Re: Opinions as Evidence

Robin Hanson (hanson@hss.caltech.edu)
Fri, 28 Mar 1997 10:37:50 -0800 (PST)


I asked for examples of the following:
>1. There are (at least) two communities, each aware of the other's views.
>2. These views are relatively stable and substantial different.
>3. These views reflect actual beliefs, not just verbal posturing.
>4. The dispute has lasted long enough so both sides know that
> both sides know that both sides know (etc.) ... that they disagree.
>5. Both sides see the issue as a dispute over fact, not values, and
> believe that the other side does as well.
>6. We can't easily dismiss one group as small and primarily composed of
> people who are much more "cognitively broken" than most.
>7. We can't easily dismiss the type of issue as exceptional in
> inducing "broken cognition" even in ordinary people.

Curt Adams said thought "everybody here" was "passably familiar with
these issues enough so to say" that most of Michael Lorrey's list
satisfied all of these conditions. Specifically, he included:

>>the Big Bang

which I'm not aware has substantial disagreement.

>>Capitalism
>>COmmunism
>>Socialism
>>SDI

which I'm not convinced here that most people do see as disputes over
facts rather than values.

>>JFK assasination
>>Roswell Incident
>>MLK assasination
>>The Philadelphia Experiment
>>the Rosenberg Trial
>>the Alger Hiss trial
>>Contragate (esp. Ollie North)

where I know lots of people suspect foul play, but I'm sure most people
really disagree on estimating the chances of such.

I really think these three word titles are not a sufficient
description of each dispute to evaluate their status regarding this
issue. More details please.

>All the literature on human reasoning I'm aware of indicates the people have
>many blatant errors in their reasoning, especially with respect to
>probabilities (particularly important to Bayesian inference, which is based
>on modifying probabilities of conflicting theories.) What evidence is there
>that *any* people, never mind ordinary ones, act like good Bayesian agents,
>apart from in science?

With "like" being the key word, lots of evidence. Of course there is
also evidence suggesting various sorts of anomolies & deviations. But
it is a stnadard first approximation. The question is: how good an
approximation is it in different contexts?

Robin D. Hanson hanson@hss.caltech.edu http://hss.caltech.edu/~hanson/