The Low Golden Willow responded:
>I'm not sure the benefit would work at this scale. Between limited
>choice of your representative, and _one_ random person from your
>district, I'm not sure which is the rational choice. Statistics work
>best on a large scale.
You could replace yourself with a jury of twelve voters instead.
Anyway, you've got statistics even in this case. You're looking at a
legisture of N people composed of two random populations: N-M people
picked the usual way and M people picked this new way. I don't see why
you'd expect a nonmonotonic effect, so that you'd prefer M = N to M =
0, but prefer M = 0 to M = 1.
Robin D. Hanson hanson@hss.caltech.edu http://hss.caltech.edu/~hanson/