From: Eliezer S. Yudkowsky (sentience@pobox.com)
Date: Wed Dec 11 2002 - 02:27:23 MST
Olga Bourlin wrote:
> From: "Lee Corbin" <lcorbin@tsoft.com>
>
>>I still don't believe in the highly touted "scientific
>>method", and I also think that "character flaw" is putting
>>it a bit too harshly. I will agree that those who can change
>>their minds about long-held beliefs are to be praised for
>>having the capability to do so.
>
> I'm wondering, Lee, why you seem to think that the scientific method is
> overvalued. What method would you use in its place?
If Corbin doesn't defend his statement, I will: the scientific method is a
formal and social approximation to rationality, knowably inferior to
rationality in specific ways. Science is probably the best approximation
to rationality that can be incarnated *as a social process*, but it's
still not rationality. There are many rationally knowable truths which
are not scientifically knowable because of restraints on science which are
necessary to make science work as a social process; analogous to the way a
police officer may know perfectly well that so-and-so is a Mafia head, but
be unable to produce evidence which meets the deliberately higher
standards imposed by the judicial process. Science only works because the
real humans underlying the social process are free to use full-strength
rationality in inventing theories to test.
What I object to is not the existence of science, Belldandy forbid, but
rather the way in which science is sometimes mistaken for the substance of
rationality. Reality has no a priori bias to produce truths which can be
proven by the standards of science at any given point in history; it is
even possible that there are certain kinds of truths which can be
correctly deduced from confirmed theories, but which are not testable even
in theory - i.e., observable aspects of our universe's very early history
lead us to deduce the existence of causally separated regions of the
universe. If so, the correct statement would be that these truths are
rationally known, but not scientifically known - they would not meet the
deliberately tougher requirements of proof that define the social process
of science. Some people would say that such untestable extrapolations are
irrational, untrue, or even nonsensical; and *that's* what I object to.
-- Eliezer S. Yudkowsky http://singinst.org/ Research Fellow, Singularity Institute for Artificial Intelligence
This archive was generated by hypermail 2.1.5 : Wed Jan 15 2003 - 17:58:40 MST