Um, isn't this the same problem as DRM? (which we all know cannot be solved.) The fact is, if my computer can execute the code and make the request, so can I. The only thing you can do is to try to mitigate the risk or make it more difficult.
Also worth checking out is USACO. The USACO training gate allows you to start with easy problems and the build up to the point where you can do the ioi problems. Finish the entire thing and I can guarantee that you will at least have a good understanding of the ioi problems. http://ace.delos.com/usacogate
If you were on Linux, maybe you could do something with FUSE. Your FUSE driver could present the repository as a file system, with .commit_id directories for historical data.
No. It is absolutely not. I just found it on reddit the other day and thought it was interesting. I think it is an intriguing idea but I don't think it will work because tagging cannot capture enough information about a particular web page. Personally I didn't find it to be very useful, but I was wondering what other people's opinions are.
Ask HN: Do you think that the large number of sites within most categories means that it will be harder for a new player to enter the market? Or do you think that it doesn't really matter since most of them are very similar? Or maybe none of the above?