Here. Robin Hanson makes a mistake that is common to "rationalists": He examines every thread of human behavior for "reasonableness."
It is the fabric of human behavior that matters, not each thread. Any thread, if pulled out of the fabric, might look defective under the microscope of "reason." But pulling threads out of a fabric -- one at a time -- can weaken a strong and richly textured tapestry.
Whether a particular society is, in fact, a "strong and richly textured tapestry" is for its members to determine, through voice and exit. The "reasonableness" of a society's norms (if they are voluntarily evolved) should be judged by whether those norms -- on the whole -- foster liberty (as explained here), not by the whether each of those norms, taken in isolation, is "reasonable" to a pundit inveighing from on high.
UPDATE (11/01/07: Hanson has updated his post (first link above). But he digs himself a deeper, rationalistic hole when he says
I'll now only complain about [Russ Roberts's] bias to hold his previous beliefs to a lower standard than he holds posssible alternatives.He should complain, rather, about his own, too-easy willingness to reject the wisdom of inherited beliefs on the basis of statistical analysis.