Re: poly: Modeling Economic Singularities

From: Perry E. Metzger <>
Date: Sat May 02 1998 - 15:29:50 PDT

Hal Finney writes:
> Perry E. Metzger, <>, writes:
> > The method of science says that when you have a hypothesis, you should
> > be able to demonstrate its validity by experimentation. Well, just
> > show me an experiment. No need for any justifications at all -- we'll
> > treat your equations as a black box handed down from heaven. All I
> > want is one experiment. Surely if your theory is good you've tested
> > it, have you not?
> Robin's paper was about agents spreading across the galaxy.

Actually, I was refering to just the economic growth model, but...

> He was speculating about behavior given a technology which we have
> not yet achieved. How could he have tested that? The only real
> test will involve waiting a few thousand years to see what our
> descendents do.

I agree with your premise (that it is hard to test such a hypothesis)
-- I disagree with the apparent conclusion (that we should therefore
be lenient.)

> It seems fair to ask for evidence on how well this overall approach to
> modelling has succeeded in the past.

The "overall" approach isn't sufficient. If you have a mathematical
model, at the very least you have to try plugging in historical data
into the stated equations and see if it comes out with a reasonable
match. Even that isn't always enough.

> Robin has provided some references which may do this, but I suspect
> that it would require significant effort to become familiar with the
> relevant research.

Frankly, the more I become familiar with "quantitative" social
sciences, the more they resemble voodoo and the less they resembles
science. The methods of science require that you state a hypothesis
and try hard to falsify it -- not that you produce dense equations and
note that "similar" equations have some general validity.

The idea of people making extraordinary claims without so much as a
solid experiment strikes me as pretty damn amazing.

Received on Sat May 2 22:30:48 1998

This archive was generated by hypermail 2.1.8 : Tue Mar 07 2006 - 14:45:30 PST