A couple of months ago (May 2009), I created a project on Google Code to share Project Euler solutions.
At first I started with only a hundred or so answers, but I found more, and people began contributing them, and by July or August I had the answer to every Project Euler problem.
So in September, I set up a forum to discuss solutions. This quickly turned into a polymath project.
A polymath project is a large collaborative project, where many people work together on a math problem. The idea was perhaps first inspired by a blog post by Gowers.
There were previous successful attempts at polymath projects. One I found particularly interesting was this one by Terry Tao, which is about Q6 of the International Math Olympiad. This “mini-polymath” project is similar to Projecteuler-solutions in the way that both are solving already-solved problems with an unknown (but existent) solution.
I feel that this project (specifically its forums) provides a good example of a successful polymath project. From September to present, the community has solved 24 Project Euler problems. Some easier problems were mostly of individual effort (just posting the answer), while more difficult problems were of a group effort. A particularly hard problem (257 I think) took the community over a week.
Even though the project has existed for months, I still get emails and forum posts suggesting that I take the whole thing down.
Experiences with Projecteuler-solutions
Overall, I think the project is very successful. Right now, there are 200 registered users (albeit a lot of them spammers), and over 900 posts.
There are perhaps 10 or 15 serial contributers: people who come up with good insights for multiple problems.
Then there are many more people who occasionally come up with something useful. Of course, any new information is welcome in such a group effort.
There have been a couple of trolls. Although they do not contribute to problems, they provoke some entertaining discussions. As they do not really get in the way of solving the problems, I don’t ban them.
More recently, there has been a lot of spammers, linking to porn sites and similar crap. They were a bit harder to deal with, since I hosted the forums on forumer, which does not allow me to install better Captcha’s to stop the bots. What I eventually did was to limit thread creation to a specific group, which stopped the spammers.
Here are some impressions I got from being a moderator:
- The group effort is very powerful. No matter how hard the problem is, I’m confident that it will be solved eventually. Any individual will give up after some amount of time, but it’s much harder for the entire group to give up. Until it is solved, someone is always trying the problem.
- Much of the work is done offline. On one hand, we have a lot of low-quality observations. On the other hand, we have posts that make giant leaps of progress, but are done by one person. Indeed, for easier problems one person can come up with the entire solution. But for harder problems, multiple of these ‘giant leaps’ are required, sometimes by different people building up from the work done by others.
- The forum is probably the best format for this kind of project. The other polymath projects used the wiki format. I think forums are better because ideas are ordered chronologically making it easier to view progress, whereas in a wiki the ideas are organized by topic and many mini-discussions arise, while less is being discussed about the problem as a whole.
- The community usually does fine without any moderation. Members can create topics, work on the problem, and come up with solutions on their own. Even trolls are dealt with by the community, and I’m rarely forced to ban them.
Perhaps Forumer is not the best service to host such a project. I am not able to install most plugins, such as support for code syntax highlighting, for Latex, and for custom Captcha’s. But since my parents won’t pay for my own server hosting, I think Forumer is a good free service.