John McCarthy (jmc@SAIL.Stanford.EDU)
26 Jan 1995 18:04:32 GMT

In article <3g7t7j$> (David A. Johns) writes:
In article <> writes:

# I respectfully submit that our knowledge of human genetics is
# inadequate to determine a priori that it is impossible for the
# genetic differences between 'races' (as used by Murray and
# Herrnstein) to have behavioural consequents. Rather than
# rejecting their racial classification as worthless it would be
# more open-minded and scientific to evaluate its explanatory power
# on the basis of objective statistical evidence. 'Political
# implications' has no place as an argument in scientific debate.

Fine. Let me give you a head start:

1. Define intelligence.
2. Define race.
3. Devise a test that can measure intelligence reliably without
regard to a person's prior experience with the measuring
4. Devise an experimental design that can isolate a subject's
race from any other influences that might correlate with race
without being deterministically associated with it.
5. Construct a reasonable model in biological terms for any
results you get.

And by the way, you might try to avoid saying things like "you can't
prove that it isn't true" among scientists. That line has
unfortunately evolved into somewhat of an in-joke in that crowd.

David Johns's strictness certainly has advantages for racists. Until
you have defined race to his satisfaction, the proposition that blacks
were ever enslaved in the United States or are subjects of
discrimination today will be meaningless to him.

Precise definitions are good when you can get them, and when the ones
you have accurately cover the phenomenon being studied. However,
scientific progress is often made without them.

For example, for several centuries (say 1630-1860) temperature had to
be defined as what a thermometer measures. But the chemists and
physicists knew that different thermometers could not be calibrated to
always give the same readings and could argue about which was right.
In the middle of the 19th century thermodynamics allowed a precise
definition of temperature for systems in equilibrium. This definition
extends pretty well to defining local temperature in flowing gases.
It breaks down when the mean free path of some constitutent is
comparable to the dimensions of the apparatus, e.g. in the very high
atmosphere or during the explosion of a nuclear bomb.

Almost all social science and much physical science would be required
if everything had to be precisely defined before science could start.

IQ is like an old-fashioned thermometer. It measures intelligence
well enough to do quite a lot of science.

Demanding precise definitions where they cannot be had or aren't
useful is a form of filibustering. Certainly it plays this role in
arguments about intelligence.

John McCarthy, Computer Science Department, Stanford, CA 94305
He who refuses to do arithmetic is doomed to talk nonsense.