Saturday, April 19, 2008

Revealed Preference?

(1) Quoting an Open Letter to ABC:
The debate was a revolting descent into tabloid journalism and a gross disservice to Americans concerned about the great issues facing the nation and the world.... For 53 minutes, we heard no question about public policy from either moderator. ABC seemed less interested in provoking serious discussion than in trying to generate cheap shot sound-bites for later rebroadcast.

Alex Tabarrok responds with a lame 'revealed preference' argument: "who would want to rebroadcast something the public didn't want?" He really needs to read, um, Alex Tabarrok:
Early on Slee makes a good point about preferences and outcomes:

"The prisoner's dilemma shows how, as soon as one person's choice alters the outcome for another person... choices do not reveal preferences... instead of thinking about choices as revealing preferences, it pays to think of choices as 'replies' to the actions or likely actions of others. The best choice you can make is the best reply to the likely actions of others."

Given that public debate is so degraded and sub-rational, partisans will reach for every weapon in their rhetorical arsenal, including 'gaffe bombardment'. No one dares risk unilateral disarmament. But it obviously doesn't follow that they prefer this situation to some alternative where gaffe bombardment was safely off the table. Political discourse could very easily be a prisoner's dilemma in this way.

(2) Another problem I'd like to consider for sloppy appeals to 'revealed preference' is that of intra-personal conflict. Consider the unwilling addict, who is compulsively moved to seek drugs, even though he would prefer (on reflection) to be able to withstand this compulsion. Robin Hanson denies that there is any normatively relevant structure to our preferences. Instead, in a case like this, we can simply watch the addict's behaviour to see which preference is the more weighty. 'Might is right.'

But this is plainly misguided. Mere behavioural drive has no normative impact; the mere fact that my body is disposed to move in such-and-such a fashion is at most defeasible evidence - and does not strictly entail anything - about my mental states, or what I really desire in any normatively significant sense (viz. reflective endorsement). You can see this by imagining a brain implant that grants a mad scientist remote control over my body. It's not so different in principle if the source of my unfreedom is internal -- an inner demon such as mental illness, addiction, etc., may be every bit as constraining as an external obstacle. My resulting action may be no more an instance of my "getting what I most want" than when the mad scientist was controlling me.


  1. It was not partisans who chose the low topics - it was the debate host ABC. If there were gains to coordinating on high topics, the host could provide that coordination.

  2. Right, and that's precisely the quoted complaint: as host, ABC failed to provide the coordination that would make us all better off. (Alex responds that we're going to replay the gaffe moments, thus "revealing" that they are what we really prefer to see. My current post then explains why this is a non-sequitur.)

    Hopefully future candidates will punish poor hosts by refusing to appear again on their shows. It's not inevitable that debates must be this bad.

  3. I was referring to your speculation that we could have a prisoner's dilemma like situation, which is where coordination fails. My guess is that Alex is right - this was the debate that TV viewers wanted to see.

  4. We don't really disagree. I suggested that political discourse is like a prisoner's dilemma in terms of the pay-off matrix (i.e. unilateral disarmament is imprudent, even though we would both prefer that both disarm). I certainly wasn't suggesting that co-ordination was impossible.


Visitors: check my comments policy first.
Non-Blogger users: If the comment form isn't working for you, email me your comment and I can post it on your behalf. (If your comment is too long, first try breaking it into two parts.)