Speaking of the Soviets and Cybernetics... I highly recommend
Spufford's Red
Plenty <http://www.amazon.com/dp/1555976042/>. A beautiful mixture of
fact, imagination and fiction set in the Khruschev era.
Guess that has to go on the heap, given that that book has inspired a
great blog post:
http://crookedtimber.org/2012/05/30/in-soviet-union-optimization-
problem-solves-you/
(which seems to be just the tip of the iceberg discussing the book)
Apparently even if you manage to figure out how to get people to tell
you the truth, rather than what they believe you ought to hear, and
even if you manage to know precisely what you're going to want,
rather than what you currently think you'll want, and even if goods
are nearly perfectly substitutable, the mathematics of even the
simplified problem is much more intractable than I had supposed.
(which, would, however, explain nicely why we don't all take these
techniques for granted after more than a half-century)
The Gibbs phenomenon is a nice example of how taking the first few
low-order terms from something which is an approximating series in a
mathematical sense may not lead to desired behaviors in the practical
sense; it was noticed and analyzed in the XIX, back when computing
was done with brass, not with bits, using machines like the following:
https://www.youtube.com/watch?v=6dW6VYXp9HM
-Dave
Shalizi
More serious is the problem that people will straight-up lie to the
planners about resources and technical capacities, for reasons
which Spufford dramatizes nicely. There is no good mathematical way
of dealing with this.
(a point which is echoed in heartfelt manner by Gray and Reuter in
their book on "Transaction Processing)
