Bryan Caplan  

The Pledge for Efficiency versus the Attack on Signaling

Notes from the Monterey Tea Pa... Debating Health Care Reform...
Robin Hanson now has an official catechism.  A key passage:
I need not accept all clients, but for the clients I do accept I work to suggest deals that, if accepted, would get them more of what they want, relative to no deal.  I resist temptations to slant my advice to give hidden benefits to myself or my associates.  I accept my client's wants as they are, and unless asked do not preach to them on what they should want.  When actions conflict with words, I mostly infer wants from actions.
At the same time, though, Robin has waged a long campaign against the inefficiencies of signaling.  For example, he argues that we waste a ton of resources on health care because we're all trying to "show each other that we care."

I think there's a major conflict here.  Consider: According to Robin's pledge, under what circumstances is he allowed to advise people to do less signaling?  People would actually have to tell him that they want his help to reduce their signaling!  But if people are in fact stuck in an inefficient signaling equilibrium, they won't admit that they've got a problem!  After all, it sure doesn't sound caring to say, "Let's figure out how to to stop wasting resources on health care in order to show how caring we are."  The people who need Robin's help the most won't admit that they've got a problem.

The only out for Robin, as far as I can tell, is to lean on the sentence, "When actions conflict with words, I mostly infer wants from actions."  But this escape hatch is precisely the kind of "elastic clause" that Robin fears will lead potential clients to distrust him.  In fact, if you ignore words and "infer wants from actions," you have virtually unlimited wiggle room.  The louder people protest, the more you can say, "Methinks you doth protest too much."  I.e., "The more you spend on and praise health care, the more I infer that you want me to help you spend less."

I think I've got Robin cornered here.  Anyone want to help him out?

Comments and Sharing

COMMENTS (7 to date)
Peter Twieg writes:

If Robin was competent in recognizing when people were signaling, why would people distrust him? Presumably this outcome would only result if the signal-detection process employed generated false positives.

In that case, perhaps it could be said that on a pragmatic level Robin shouldn't try to extract signals. But this concern could apply to any case where a possibly-faulty signal-detection heuristic is used. If Robin's statement simply were amended for the possibility of human error on his part, would that really be such an important concession?

Sohaib writes:

The only out I see is for Robin to make public his views on signaling through links to research showing this is happening, as in "hey check out this neato article!" Or maybe just throwing his original thoughts out without having them directed at anyone as in "hey this is just so I can remember I thought of this." Haven't completely thought through if these strategies completely fulfill the criteria. Can't think of anything else to help him out either. I sure would like to considering how enlightening his views on signaling have been to my world view.

Robin Hanson writes:

I promise to find deals to better get them what they want. But I don't have to explain exactly why the deals achieve this, nor do I have to force them to overcome self-deception about what they want. People rely on experts all the time without knowing exactly why what experts advise is helpful.

I do have to worry that folks might not accept my deal because that would signal something bad things about them. While they might approve certain deals that would hurt them in order to signal, I can simply not suggest such deals to them.

Ben writes:

May he could just ask them, 'do you care more about efficiently using resources to maximise social welfare, or looking good?'?

It would work if he dropped the preference for judgements based actions over those based on words.

mobile writes:

But Robin also ought to think carefully about whether he should give more weight to actions than to words. Suppose that the best evidence shows that the minimum wage increases unemployment among minority teenagers and demonstrably makes that group as a whole worse off. Now suppose that Robin's client says he wants to make minority teenagers better off, and the client is also in favor of minimum wage laws. Is Robin to infer that the client actually wants to make minority teenagers worse off? Or just that the client has a mistaken belief about which actions are aligned with his wants?

Robin Hanson writes:

mobile, you seem to be assuming that I'd take a vote for min wage laws to be a strong indication that the voter wants to hurt minority teens; I wouldn't.

Kurbla writes:

I think it is minor thing. I do not have idea who are Hanson's clients, but if they've read this

    "I accept my client's wants as they are, and unless asked do not preach to them on what they should want. "
and they are interested in that aspect of his knowledge, they can simply ask "Robin, give me all advices you can, including what should I want." In fact, Robin can stay faithful to his principles and put these items on his menu

    * what should you do - $100 per advice
    * what should you want - $100 per advice
    * both (20% discount)
    * neither one (80% discount)
I can imagine the following conversation:

Client: Mr. Hanson, I'll invest $3000 in Mexican paprika business, spend holidays in Nebraska, marry woman with very strong, muscular legs, and wear Hawaii shirts all the time. Please, could you tell my what should I want so this choice of my actions is the most efficient one.

Hanson: (thinking thinking) ehmm ... yes! According to this paper written by young American economist T.C. you should want to maximize the chance that your oldest child will become slightly overweight police officer and divorce in early thirties from a Linux geek. If you'll want that, the mentioned actions will be the most efficient. $100 please.

Comments for this entry have been closed
Return to top