It is a long time since I last wrote. It was really a terrible time since two weeks ago. Anyway, finally it is completed, so I can change my attention to more interesting staffs.

I was involved again in NOI this time — not as a contestant, but an (likely) observer. Due to the community service requirement of Tsinghua on issuing PhD degrees, I went to Mianyang, Sichuan this summer. Put aside the numerous critics on keeping this requirement since ’77, I was “teaching” the high school kids something for NOI. I quote the “teaching” part because for olympic students, the usual way is to learn by themselves. But it seems that this school is different from the others, since the teacher here decides most of the process. Actually I think it is quite a bad idea, because the olympic problems change every year, and I am sure that none of the high school teacher has a PhD degree in CS, so who will expect one that does not understand why regular expression is equivalent to DFA to catch up?

Another thing impressed me is the problem is becoming harder and harder. Not in the sense they bring more ideas and sights into the contest, but they are trying to push technical parts further. I don’t like the idea of discussing whether being technical or conceptual, but I think being technical too much just make the contestants to recite algorithms, which was clearly a sign of “bad contest”.

Fortunately this year’s NOI is better than the last two years, which is focused on some simple ideas. But they are still kind of “dirty” if compared to this year’s CEOI problems. I was very astonished to see Mihai was on the committee of deciding the problems of CEOI. Generally speaking, CEOI has more neat problems. Paraphrasing from the teacher, “you will know what you are going to do when you see the problems”. I admit that the problem sets in China are more or less more on modeling part, but not on the algorithm part. And I think it was intentional because if they are too obvious, then some of the kids will spend their time sleeping in the contest . It is very easy to get problems everyone knows how to solve, or the problems that no one knows how to solve. But for the problems good for a contest, it is kind of hard to make. I have to say CEOI is better than NOI on this point, for they are trying to increase the internal difficulty of the problems, and we are trying to make the problem more obscure. Although they also suffer from the problem of getting too much complexity on formulated optimizations (like this), but more problems are good even for talking after dinner (like this).

By the way, this year’s IOI is in progress, which is located in Bulgaria.