A Bayesian approach to legal gay marriage

Ezra had this good bit:

Due to D.C.'s strange system of governance, the District's laws are
subject to approval by Congress. If D.C. passes a gay marriage
ordinance, the House Committee on Oversight and Government Reform and
the subcommittee that handles District matters will have to either
reject D.C.'s decision or accept it. If they reject it, the outrage
from gay donors and activist groups will be overwhelming. If they
approve it, even on federalist grounds, the Right will argue that
Congress has literally approved gay marriage.

The interesting question is why there is so much opposition to legal gay marriage (which I favor).  You can cite various evil opponents and their evil motives, but there are many good people who aren't all that enthusiastic about the idea.

Jennifer Roback Morse offers the argument that gay marriage requires ongoing statist intervention (as does the notion of a corporation) and will drive the Mennonites from Quebec.

I have a simple hypothesis about the cross-sectional econometrics.  If you take the heterosexual couples who engage in the practice which is sometimes "associated" with male gay marriage, I predict those couples will favor legal gay marriage to an astonishingly high degree.  Their marriage is already "affiliated" with that practice, and so the notion of legally married gay men (and the practices which go along with that) does not constitute an extra and unwanted affiliation for their marriage ideal.

Now, if you are a rational heterosexual Bayesian and neither engage in that associated practice nor favor legal gay marriage, and then you learn about these cross-sectional econometrics, what should you infer about the correctness of your point of view?

If you're still not sure, reread Gulliver's Travels and get back to me.

Addendum: Comment 51 is apparently from Roissy.