Thursday, November 29, 2007

How copyright suppresses knowledge

On the web, I find a review of an interesting-looking book on cake-cutting algorithms. The review cites The Computation of Fixed Points and Applications by M.J. Todd (Lect notes in Economics and Math Systems; Springer-Verlag, NY 1976). Now, that in turn looks like a book I should be interested in, BUT, it's not online, and it's not in the library.

Well, I could request it by inter-library loan, but that's too much hassle for a book that may or may not turn out to be useful. Maybe I can buy it using money from a research grant, but even if a copy can be readily bought, will it be worth it, even if there's plenty of money in the grant? (And, that's also hassle.)

No-one really expects to make a profit selling this book right now, do they? (Maybe 30 years ago, you'd be ripping someone off if you copied it illegally.) But, copyright law, and in particular the absurdly long duration of copyright, prevents anyone from putting the book online. The result is that I probably won't ever read it, and the book will effectively vanish, at least for another 60 years, or thereabouts, when it may get to enter the public domain.

Of course, the above is just a specific instance of a general problem that recurs time and time again. Not just in a professional context either -- there are lots of miscellaneous books I've headr of and wanted to look at, which are not profiting anyone as intellectual property. If I write a book, I would want it to enter the public domain after about 10 years after publication.

Wednesday, November 28, 2007

Current state of my to-do list

Gray is stuff I have started on...

Update COMSOC flyer

dept web pp, update links to PhD study pp

research group web pp, add newphotos

chase missing amazon order

missing order for laptop

3rd year project design demos

read draft of approx NE paper

get P to attend one of my lectures

discuss networks funding with L

ask X to write letter on behalf of my grant application

read A's thesis chapter

review paper for STOC

reveiw paper for GEB

Make web page on Dagstuhl talk

Saturday, November 24, 2007

Buy Nothing Day

This evening I find out that today was Buy nothing day; too late since I already went to Tesco. It's a shame, I would have been happy to participate.

I get my ballot paper for the Lib Dem leadership contest, and vote for Nick Clegg. The differences betweeen the two candidates are really rather nuanced; this is a rather marginal choice.

talks; return home

I return home yesterday, with a four-hour delay to my flight.

Let me mention 2 other good talks. Uri Zwick gave another nice one yesterday on a recent result he co-authored (on a deterministic sub-exponential algorithm for parity games). The hallmark of a great talk is to strip away the layers of definitions and notation that are needed for a precise paper, and to get to the heart of the fundamental idea, and make it look simple. It is hard to do this with one's own work; we do not like to make our own work look simple. I also liked Ronald Peeters' talk on homotopy methods on equilibrium computation - again, a simple idea: you want to compute an equilibrium of a game, so you start out with a game for which you know the equilibria (having the same number of players and strategies) and you gradually move all the numbers that define that game, towards the numbers that define the game of interest. As you do so, try to keep track of one of the equilibria, which are themselves moving continuously as this process goes on. Apparently the Lemke-Howson algorithms can be thought of as this kind of process, which gives me a new way of thinking about the L-H algorithm.

One thing this workshop has done for me is, give me an impression of the interesting and varied ways that people have tried to implement algorithms for computing Nash equilibria in practice - my own work has just been on the analysis of algorithms in the abstract.

Thursday, November 22, 2007

Talks

I give a one-hour introduction to the complexity class PPAD. Mostly done on the blackboard but using the laptop to show some nice depictions of Sperner's lemma. A nice talk today was by Martin Hoefer on pricing edges in a graph for buyers who want to obtain spanning trees; it started out with examples that explained the scenario very nicely. Another talk I liked was Uri Zwick (on Monday) on introduction to parity games and related games; it was very informative.

Later, I distribute some flyers for COMSOC-2008. I play some pool and ping-pong.

Tuesday, November 20, 2007

Stackelberg schedule

I give my talk on the price of Stackelberg leadership in network games (joint woirk with Pattarawit Polpinit). As the last speaker of the session I am in the unhappy position of Stackelberg follower, in that previous talks can overrun but I am cut off by the dinner bell. That's a nice analogy with the outcome of Stackelberg leadership in the context of Cournot competition, but not one that I had time to dwell on! I was also asked to give a general introduction talk to the complexity class PPAD, which I will do on Thurday morning.

Sunday, November 18, 2007

Dagstuhl

I am once again at Dagstuhl for a workshop on Equilibrium Computation. Tomorrow I give my talk on the Stackelberg version of a well-known network-sharing game (joint with Pattarawit Polpinit; it is part of his PhD work).

Friday, November 09, 2007

Spiked online

I recently found the web site Spiked-online, sort of an online newletter, with social-comment articles having a libertarian bias (but it's actually a UK web site!) Despite strongly disagreeing with some of their stuff (notably the critiques of environmentalism) I found it to be quite a refreshing read. Indeed, the anti-environmentalism articles had some quite valid attacks on the less intellectually regorous aspects of environmentalism, or which there are many... I will read on.

Thursday, November 08, 2007

another project...

I enter into research discussions with Patrick Briest, Piotr Krysta and Heiner Ackermann, who is visiting from Aachen (an increasingly well-worn trail between there and Liverpool!). Best-response dynamics in splittable-flow network games.