2009

It could be even better. If not everyone spent the day with thinking up 101 reasons for not to write something into the wiki, but just did it. Okay, I stop now. :-)

No, Urs. It’s not about thinking up 101 reasons for not to write something into the wiki. It is much more about me going to the Secret Blogging Seminar for a certain kind of discourse, and responding to comments here while I read this blog, it’s about me writing at Wikis when I feel the need for it, or the urge for it, or have something to write (see, for instance, http://haskell.org/haskellwiki/User:Michiexile/MATH198 or http://en.wikipedia.org/wiki/Operad_theory or http://en.wikipedia.org/wiki/Ext_functor); and it is about perceiving different communities online as having different registers of discourse, different modes of communication, with different things feeling appropriate in different venues.

A random thought will garner an IRC channel comment, or a post in my Livejournal. A serious random thought will garner a few minutes to formulate it, and get posted to my blog. A weighty question now has a venue: I can post it to MO. And an urge for writing up a good exposition of an obscure topic would go either to the nLab or to Wikipedia, depending on the time investment and the … flavour of the idea.

I would be less dedicated in continuing this discussion with you if I didn’t feel some amount of adversariality leaking out from your comments – and I’m saddened at this adversariality, since it seems to me to be not entirely warranted.

38. Anton Geraschenko - October 15, 2009

But the main point is: Few people will have time to play secretary and try to build cleaned up entries out of the bits other people dropped.

I think you’ve hit the nail on the head. But as Noah points out, people also don’t like to play secretary with the bits that come out of their own heads. If somebody has a great idea, I’d rather get it in crude form than not at all.

You don’t need to worry about appearing to undermine my efforts. I’m already quite sure that MO is providing something new for the mathematical community, and it seems to be doing very well so far, so I don’t have any kind of “protect my baby” instinct kicking in. This kind of discussion is important to figure out exactly what this tool is good for.

Duplicate material between different sites is a tricky issue. If the answer to an MO question is on nLab, then an answer consisting of a link is totally fine. But I don’t know if I would go so far as to say that somebody shouldn’t ask a question on MO because the answer is on nLab somewhere. I certainly wouldn’t close a question as a duplicate because it was a duplicate of something on another site. If something comes up on MO that isn’t on nLab, and it gets some answers (let’s say they’re messy answers), I’m not sure where to post a cleaned up version. I like the idea of a “community wiki” answer on MO, but I guess it would be fine to make a page on nLab and make a community wiki answer on MO consisting of a link. Having an entry in both places feels like it would be problematic, but given that the style of presentation is likely to be very different, maybe that’s the right thing to do. In any case, liberal use of links ensures that people can find the information they’re looking for, which is what matters.

39. Scott Carnahan - October 15, 2009

If only we had a Wave option for questions and answers, allowing real-time edits. This could make MO even more addictive than it already is.

40. Noah Snyder - October 15, 2009

For me there’s a gap between when I understand something well enough to feel I understand it/can informally explain it and when I’ve sorted out how to write it up. The latter stage is very time consuming and I don’t enjoy it. I already have to do it to write papers, but doing more of it is not something I’m likely to do. MO and blogging feel more to me like informal explanations while nlab feels more to me like writing a paper. Since my writing paper time is already as much as I can bear, serious wiki writing (for me personally) would be time not spent publishing. However this would have been different if nlab existed say my third year of grad school.

41. Andrew Stacey - October 15, 2009

I sometimes feel that the nlab has the wrong name. It ought to be something more GNU like: NAW: NAW Ain’t Wikipedia.

We’re still not completely settled as to what the n-lab is, but one thing that is fairly clear is that it is not Wikipedia. The purpose is not to make polished expositions of material, that’s a happy by-product. The purpose, and we’ll see if Urs agrees with me here, is to provide a public place where people can make notes about stuff.

We all make notes as we read papers, read books, doodle on pads of paper. The n-lab is somewhere to put all those notes, and incidentally make them available to others. Others might then read them and add or polish them. But even if they don’t, it’s still then easier to link to other notes that you’ve made.

So, Noah, I’d say you’ve gotten it absolutely wrong. The n-lab should feel exactly like the informal explanations. It should, in fact, be even easier since you don’t even have to get it right – someone else will come along and make it right.

Of course, it’s our fault for not making that clear on the n-lab. It’s interesting to hear what everyone else thinks of the n-lab and how that compares to what those a little more involved think.

42. Why do I find MathOverflow fun and nLab not? « Secret Blogging Seminar - October 15, 2009

[...] in websites, wiki. trackback There’s been a very interesting dicussion in comments on Scott and Anton’s post about the strengths and weaknesses of MathOverflow and nLab, and I thought it might be good to [...]

