Ravi Vakil, Anton Geraschenko and I are writing an opinion piece for the Notices about Math Overflow. Following in the fine tradition that John Baez started with his opinion piece about mathematical blogging, we’d like to post our draft here, and ask for suggestions and criticism!
Note that we’re working inside a fairly strict ~800 word limit for an opinion piece, so if you tell us to add a whole new section, expect to be disappointed. (On the other hand, we are listening for ideas about things that you feel should be covered in a longer article, as we’re writing one of those too!) Particularly helpful would be advice from anyone who’s standing a little bit further away from Math Overflow than we are, and can point out any background or context we’re implicitly assuming of the reader.
Here’s what we’ve got:
The internet has changed the way mathematics is done. Online journals, MathSciNet, and the arXiv have existed for some time. More recently a number of blogs have become home to rather sophisticated mathematical discussion [1,2,3,4]. There are also mathematical wikis combining exposition and research [5] and several massively collaborative online “Polymath” projects [6].
On the new site Math Overflow, users ask and answer focused mathematical questions. They also vote on questions and answers, making it easy to sort out the good stuff. The majority of active participants are graduate students and faculty, including many leading researchers, along with a few advanced undergraduates and non-professional mathematicians. On a typical day, the site receives about 30 new questions and over 30,000 page views from around 2500 different users.
Math Overflow is a natural tool since doing mathematics is itself largely driven by asking questions. We often formulate a problem as a series of small focused questions and then try to answer them. Sometimes the answers are surprising and lead to more questions, and off you go on an adventure. Other times you get stuck, in which case you might try looking online. Math Overflow is heavily indexed by search engines, so you’re likely to find your question being discussed there. Since many mathematicians first visit the site in this way, the system is designed to make it easy to jump right in. You can start posting questions and answers without registering, and \LaTeX\ “just works.”
The site has a personal feel to it. It’s very satisfying to know that you’ve helped someone with their problem or that somebody has taken the time to help with yours. It’s like a global math department tea. You interact with excellent mathematicians of all ages and you get to do some really fun math. We know of cases where someone was stuck on a problem in their research, asked it on Math Overflow (why not?), and got a satisfying answer within an hour, or even minutes [7]! The site launched in October of 2009, and already questions have led to research papers with the asker and answerer as coauthors: [8] was a direct result of [9].
Upon first visiting Math Overflow, people are often surprised by the high quality of the content and the sophistication of the mathematics. This is made possible by features which allow the community to moderate itself, keeping the site from becoming anarchic. For example, established members of the community have increased power to make decisions. This ability is measured by a point system, somewhat unfortunately called “reputation.” Votes for something you’ve written increase your reputation. As your reputation grows, you gain abilities that allow you to organize and improve the site, so moderation duties are efficiently shared amongst the most active users. At first you can only ask and answer questions, but soon you can vote (one good question will get you this far), then you can create new tags, and eventually you can edit other people’s posts and vote to close bad questions.
A “bad question” is one which is either irrelevant to mathematicians or one where it is unclear what constitutes an answer. Homework questions, on the rare occasion that they occur, are closed and the asker is politely directed to one of many websites where such questions are appropriate. Blogs often host discussions, and wikis compile expository material; Math Overflow fills a different niche. It is optimized for focused questions which (could) have clear answers. As a result, it is awkward to use it for broad or open-ended questions, and such questions are often closed. Community norms dictate that you should use Math Overflow when it is the right tool for the job. The best questions include motivation and background, and have a clear goal.
The underlying software is StackExchange, the engine behind the extraordinarily successful programming site Stack Overflow. All contributed content is under the same Creative Commons license used by Wikipedia, and database dumps are regularly made available for analysis, reproduction, and future-proofing. [10]
It is hard to explain Math Overflow without showing it to you, so please visit http://mathoverflow.net and poke around! If you have a question that’s been bothering you that you are pretty sure \emph{someone} must able to answer, try asking it. If you find the main page overwhelming, go to http://mathoverflow.net/tags and click on a tag corresponding to your specialty. Currently some areas of math are disproportionately represented, but just as with the arXiv, it is gradually “filling out” as it matures. We’d love to have mathematicians from more areas actively involved, and already you’re sure to find something you’re interested in.
[1] John Baez, Notices of the AMS, Febuary 2010 Volume 57 Issue 03 http://www.ams.org/notices/201003/rtx100300333p.pdf
[2] Secret blogging seminar, https://sbseminar.wordpress.com/
[3] Terry Tao’s blog, http://terrytao.wordpress.com/
[4] Gowers’ blog, http://gowers.wordpress.com/
[5] The nLab, http://ncatlab.org/nlab/show/HomePage
[6] Polymath project, http://polymathprojects.org/
[7] Shtetl-Optimized, “Prove my lemma, get acknowledged in a paper!”, http://scottaaronson.com/blog/?p=432
[8] F. Calegari, S. Morrison, N. Snyder, “Cyclotomic integers, fusion categories, and subfactors”, to appear.
[9] N. Snyder, “Number theoretic spectral properties of random graphs”, http://mathoverflow.net/questions/5994/.
[10] Mathoverflow database dumps, http://dumps.mathoverflow.net/
We’ll be listening in the comments!