Re: Opinions as Evidence (Was: What is to be done?)

Wed, 26 Mar 1997 02:35:54 -0500 (EST) (Robin Hanson) writes:

>There are lots of explicit models of situations like these, most but
>not all using Bayesian inference, and the robust result is that other
>people's opinions should be given a lot of weight, typically much more
>than the weight given to your personal experience. In fact, it turns
>out to be difficult to explain persistent divergence in opinions.

Clearly, then, this theory has two *major* defects.

a) The population as a whole is often wrong on major, simple issues (e.g.

b) Persistent divergence is downright ubiquitous

The obvious conclusion is that most if not all people do not use even a
passable approximation of Bayesian inference. I'd say most if not all people
have their opinions driven by memetic replication (as exemplified by fashion
and religion) and not a passable Bayesian approximation.

>From what I've read on Bayesian inference, judgements not based on Bayesian
methods or methods with indistinguishable results are objectively bad. If
most people are using an objectively inferior inference process, what does
that do to the value of other people's opinions for a rational Bayesian?

Interestingly, science seems to conform to the predictions of this "community
of Bayesian agents" model much better than society as a whole. Persistent
disagreement is common only when evidence is scanty and it's hard to think of
examples where the scientific community sticks to objectively bad theories
(as opposed to holding them in the past when evidence is inferior). This
would indicate that scientists, at least while working in their fields, are
more "rational" than the population as a whole in the sense that they are
closer to Bayesian agents.