43. Qiaochu Yuan - October 15, 2009

Some people on MO don’t seem to realize that answers to a question change order depending on the voting and are responding to previous answers, which gets very confusing. If you want to insist that MO is not a place for discussions, perhaps it would be appropriate to make this clearer so that answers are more self-contained.

44. Urs Schreiber - October 15, 2009

The purpose, and we’ll see if Urs agrees with me here, is to provide a public place where people can make notes about stuff.

Yes, I perfectly agree with this and with everything else you say. In fact, we adopted your suggestion that the best imagery for the nLab is that of a Lab book and moved that statement to the very top at Home page.

But I also clearly see that just a few select get the point of the nLab. Not just here, even on our mother blog, the nCafe. Maybe we need to do something about it. But maybe not. I tried a lot. But I’ll stop.

I am remineded of Zoran’s objection even against the nforum. Maybe there is a cultrual divide here between those who do math in quiet, just me the pencil and the paper, or me the blank nLab page and my keyboard, and those who want crowds and eye candy and virtual points assigned to them. Should I add a smiley? :-)

45. Todd Trimble - October 15, 2009

@ Noah, #30: If I were unhappy with the notion of say simple object but didn’t want to go to the bother of tracking down the precise or best definitions, I’d be inclined just to register my discomfort in a query box and have a little chat about it, as if it were on a blog. I notice this is the way a lot of nLab entries grow.

46. Scott Carnahan - October 15, 2009

Somehow, I had failed to understand the “nLab as public sketchbook” idea until this conversation. I had the impression that it was an encyclopedia that I couldn’t understand (plus some unpleasant editorializing about my friends, the vertex algebras). I think I might be more willing to jump in now.

47. David Speyer - October 15, 2009

On a separate topic, it might be useful to add a portion of the MO FAQ explaining the color scheme. I see that each question has next to it the number of answers it has received, and that the color of the surrounding box varies, but I can’t figure out what the different colors mean. Similarly, why are my answers sometimes displayed on an orange background?

48. Andrew Stacey - October 15, 2009

Scott C. Please do! I want to understand vertex algebras so would love it if you could write something on them.

And to add to Urs’ comment about public versus private maths. I do some of both. There’s projects that are very much in their infancy which I only have a vague idea of what to do with them so I start them on the n-lab in the hope that I’ll get a little feedback, maybe even a collaboration!, and so be better able to direct my thoughts. Some of my more established projects, then I’m still doing on paper (and consequently losing in the morass that I call a “desk”).

49. me - October 16, 2009

Hey I’m reading MO for a few days now, haven’t made an account yet so I don’t know if this is already possible but –
Could we make it a rule that when people post a link to a book which they suggest, they use the tag “book” and when that book is freely available (in PDF for example) they use “free-book”…
As I’ve said, I’ve been reading the site for only a few days and I’ve already found a number of very interesting reading material…

50. Scott Morrison - October 16, 2009


it’s hard to make rules per se, but they are mechanisms for ecouraging positive behaviours. It’s thus up to us – the early participants, to create the community we want. In particular you might:

* do this on your own questions
* leave comments with, e.g. “Suggested tag: book”
* upvote and leave comments explaining why, e.g. “+1 for linking to online copy!”

Of course, you need a little reputation for this, so go create an account and ask a question. As the user base grows, it’s easier to quickly pass the reputation requirements, as a single good question will receive plenty of upvotes.

51. Andrew Stacey - October 16, 2009

I have a comment regarding bounties. I asked a question a short while ago which hasn’t had an answer. I would quite happily offer a bounty for that but if I don’t get a satisfactory answer then I wouldn’t want to award the bounty at all.

That particular question is perhaps not the best example. So here’s a hypothetical. I ask “How do you prove the Riemann hypothesis”. No one answers. I offer a bounty. Someone posts something which isn’t an answer (but which isn’t daft enough to get deleted). After a week, that answer gets (half) the bounty whether or not it was actually an answer.

Is there some way round this? Obviously, if I had complete freedom to withdraw bounties then that could easily be abused, but the system ported from stackoverflow doesn’t quite fit the right-or-wrong world of mathematics.

52. Anton Geraschenko - October 16, 2009

@David #47: For some reason, I feel like the FAQ should be short, but I’ll add something about the colors to the Tips and Tricks page. Questions with no answers have a reddish box surrounding the number of answers; those that have answers have a light green box; those that have an accepted answer have a dark green box. Accepted answers appear with a light green background (and a check mark next to them), answers by the person who asked the question have a light blue background, and deleted posts have a light red background. I don’t think there should be any orange backgrounds. Can you give an example? Are there any other colors that need explanation? Edit: I’ve updated the Tips and Tricks page (colors are Tip number 10). Feedback is welcome.

