Re Allin's [5538]:
> A causal theory is capable of
> generating predictions _conditional_ on
> knowledge of the values of the
> theory's "independent variables". If these values > aren't in fact known
then one can't make
>_actual_ predictions, but that doesn't mean
> the theory is non-predictive in the sense Paul is
> using.
Do all causal theories, though, also claim to be
predictive? I think not. Paul and you are using
one understanding of causal theory which is not
the only understanding of the relation between
causal and predictive theories. I tend to agree
with (of all people!) Ludwig von Mises who
wrote that predicting the future is "beyond the
power of any mortal man".
One reason we don't know what the
values of the unknowns will be -- in addition to
the sheer quantity of unknowns -- is that
a change in the 'value' of one unknown can
then cause a change in the value of the other
unknowns which can in turn change the value of
what we took to be known (including the value
of the initial unknown that changed).
Let us not forget, Occam's razor notwithstanding,
that capitalism is inherently a very complex
system. If one, for example, makes predictions
about the long-run doesn't one have to have
a fully specified non-linear dynamic model?
Chaos rather than predictability would seem to
be the norm in this context. Rather than being
able to make reasonable predictions, it seems to
me that *the best* that is possible is to identify
what _looks like_ the most likely *alternative
scenarios*. In thinking about the future
under capitalism, after all, a certain amount of 'fuzzy logic' is required.
In reference to Occam's razor, I think that David
McCloskey offers a valid observation:
"Above all the economical reader delights in
the simplest argument available. Economists
shave dangerously close with Occam's razor.
If some apparently complex behavior can be
reduced to a slogan or a three-line proof the
economist can be relied upon to seize it"
(_The Rhetoric of Economics_, Madison,
University of Wisconsin Press, 1985, p. 135).
I think that many Marxists (particularly those
on other Internet mailing lists) also shave 'dangerously close' with Occam's
razor. Slogans
are much easier than an explanation of complex
systems and relationships, after all. [NB: this is
not a criticism of AC/PC.]
I am also highly skeptical of how predictive
models are developed in practice, e.g. market
forecasting models. All of these models, in
order to make any projection at all, must
estimate values for all of these unknowns. When
it gets down to it, though, even if these are
developed into sophisticated computer simulation
models, this estimation is simply GUESStimation.
And these guesstimates implicitly contain the gross
ahistorical fallacy that trends which have been observed in the past are
assumed to continue
in the future. This crucially ignores how trends can
*change* _and_ the role of *uncertainty* in
capitalism. If we come to think that we know what
_will_ happen in the future, then our theory must
be over-simplistic since we have eliminated the
essential role of uncertainty that is given by
the nature of capitalist production and
circulation.
It seems to me that if a theory claims to be
predictive, then it should make predictions.
This is expected of fortune tellers, after all,
who make predictions. If economists go around
saying that their theory is predictive then
truth-in-advertising should demand that they
show us what they can do -- and then have their
theory judged in part on the basis of the accuracy
of the predictions (after we have adjusted for 'external shocks' like
natural disasters that no
social theory can be expected to reasonably predict, but which can
nonetheless cause a
divergence between the projection and
what actually then happens.) Would we take a 'fire-eaters' words for it
that s/he can eat
fire or would we want to see her/him eat
fire?
In solidarity, Jerry
This archive was generated by hypermail 2b30 : Sat Jun 02 2001 - 00:00:06 EDT