Boccara on Modeling Complex Systems
Many moons ago, Physics
Today asked me to review Nino Boccara's Modeling Complex Systems.
A somewhat smaller number of months ago, I sent them my review. They now tell
me they're going to be getting in touch with me about production within a few
weeks, so this is as good a time as any to put the review online. Comments
are welcome; this is a last chance to fix things...
Here are some things I'm aware of, but can't really change, for reasons
of space and/or tact.
- A list someplace of all the examples would've been nice.
- It's too expensive, but sadly enough not out of line for a
textbook. A paperback edition would be a Good Thing.
- It's not a good introduction for people who don't know
statistical mechanics at the level of, say, Chandler.
Such people do not read Physics Today.
- He could have said a lot more about statistical and stochastic
techniques. I'm kind of glad he didn't, though, because it means there's a gap
in the market, which, by expanding on this, I might be able to fill.
- Boccara defines an agent-based model as an automata network, a set of
points in (discrete) space, each of which has some state, with the
configuration of states changing due interactions between the points. Computer
scientists or social scientists define ABMs as a set of agents with internal
states, spatial locations and interactions between agents. The two definitions
are equivalent (or can be made so), but they're rather different perspectives;
it's like the difference between a field picture and a particle picture.
- My account of how physicists got into complex systems omits two causes
which I think are very important. One is the exhaustion of "low-hanging fruit"
in conventional statistical physics, i.e., the comparative dearth of
siginificant problems which can be straightforwardly solved
by routine applications or minor extensions of known
techniques. (I emphasize those words to ward off mis-understandings.) There
are many really important problems left, of course, like high-Tc
superconductvitiy, turbulence, etc., but they're all really freaking hard. The
other cause is physicists, especially theoretical physicists, tend to have a
strong conviction that we're smarter than other scientists, hence we should be
able to solve their problems. (Sadly, this is not true.) But more on this
theme will have to wait for my book On the Genealogy of
Complexity, which it will not be safe to write until I have tenure (for
(Oh, and to whoever recommended me as a reviewer to Physics
Posted by crshalizi at October 03, 2004 11:11 | permanent link