@Andrew #51: To avoid gaming, you never get any of your bounty back. If you’re willing to accept that, it’s a question of how much of your bounty gets awarded to somebody else. If you accept an answer, that person gets the whole bounty. If you don’t, I think half the bounty is divided among the top voted answers, and the other half disappears. If all of the answers have <2 votes, then nobody gets any of the bounty. I’ve updated the FAQ.

53. me - October 16, 2009

@50 Scott
OK I’ve created an account, but all of the questions asked so far are out of my league :/ I am just about to start college and the FAQ says this website is for “professional mathematicians, mathematics graduate students, and advanced undergraduates” and I’m neither of those …yet.
But OK I intend to ask a question about books and if it gets buried that would be OK…

54. Lane - October 16, 2009

As a site for and about math, I have seen a question met with hostility right away here and that this is reinforced by the moderators of the this site.

Behavior like this will lead to a bad reputation for the site. This is a community site and closing legitimate questions forces people out of the community right away. A philosophy should be adopted of “If it is about math, this is a good place for the question.” I think that the FAQ states this, but this is not the way the moderators (Anton) run the site.

As my interest is only in part math, I am not vested in the success of your site. But you should be.

There are many reasons stackoverflow as been successful. Take note that Joel himself has said there is no question “Too Noob” and to prove his point he posted a question about how to move the “Turtle in Logo” on stackoverflow. This page is probably one of the best pages to learn about logo on the internet.

Link to Joel’s question:

Your site will get both novice and advanced questions as your site grows and you will have to find a better way to handle the novice ones.

A question I point to is:

This question is clearly about math and the asker met with a condescending response of “Don’t ask calculus homework questions here.” I will admit that this question is mine and it has been a long time since I have taken Calc 1, 2, 3 and 4. But continuing to work on the problem I came to the same solution that was posted. But since the question has been closed I can’t post an answer that explains why the given answer is correct. With out explanation, how can we improve the quality of math help out there and take advantage of what makes stack exchange so useful.


55. Jason Dyer - October 16, 2009

Just to chime in re: nLab. When I first heard about it I presumed it was designed solely for the particular sort of topics on the n-Category Cafe; hence I was not interested. Whereas Math Overflow when announced was quite clearly much more open in scope, so I was intrigued.

56. Jason Dyer - October 16, 2009

Lane, unlike in computer science, math sites suffer a plague of students trying to get other people to do their math homework for them. So there is something of a natural suspicion when something elementary comes up. (Unlike students posting their homework, however, you had essentially the entire problem finished except for the last step.)

But your Logo example is a good one and I’m not sure if there’s some way to compromise to allow interesting elementary questions while still being a helpful environment to the mathematicians it targets.

This might be a case where someone with editor ability could jump in and generalize the question to more inclusive of deeper mathematics. At the moment there aren’t enough people on the site with a high enough reputation for that.

57. Mikael Vejdemo Johansson - October 16, 2009

Just like Jason pointed out above, there is a strong tendency for mathematical fora online to get bogged down, heavily, in students asking for homework help or outright cheating assistance and cranks pushing their own ideas louder than anything else around.

This bad signal/noise ratio for research discussions online has already driven me away from several different potential fora – sci.math and sci.math.research once upon a time, several IRC channels, et.c. And one of the very few ways to deal with it in building a forum specifically for active mathematics researchers is to clearly delineate the scope of the particular forum, and to moderate away anything falling outside the scope.

This moderation, though, can, and will, hit false positives. It’s a difficult skill to moderate well and foster community through moderation. But the alternative to moderation is all too often that the forum in question loses all value as a tool for the research community.

58. Qiaochu Yuan - October 16, 2009

Lane: I’m sorry if my first comment was rude. The sentence I wanted to emphasize from the FAQ wasn’t the first one, but the last one.

I stand by my second comment, though; artofproblemsolving.com is a great place to ask elementary questions and plenty of bright high-schoolers would’ve been happy to answer your question there. But MO, I think, will come to occupy a different niche.

59. Anton Geraschenko - October 16, 2009

@Lane: I’m sorry you had a bad experience, and I hope you don’t have any hard feelings. I should have left a comment better explaining why your question was closed. We are working on the FAQ to make it clearer what kinds of questions are appropriate for the site. Let me explain here why your question was not appropriate. By the way, I felt it was perhaps borderline, so I waited for somebody to answer it (and you even accepted the answer) before closing it. You can still contribute comments to the question and the answer if you feel you have something to add.

