This is a post I’ve been meaning to write for some time. As most readers will know, at the very end of 2008 Alex Frolkin, Olof Sisask and I started the Tricki, a wiki-like website where people could post articles about mathematical techniques. The hope was that after the site had reached some level of completeness, it would be possible to take a mathematics research problem (or subproblem) and search efficiently for known techniques that were likely to be relevant. It would be doing something a little different from Wikipedia mathematics articles, which concentrate more on what I like to think of as “things with names”. For instance, if you suspect that discrete Fourier analysis is likely to be a useful tool for your problem, then you can type “discrete Fourier analysis” into Google and find many links, including to a Wikipedia article that contains many of the basic facts. But what if it doesn’t occur to you that discrete Fourier analysis is what you need (even though it in fact is)? The idea was that, using easily identifiable features of your problem, you would be able to follow links to ever more specific Tricki pages until you would reach a page that explained when discrete Fourier analysis was useful and how it was used. In general, the whole site would be about how to do mathematics rather than about lists of facts.
I still believe in this general concept, but for a wiki-type site to be successful it must reach the stage where people think it is worth contributing. Since writing a good Tricki article takes a bit of work, the motivation to do it has to be particularly high. And it seems that it is not high enough for the site to have taken off. I hasten to add that I myself am just as guilty as anyone else — after an initial burst of articles I got distracted by things like Polymath and, dare I say it, conventional research, and haven’t written a Tricki article for well over a year.
During that time, a new factor has come into play: Mathoverflow. For the Tricki to be successful, it had to do something that Wikipedia doesn’t do. And now, to be successful, it would also have to do something that Mathoverflow doesn’t do. This is a serious point: I used to think that one of the main functions of the Tricki would be to make it much easier for people to find out what was known about a problem, but Mathoverflow seems to me to be a better way of doing that.
What I’m asking for here is a bit of feedback. Is the Tricki useful even in its very small and incomplete form? Is there, at least in principle, still a niche for the Tricki, or is it squeezed out by Wikipedia on the static side and Mathoverflow on the dynamic side? Should the Tricki just be allowed to die a dignified death? (Of course, the existing articles might as well stay there, so this would be more like the death of a certain dream.)
Here are a few miscellaneous thoughts.
1. Perhaps some kind of reputation system would provide just enough extra motivation for people to want to write articles. (However, this becomes more problematic when it comes to major edits of existing articles.)
2. Perhaps the Tricki could, at least initially, achieve some success by restricting itself to one or two technique-heavy areas of mathematics such as extremal combinatorics. I have a theory that the success of Wikipedia is partly due to the fact that it has reached the point where the default is that it has an article on something, so that if it doesn’t then that is perceived as a gap that needs to be filled. I think people are in general much more motivated to complete a task that feels as though it is well over 50% done than to do the earlier work.
3. I am prepared to have another blitz on the Tricki (though I make no promises about when), but I don’t want to do so unless it is actually going to be useful to people.
4. I think that there are at least some questions the Tricki might be more suitable for than Mathoverflow. I’m thinking of questions along the lines of “This problem that has just emerged is of a kind I haven’t seen before but it feels as though people ought to have thought about similar things; what should I do now?” Obviously Mathoverflow would be useful for some such problems, but sometimes they are a bit too vague, sometimes one might have a subproblem that one does not wish to make public, and sometimes they may be sufficiently easy that one would prefer just to look up the answer rather than bother other people with the question.
5. If the Tricki became more complete, then one could browse it more systematically than either Wikipedia or Mathoverflow. One could for example decide to read up on all the standard techniques in some subarea.
6. Perhaps part of the reason the Tricki hasn’t taken off is that a high proportion of the existing articles are about undergraduate-level mathematics rather than research-level mathematics. Eventually I would hope for both, so it seemed natural to start with low-level articles and work up, but this may have given a misleading impression of what a complete Tricki would be like.
I’ll leave it there. My two main questions are (i) whether it would be worth my while to put some more effort into the Tricki and (ii) whether there is any chance that if I did so then eventually it would continue to grow without substantial input from me.