[OPE-L:7349] [OPE-L:879] Re: Re: Hodgson

Paul Cockshott (clyder@gn.apc.org)
Mon, 12 Apr 1999 22:03:39 +0100 (BST)

Rakesh asked for my views of Hodgsons new book.

I have obtained a copy of Hodgsons New book and read
about a third of it, including the section that
criticises the propositions put forward by Allin and I.

I will deal with some of his criticisms of our work

'However it is still an open question whether modern
computers can handle the actual amount of information
involved in the modern context.
Today it is likely that the number of different types
of commodity would vastly exceed 10 million. Many of these
individual commodities have variations and varied
specifications. Stiglitz noted that the specifications
of the characteristics of a particular, but standard,
white T shirt, filled up thirty small-print pages. If this
amount of information applied to each one of millions
of commodities would even the fastest of modern computers
be able to cope?'
(Hodgson 99, p54)

It strikes me that he is using language very loosely here,
what does he mean by vastly more?
Does he mean twice as many, 10 times as many or 1000 times
as many products?
The figure that we took in our argument came from Nove
who estimated in the 80s that the number of distinct
commodities in a continental sized economy was 12million.
Whilst the passage of a couple of decades might have increased
the number of products at a rate of a few percent a year,
this does not seriously affect our arguements for the following reasons:

1. The number of distinct varieties of commodities will be
bounded by the working population. In general each product
involves the cooperation of several people to make it, so
the number of products will be less than the population.
2. The rate of growth of the number of products even if it
is as much as a few percent a year, is so far below the exponential
growth rate in computer power, that it can not seriously
constrain the possibilities of planning.
3. We go to considerable length to show that the complexity of
the planning algorithms are low, being polynomial of degree less than
2, thus they are relatively insensitive to plausible linear
increases in the scale of the problem.

As to the fact that commodities may have technical descriptions
that may be voluminous, this is of course true, but:

a. It is largely due to the development of information
technology that these voluminous technical descriptions
can be handled and stored anyway, so that computers are
already being used to handle and distribute them.

b. It is simply irrelevant to the planning problem, as
these specifications relate to the dimensions, properties
uses etc of the goods. What is relevant for planning
purposes are what goes into making them. We argue that
the mean number of inputs per output grows as a function
of the form f(n) = n^k where 0<k<1. He does not
challenge this in any mathematically serious way. Vague
annecdotes about Tshirts are not a serious response.

Appart from this the major thrust of his argument is that
we have not adequately responded to Hayeks writings on the
importance of tacit information. Prior to publication of
Hodgsons book, but possibley after he wrote it we published
'Information and Economics: a critique of Hayek'(1), which,
I believe, responds to all of the points about tacit
knowledge that Hodgson cites Hayek's authority on.

(1) Published in Research in Political economy, and also
available from www.gn.apc.org/Reality/econ/index.htm.

Paul Cockshott (clyder@gn.apc.org)