
I am having a good time at
Dagstuhl Seminar 07431 on Computational Issues in Social Choice. Almost all talks are very interesting and I had some good conversations with some of the people here.
On Tuesday there was an open discussion about complexity of voting. While participating in this discussion, it became clear to me that there is something very wrong with most of the existing works on complexity of manipulating elections, and only very few papers dealt with the problem in the approach I consider more correct.
[If you are not interested in details about my research, skip the next two paragraphs]It turned out that the principal authors of two of these papers are here at the seminar. I spent the night* between Tuesday and Wednesday thinking about this problem, and on Wednesday morning I had a developed idea. After telling
Vince about it, he reminded me of the
general Gibbard theorem, a corollary of which removes any hope of pursuing my crypto idea.
So, I let go of the crypto direction, and instead considered voting under partial information. There was limited work done on the subject, and I had some good ideas on how to model the problem. On Wednesday after lunch I got Vince interested, and together we managed to prove two interesting impossibility results and have some very important observations regarding this problem. As it seems, this work is on the way to become a paper.
I am very happy to be able to write a joint paper with Vince Conitzer. I have known him since the first conference I attended in my PhD, which, as luck may have it, was a
Dagstuhl seminar. Since then, I have met him in every conference I have attended. He has published over 40 papers, even though he has just recently finished his PhD, some of them with groundbreaking results.
* The reason I am working nights is my partial adaptation to jet lag, I go to sleep after dinner at 19:00 and wake up at about 3:00, I get enough sleep and don't miss any talks, even though I don't really live in the right timezone.