Meta-Rationality (was: The Extropian Religious War)

From: Robin Hanson (
Date: Sun Jul 15 2001 - 12:32:42 MDT

Eliezer S. Yudkowsky wrote:
>In particular, rationality has the power to judge which "arational"
>elements of the mind are important, whether or not they're too dangerous
>to be used, and where their place is in the larger scheme of things. One
>of the things I really object to is the idea that rationality is somehow
>"incomplete", that there's something outside rationality that has the
>power to dictate whether I should be rational, rather than the other way
>around. I've simply never found anything like this. I've never found
>anything even close.

This is the essential idea of what I've called "meta-rationality",
which I think is a key issue in economics, politics, and policy.

Once upon a time, the main issue was: which is worse, "market failure"
or "government failure"? That is, are rational people who know what
they want but may not know relevant facts better able to coordinate
actions and information aggregation through market or government
institutions? In this approach there is little to complain about
independent individual choices - complaints only make sense when
choices are interdependent.

Now the main issue seems to be turning to how various institutions
exacerbate or mitigate "human failure", or the tendency of humans to
make bad choices even when those choices are not interdependent.
The rise of behavioral economics has highlighted various phenomena,
such as addiction and belief biases, which seem more easily explained
in terms of simple human irrationality.

Often authors recommend the policies by which a wise and benevolent
power would seem to be able to counteract the biases produced by
typical irrationalities. This approach, however, doesn't usually
deal seriously with the possibility of meta-rationality. That is,
if people are capable of realizing that they have certain biases,
and have even a few simple ways to counter these biases, then it
can be enough to just tell them about their biases, or to give them
more ways to control themselves.

For example, if people realize they might become addicted to
drugs, this can be enough to deter them from risking addiction.
If people realize that their future selves might "live for the moment"
and spend all their assets, they might choose savings plans that
do not allow all assets to be withdrawn at once. If people realize
that they may irrationally sign certain kinds of contracts, they
can choose their contracts to be enforced by a legal regime where
arbitrators can void contracts based on such considerations.

Of course it is also possible that most people are not meta-rational.
If so, benevolent paternalistic regulation might benefit most people.
Of course one might argue that the regulators are typically just as
irrational as the regulated. Or that any excess rationality of
regulators is outweighed by harm from their divergent interests.
One might also argue that while regulation is good for those fools
over there, *you* should be exempt since you can see that you are
meta-rational. But you may be self-deceived about this.

We're hardly at a place where we can settle these issues. But
maybe we have a better handle now on what the real issues are.

Robin Hanson
Asst. Prof. Economics, George Mason University
MSN 1D3, Carow Hall, Fairfax VA 22030-4444
703-993-2326 FAX: 703-993-2323

This archive was generated by hypermail 2b30 : Fri Oct 12 2001 - 14:39:48 MDT