The FAQ certainly does not say “If it is about math, this is a good place for the question,” and I don’t think that would be a good philosophy. The FAQ says roughly “If it is of interest to mathematicians, this is a good place for the question.” A mathematician is a person whose primary occupation is doing mathematics. Just as the primary goal of Stack Overflow is to make better the lives of professional programmers, the primary goal of Math Overflow is to make better the lives of professional mathematicians. It happens that professional-level programming is somewhat more accessible than professional-level mathematics. The goal of Math Overflow is not to get the largest possible audience. We’re not trying to monetize the site, and we’re not trying teach mathematics to the general public in any direct way. The kinds of “newbie” questions that we allow are the sorts of questions an already experienced mathematician might have as she is exploring a new field.

If you don’t agree with the analogy to Stack Overflow, that’s fine; we can disagree. Maybe Joel and Jeff disagree with me too. Ultimately, Math Overflow has a particular target community (professional mathematicians), and the content of the site will be dictated by what that community does and does not want to see. Right now, the site is young and there are no high-reputation users to vote to close inappropriate questions, so the moderators do their best to prune questions we think professional mathematicians don’t want to see on the site.

Edit: I’ve updated the FAQ. If anybody has suggestions about how to further improve it, please post in this thread.

60. Akhil Mathew - October 16, 2009

@Anton: Could you please clarify the meaning of “interest to professional mathematicians?” For instance, would it be ok for someone to ask what Yoneda’s lemma means, and is there a cut-off level: say, no questions on baby Rudin, Ahlfors, or Herstein, but questions are allowed at Hartshorne type material?


61. Anton Geraschenko - October 16, 2009

@Akhil: I actually intentionally left the word “professional” out of that phrase and defined a mathematician as somebody whose primary occupation is doing mathematics. This is because I definitely want to allow questions like “What does Yoneda’s lemma mean/why is it important?” An advanced undergraduate who plans to go to graduate school is a mathematician by my definition; her primary occupation is doing mathematics, even though she is not a professional mathematician. Also, it’s entirely possible that somebody is a professional mathematician, but doesn’t know category theory and is trying to learn it, in which case he would be very interested in such a question.

I don’t think there should be any cut-off level. I can’t give a formula for determining if a question is appropriate. The way I personally decide whether to close a question is by asking myself, “is this relevant to mathematicians?” and “do I want to frequent a site with content like this?” When there are >3000 rep users (you need 3000 rep to vote to close a question), I expect them to do the same thing.

If you’re not sure if you belong at MO, you can lurk for a while to get a feel for what is acceptable, or be adventurous and try asking some questions and see how it goes. If your question gets down-voted or closed, then you know something about it isn’t right for MO, but it’s not the end of the world. Even Ben has had one of his questions closed.

62. Ben Webster - October 16, 2009


I would encourage you not to write a question that was just “What is Yoneda’s lemma” before you read the wikipedia article, but if you read that, or a standard reference, and wanted more details or clarification on something, I think posting a question that would be quite reasonable.

63. Ben Webster - October 16, 2009

@Lane in particular, but the world in general-

Please check Wikipedia and Google first. I found the answer to Lane’s question in 15 seconds on Wikipedia (I didn’t even have to put down my beer) and it would have actually been quicker to Google “number of divisors.” Which is not to say a question can’t be asked if there’s an answer on Wikipedia; maybe someone on MathOverflow has a better answer, but take what’s on Wikipedia and other easily accessible sources into account when writing your question.

64. Lane - October 16, 2009

To respond to your comment “unlike in computer science, math sites suffer a plague of students trying to get other people to do their math homework for them.”

I disagree with this. Stackoverflow has had much discussion about how to deal with people who ask homework problems, and another classes of questions where people flat out want someone else to do their job.

Here is a few things that the stackoverflow community has decided about homework after long discussion.

Basically “It would be impossible to stop it all even if we wanted to.” and “StackOverflow exists to help programmers learn and provide a standard repository for programming problems, both simple and complex, and this includes helping students.”

65. Lane - October 16, 2009

It is your site and I will just have to disagree with you.

66. Urs Schreiber - October 17, 2009

Scott writes:

unpleasant editorializing about my friends, the vertex algebras

The entry on vertex algebras is evidently stubby and waiting for people to improve on it. It would be more than appreciated if you added a paragraph or two.

I must say I don’t really know what “unpleasant editorializing” refers to exactly. Would be more than happy to see a green query box at that entry where you voice a concern, if thta’s what you mean to do. See How to leave comments and questions.

67. Jason Dyer - October 17, 2009

Lane, the level I’m talking about here is *not* the level at any CS site. This is because there are many more students taking math classes than CS classes. (The discussion you link to is interesting reading, though.)

