(I was particularly intrigued by it as I'd encountered SO only recently and had posted on the algebraic-topology mailing list about using the model for a review site for mathematics.)

But to return to my point -- which I don't imagine anyone disagrees with -- I just think it would be nice if it were clear who exactly the funding comes from. Clearly, whoever it is deserves credit for it, whether it's Ravi in person or a big impersonal funding body such as the NSF.

As I said before, I'm also impressed by his generosity. I vaguely remember him saying something along the lines of, "the $1560/year is about the cost of flying in a speaker or two; I think this project is worth that much."

(For a while I thought this was an invitation on python-dev mailing list, but I think this is not true. I wonder what was that list/site?)

Anyway, I consider myself very lucky: the email by Anton went on 9/28 and it was clear from the start that MO will work, largely because it was already well thought-of. The rest, as they say, is history.

Over the last few years, I've had several conversations with various people about how it would be great to have some sort of an online math wiki-type thing. The details would vary a fair amount; sometimes it was a (counter)examples database and sometimes a wiki for recording generalizations of results in EGA. I assume everybody has dreams about organizing and sharing everything they ever think of, or of how great it would be if everybody in your field shared a big brain like the borg, so I won't go into it.

In early May of 2009, Dave Brown (one of the people I'd had these conversations with) sent me a link to Stack Overflow and said, "I wish there was something *exactly* like this for pure math...". I suggested that we might be able to convince the SO guys to run a sister site for us (Server Fault, the first sister site, had just come out), especially if we could find a pile of money to give them. We figured the hard part would be convincing them to host a site for us (or sell us their software), and the easy part would be finding some money, so we started putting together a case that the math community shared the right characteristics with the programmer community so a math version of SO would be successful. We watched this awesome talk by Joel Spolsky about SO, I started listening to all the SO podcasts and lurking on SO and on meta.SO, and in between chatting about math, Dave and I would chat about what the SO guys wanted and how we could best pitch to them.

At the end of June, Fog Creek announced Stack Exchange, which made us really happy. It meant that the hard part was completely taken care of. I signed up for the SE beta and registered mathoverflow.[net|org|com]. I sent out a few emails asking various people about how to get funding for the project. In retrospect, those emails were far too salesy. One of the people I emailed was David Smyth, who recommended I talk to Dan Erman (another person I'd had those conversations with, and a student of the very well-connected David Eisenbud). I met with Dan over coffee, and I don't remember all the details of what we talked about, but as I thought about it over the next couple of days, it thoroughly changed my perception of how we should get MO off the ground. The right thing to do was to get people involved in the project and worry about finding some money later (if I'd known how long the free beta was going to be, this probably would have been even clearer). A couple of weeks later, Dave Brown, Dan Erman, and I met with Ravi Vakil over lunch to chat about MO. We had a very good chat about what would be important in getting MO to work, and compared the MO idea to various similar ideas (both successes and failures). To our great surprise, at the end of it Ravi said he would fund MO at $129/month for five years!

On September 28, I finally got into the SE beta. I set up the site, deputized moderators Dave Brown, Dan Erman, and Scott Morrison, and sent out an email to everybody on the beta sign-up list. We had always planned on using SBS to get the word out, but we wanted to wait for a few bug fixes (one in particular) before publicizing widely. At some point, we got hit by reddit, so I recruited another moderator, Ben Webster, to help us keep MO on topic. On October 14, Scott and I announced MO on SBS, and it took off beautifully. Lots of people started providing great feedback. Pretty soon we set up meta (Andrew Stacey's recommendation), hosted by Scott. A bit later, we added jsMath support (files also hosted by Scott). Now here we are.

I was wondering how it came about ? Who proposed it over the dinner table ? Was it supposed to replace the newsgroups way of asking questions ?

Then comes Mariano: This is the canonical question that should be asked on the meta site! See meta.mathoverflow.net.

And the OP disagrees. I agree with Mariano, and I am also curious to find out, so I ask it in meta anyway. This could also be a discussion whether it was indeed an appropriate question for MO, and whether my impulsive action was appropriate.