A skeptical reader made the following comment to a previous post:
What exactly do all those AGT researchers do to increase profitability?
…
Shouldn’t we stop pretending that AGT has any relation to the real world?
The theory vs. practice debate in every field is interesting, but for now just a few quick comments:
1) Actually it seems to me that ignoring the relation of AGT to the real world is pretending. It is a fact that many companies are interested in this work, whether or not justifiably so.
2) At least what I did for a while in Google can be seen in this paper.
3) While there are a few “deep” theoretical problems in AGT, most of the interest in this young field is actually figuring out the right questions and models. An early split from the “real world” would make sure that we end up with the wrong questions and models.
4) It seems to me that this comment is part of a game we theoreticians like to play: “we don’t do anything useful — we just pretend so in order to get grant money”. I don’t buy.
5) However, I can’t resist participating a bit in this game and sharing a strategy of pretending to be useful: when I started working in the econ/CS border I observed a difference in the way economists and computer scientists gave examples. While in CS examples would look like “Suppose Alice bids $3 for a pair of socks“, economists would give the same example as “Suppose ATT bids $3B for a spectrum license“, which is so much more practical. My new suggestion for computer scientists is to move to “Suppose Alice the plumber bids $0.003 for an ad slot”, trumping the $3B, I would say. (Which makes me wonder how many other human inventions function over a range of twelve orders of magnitude like auctions do?)
There is definitely a really strong connection between theory and practice in AGT.
Yesterday Jennifer Chayes had an amazing high level talk about this connection and in particular about the connection between AGT and real-world networks (those that probably Microsoft research center in Boston works on).
I think, a lot of research centers are doing the same thing and of course they are not doing that because they want to spend their money without any practical effect of such a research.
[…] is part of a post titled AGT and practice, which is worth reading along with the paper. In the paper, Nisan and his coauthors first describe […]