Also a good number of those in CS classes are intending to be (in some sense) professionals in the field, so the attitude and approach are different.

In any case, I’ve been mulling the issue and wondering if there is a place for an elementary-mathematics tag. Example: recently various math blogs have been writing about why a negative number times a negative number is a positive number, including some professional mathematicians. The discussion has been fruitful and produced some material not available on other sites.

However, using such a tag would be conditional on it being a _general_ question. For example, “perform the division: 3x^2 + 5x – 3 / x + 3″ would be unacceptable, while “explain why synthetic division works, and is there some way to expland the same system so it works in more cases?” would be acceptable.

I understand if the idea gets voted down, but I wanted to at least set it out.

68. Two massively collaborative mathematical websites that readers may like « Delta Epsilons - October 17, 2009

[...] 2.0, Math Overflow, nLab, open source triumphalism trackback I realize that I’m late to the party on this, but there is a new mathematical website precisely for answering questions: Math [...]

69. Anonymous - October 17, 2009

I’d recommend editing the mathoverflow FAQ to describe more precisely what is meant by “mathematicians” and “doing mathematics”. For example, consider someone with a master’s degree in mathematics, working in government or industry, and doing statistical analysis, mathematical modelling, advanced software development, etc. (but nothing publishable). Would this count as a mathematician doing mathematics? My guess is that it’s not what’s intended, but lots of people fit the broader definition and describe themselves as mathematicians. (And this is perfectly reasonable: there’s no reason the word has to be reserved for math faculty or the equivalent.) I bet the FAQ is going to offend mathematicians in the broader sense who discover that mathoverflow doesn’t consider them to be mathematicians doing mathematics when their questions are abruptly closed.

Here is a few things that the stackoverflow community has decided about homework after long discussion.

One critical difference is that mathoverflow is run by people who assign homework to students, while stackoverflow is not, so they come at it from different perspectives.

70. Mark - October 17, 2009

I don’t think the site will be particularly useful till there’s support for math expressions. Just IMO. It would also be nice if you could somehow include on of the open source algebra systems so people could also evaluate stuff.

71. Qiaochu Yuan - October 17, 2009

Jason, I like that idea, but I’m not certain MO is the place to implement it – it feels too “discussion”-oriented, although it could certainly work, especially if the focus of the questions was on generalizations and suggestions for further reading.

72. Ben Webster - October 17, 2009


While I admit that isn’t really the audience we have in mind, interesting questions in applied mathematics are fine. I’m personally not that interested in statistical analysis, but I’m also personally not that interested in model theory. The issue isn’t really the definition of “mathematician,” but the definition of “interesting.” Lane’s question, for example, was certainly about mathematics, but it wasn’t interesting. It had a simple straightforward answer which is explained very clearly on a Wikipedia page, so it was a waste of everyone’s time (his included) for him to put on MO.

73. Short Items « Not Even Wrong - October 17, 2009

[...] a wonderful new research mathematics site: Math Overflow. For some discussion of it, see here and [...]

74. Michael Lugo - October 17, 2009

There’s a new thread about Math Overflow over at metafilter.

75. Peter McNamara - October 18, 2009

Some comments/questions about typesetting mathematics on mathoverflow:

I was able to find some advice for using HTML in an old question on the site, but it was not clear to me from the FAQ that such advice existed, and suggest that the FAQ be updated to mention the tricks available for typesetting.

Of course latex support would be much better, since it is a language we all already speak. Speaking of latex support, the few comments I have heard regarding the future availability of latex appear to be vague and leaving me with more questions than answers. For example it is not clear to me who (if anyone) is currently working on implementing latex, or if there is any proposed timeframe for this.

And speaking of which, I would like to record here that I hope for native latex support, as opposed to the wordpress “dollarsign latex blah dollarsign” version.

76. Qiaochu Yuan - October 18, 2009

It might not be a bad idea to include something in the FAQ about Googling a question before asking it; for example, the same question might’ve been asked on sci.math.research a long time ago. This might also help with the elementary question debacle. Then it would be easier to help the people who really need the attention of another human, i.e. people who don’t know the terminology of a field well enough to Google the question.

77. Anton Geraschenko - October 18, 2009

@75 Peter: I’ve updated the FAQ. Feedback is welcome: http://mathoverflow.net/faq#latex

Unfortunately, I can’t give much of a timeframe for when LaTeX support will be worked into MO. I’m pretty sure it won’t happen before the Stack Exchange software is out of beta; I imagine this will be another month or so, but I don’t know for sure. After that, I think LaTeX support will come quickly since it is the highest voted feature request on meta.stackexchange.com (btw, if you have a meta.SE account, please vote that question up further). MO is arguably the most active Stack Exchange site (the only two sites with more content are twice as old as MO), which I hope will give us some weight in the way Fog Creek implements the feature.

