Excellent book --- elementary topics in statistical mechanics from an advanced
point of view.  The writing is deceptively straightforward, and there are a
lot of points where someone who is not already fairly experienced will
be a bit lost about what point Albert is making, when those of us who've
argued these matters over coffee in the halls approximately 10,000 times will
see exactly where things are going.  (The paucity of references contributes
to this effect.)  Also, some of his sentences are just too damn long.

He has persuaded me that Maxwellian demons are, in principle, possible.  In
fact I find it extraordinarily suggestive that his argument relies on their
being sensitive to microscopic facts and capable of amplifying them into
macroscopic ones; something close to the roots of intelligence, I think, is
at work here.  (If I knew how to articulate that hunch any better, I'd be
working on it.)  Describe his demonic argument.

I am less taken by his remarks on quantum statistical mechanics, because I very
much like what he calls "the bare theory", in which account there is no
measurement problem to be perplexed by, or any weird kind of discontinuity
between normal quantum dynamics and measurements.  So I'm not very taken by
the GRW ideas that he is.

Also, I think there's a larger role for ideas of ergodicity and so forth
than he does.  In particular there's a passage where he talks about how
deterministic systems couldn't look like they have stochastic subsystems
which seems clearly wrong (cite Ruelle).