This brings up an important point: how do we want to implement LaTeX support? I feel like converting to MathML (like on the n-Category Cafe) is the right way to go, rather than converting to an image (like on this blog), but I’d like to know what other people think.

@76 Qiaochu: I disagree. I don’t think adding something to the FAQ will discrouage extremely googleable questions (like “what’s the definition of xxxx?”) since those people won’t put in the effort to read the FAQ; the right way to deal with such questions is by downvoting/flagging/closing them. But if a question was asked on sci.math.research a few years ago, I don’t want everybody who is interested in it to have to find it that way. I’d rather somebody post the question and somebody else post a link to sci.math.research. Then it will be easier for the next person to find.

78. David Speyer - October 19, 2009

@77, regarding LaTeX support. I prefer converting to an image. Many browsers do not support the MathML fonts, and I think our equations look cleaner than the n-Category Cafe.

I would also suggest that MO make sure to put the original LaTeX in the alt tags of the images, as we do here. This should make it easier for scripts to deal with the MO pages, and will be more accessible for people with vision problems or low tech browsers.

79. Andrew J. Leer - October 19, 2009

@79 I like this idea. This way you’ll still be able to copy content that you find on the site into something else that you might use to store math like ascience pad.

Maybe we should have people list what sorts of string formats they use in other programs to store equations/maths, and some form of javascript popup can be attached to the math images you mentioned so that one can copy it from the math image into another program of their choosing.

I was thinking that this could be similar to the way that some blogs display social networking sites and social bookmarking sites at the bottom of their blog entries, except that it would give you the option of which math format you might want to display the equation in the image as for copying into another program/website.

For instance you could have LaTeX and MathML be two of the options. (Those are really the only two I can think of but I’m sure there are more if someone would be kind enough to list them here).

80. Andrew J. Leer - October 19, 2009

I meant @78. My bad.

81. Anton Geraschenko - October 19, 2009

@78 David: One idea I had (which I don’t think it crazy, but might be) is to have a user preference for whether LaTeX is rendered as an image of as MathML. By default, it would be set to “image”, at least until almost all browsers render MathML properly without any work from the user.

82. Colaboração e Ciência: tudo a ver… « Ars Physica - October 19, 2009

[...] Math Overflow [...]

83. toomuchcoffeeman - October 19, 2009

Naive question/admission of stupidity here: I originally left some comments as an unregistered user (http://mathoverflow.net/users/474) and have now finally signed up as a registered user (http://mathoverflow.net/users/763/yemon-choi). Is there any way to merge the two profiles? If possible, it would be nice to transfere what meagre credit I’ve earned from the 1st profile onto the 2nd, but failing that to at least get rid of the 1st one to avoid duplication & clutter.

84. Anton Geraschenko - October 19, 2009

@83 toomuchcoffeeman: I’ve merged those users.

85. toomuchcoffeeman - October 19, 2009

wow, that was quick! thanks

86. Jacques Distler - October 19, 2009

I think our equations look cleaner than the n-Category Cafe.

I would strongly disagree, but perhaps that is a matter of taste.

I would also suggest that MO make sure to put the original LaTeX in the alt tags of the images, as we do here. This should make it easier for scripts to deal with the MO pages, and will be more accessible for people with vision problems or low tech browsers.

There no existing software that will make your output accessible.

It’s true that screen readers will read the @alt text aloud. But they do a particularly crappy job reading LaTeX markup, so that’s not much of a help. By contrast, MathPlayer does quite a creditable job of reading MathML equations aloud. And it can also convert them to braille.

And, for the merely vision-impaired (as opposed to blind) users, scaling your site up by 200% or 500% (which is what the visually-impaired do) makes the equations practically illegible.

If you want an idea of what your site is like, for the visually-impaired, don a pair of glasses, smear the lenses with vaseline, and surf your site at 500% magnification. I think the exercise will give you a different perspective on the images-vs-MathML debate.

87. Toby Bartels - October 19, 2009

Math Overflow rocks!

(Actually, all that I want is to be notified of comments to this thread by email, and I don’t see how to do that without making a comment. So you can delete this comment if you like without risk of offending me. Still, what I wrote above is true.)

88. Noah Snyder - October 19, 2009

People aren’t voting enough. There are questions with thousands of views and half a dozen votes. This is bad. Maybe we should put “remember to vote” in the header somewhere?

89. Mikael Vejdemo Johansson - October 19, 2009

I’m sure some people, just like me, are not certain, or have a higher threshold than they should, for what constitutes a vote-worthy question.

Also, does several thousand views really mean several thousand _vote-entitled_ views?

90. Noah Snyder - October 19, 2009

The problem with having a high vote threshold is that it means people’s reputation goes up slowly, which is bad. We want people to be able to leave comments, for example.

91. Qiaochu Yuan - October 19, 2009

Part of the problem with voting is that it’s hard to tell if a question is a good question if you have no background in the relevant field, so for example it’s hard to tell if a question is deep or a trivial exercise.

92. David Speyer - October 19, 2009

@Jacques Distler — I acknowledge that I do not know how most blind or visually impaired mathematicians work; and would be glad to have data. I have had dilated pupil eye exams, and wanted to read some math before my vision returned to normal; my solution was to directly read TeX source with my text editor at 300% magnification. I imagined that people with visual difficulties would rather have direct access to the source than to any visual format. But I’d love to hear from people more knowledgeable than me.

93. Scott Morrison - October 19, 2009

@88, 89, 90

Really appalling is all the questions with more answers than votes! If you’re giving any answer besides “the question is lame, because …”, then the question is good enough for you to vote on it!

At this point, I’m pretty happy to even just vote for okay questions in subject areas that I’d like to see more questions from. In particular, at the moment voting is an excellent way to encourage mathoverflow.net to develop in the directions *you* want it too.

Any questions which is clearly written with care and thought (regardless of whether it turns out to be “deep” or “an exercise”) deserves a vote too.


94. Anton Geraschenko - October 19, 2009

@88: There are “remember to vote” and “remember to accept” banners all over the page. You can’t see them because you have >200 reputation. Try browsing while logged out.

also @89, 90, 93: It’s good to encourage people to vote more, but don’t feel too bad when it doesn’t work. I think it will eventually get better. The problem is that people have no idea what constitutes a large number of votes, so they make something up. When they see a good question, they say to themselves “I think this is worth 7 votes. Does it have 7 votes yet?” rather than “Do I think this is worth voting up?”

95. Andrew Stacey - October 19, 2009

Re: voting.

I don’t tend to vote for questions because I find that they come into one of three categories:

1. Dull
2. Interesting and answerable
3. Incomprehensible

I obviously don’t vote for questions in category 1. I suppose I could subdivide category 3 into “Incomprehensible but elegantly written” and “Incomprehensible but really incomprehensible” and vote for the former. I haven’t been voting for 2 because I tend to try to answer those questions and it seems Bad Form to vote up a question that I’ve tried to answer. Maybe this is the wrong attitude.

Re: MathML versus Images

One great thing about MathML, which is implicit in what Jacques said, is that if you, the user, don’t like what you see then you have a measure of control over how it is displayed. Because it is *your* browser doing the rendering then you have control. Shipping something off somewhere else to produce the images means no control at all. As a naff example, note that mathematics in the comments here has the wrong colour background (it’s also a bad example because that’s easily fixed).

96. Math Overflow « mathematik, bücher & meer - October 19, 2009

[...] am 14. Oktober 2009 und am 15. Oktober 2009 im Secret Blogging Seminar [...]

97. David Speyer - October 20, 2009

Thanks for the feedback on vision and MathML. It does sound like MathML would be better for this purpose. Thanks to your arguments, if the fonts became standard on most browsers, I would be in favor of switching to it. As is, it sounds to me like the best possibility would be to support both with default on images. Of course, I imagine that what will really happen is that this will be decided by whoever is generous enough to implement it.

My previous standard for voting up a question was that, if a colleague asked me in person, I would say “Huh, that’s an interesting question!” It sounds like I should be more generous.

98. Noah Snyder - October 20, 2009

@94: I didn’t realize messages were easily targetable. Is this a hack where they’re actually “ads”?

@95: There’s no reason not to vote up a question just because you wrote an answer for it. I don’t consider that poor form at all. What do other people think?

@97: I mean don’t vote for questions that are uninteresting, but I’m just suggesting that a lower bar helps integrate people into the site faster.

99. Miscellaneous matters « Gowers’s Weblog - October 20, 2009

[...] From Quomodocumque I learned of a new website, Math Overflow, where you can ask and answer mathematical questions. It seems to be very active, with a lot of users, rating systems for comments and commenters, and the like. So in principle it could be another mechanism for pooling the resources of mathematicians with the help of the internet. For example, if you need a certain statement to be true and do not know whether it is known, then my guess is that you could find out pretty quickly if you post a question there. For more discussion, see a post over at the Secret Blogging Seminar. [...]

100. Ben Webster - October 20, 2009

If anything is unseemly, it’s answering a question you wouldn’t vote up (though I certainly do that from time to time).

101. Anton Geraschenko - October 20, 2009

@95: You should definitely vote up questions in category 2.

As Scott (93) and Ben (100) said, if you think a question is good enough that you’re willing to spend your time answering it, then it’s probably worth an upvote.

102. Andrew Stacey - October 20, 2009

Okey-dokey. I’ll be more aggressive in up-voting questions (and maybe even down-vote one or two. Maybe all those irritating algebraic-geometry ones that just clutter up the view.)

103. Scott Morrison - October 20, 2009

@102. I know you’re at least half-joking about the algebraic-geometry questions, but I wanted to point out to everyone that you can “ignore tags” to filter subjects out.

104. Scott Carnahan - October 20, 2009

I submit that if your reputation is more than 10 times your vote count, then you should be voting more. I think we’re looking for an inflationary dynamic here.

105. A.J. Tolland - October 20, 2009


I’m just thinking aloud here, but — if it’s possible to filter by tag, then maybe it would be useful to tag comments with the difficulty of the mathematics involved? It’s probably not necessary now, but it might be useful someday to be able to sort problems according to whether or not they need expert attention or could be answered by anyone with basic knowledge of a field.

106. Anton Geraschenko - October 20, 2009

@102, 103: if you add a tag to your “ignored tags” list, then questions with that tag are grayed out, but if you go to your user profile page and click the “prefs” tab, you can set it so that questions with ignored tags don’t appear at all.

@105: I’m pretty strongly opposed to tagging things according to their difficulty or sophistication. I think it would discourage people from looking at “difficult” problems because they expect not to understand them, or it would discourage people from looking at “simple” problems, thinking that they’ll be uninteresting. Worse, who decides the difficulty of a post? I can just imagine cocksure hot-shots deliberately not posting background information or obfuscating their post in an attempt to cultivate some official recognition of how smart they are. Even if you could avoid all of these problems, I don’t really see what good it does you.

107. Charles Siegel - October 20, 2009

Is there any way to tell who voted something up or down without hunting through profiles and clicking on the “votes” tab? It’s public information as it is, I was wondering if there was an easy way to search through it.

108. Anton Geraschenko - October 20, 2009

@107: You should only see a votes tab on your own profile page. If you can see how other people have voted, please report the bug to me. You should only be able to see the total number of up/down-votes cast by other users.

109. Charles Siegel - October 20, 2009

Ahh, ok, I hadn’t looked carefully. I don’t see how others voted, and had just assumed.

110. Chris Leong - October 20, 2009

I think the best way to handle “homework questions” is to only answer them the first time with the general method and then to afterwards downvote/close subsequent questions.

111. Scott Morrison - October 20, 2009

@110 re homework questions.

Why do you think this is better than “closing with extreme prejudice”? I’m more than happy (overwhelmed) by the early popularity of the site. Given that already we have fantastic participation relative to what we were expecting, I have no desire to be unnecessarily welcoming.

I think we should decide what community we want, encourage that, and for now discourage deviations from that.

112. Ben Webster - October 20, 2009

You see what happens when you don’t use your real name…

113. Anton Geraschenko - October 21, 2009

Doing this meta discussion in the comments is getting a little hairy, so we’ve set up a meta discussion forum at meta.mathoverflow.net. Please direct your meta discussion there.

114. Curl - October 26, 2009

Don’t know if this has been mentioned: You are able to list all questions, but how about only listing questions that has a tag which is in your “interesting tags” list?

115. Scott Morrison - October 27, 2009

Anton just showed me this:


It appears to be a copy-and-paste followed by poor-use-of-thesaurus job on this exact post. Imitation is the highest form of flattery??? Hilarious.

116. A.J. Tolland - October 28, 2009

What is that? A english->russian->english translation?

117. Test Your Intuition (10): How Does “Random Noise” Look Like. « Combinatorics and more - November 5, 2009

[...] I should try to ask the problem also on “math overflow“. See also here, here and here for what math [...]

118. Math Overflow « What’s new - November 19, 2009

[...] is an active discussion of Math Overflow over at the Secret Blogging Seminar.  I don’t have much to add to that discussion, except to say that I am happy to see [...]

119. Tara B - December 28, 2009

@116. I couldn’t say which language, but it’s pretty definitely been machine translated into another language and back. Quite amusing!

120. It’s 2 am, where is my long term strategy? « The Accidental Mathematician - October 8, 2010

[...] Overflow, which has just celebrated its first birthday and earned an article in the Atlantic, was created by a few people who thought that mathematicians could use a site similar to Stack Overflow in [...]

Sorry comments are closed for this entry


Get every new post delivered to your Inbox.

Join 610 other followers

%d bloggers like this: