I'd like to add a search engine, but tonight I find that the Pico engine does not work because the entry point fails, probably because of some Indiana University security feature, since an entry point in Geocities does work. Maybe a workaround would be to post a duplicate of this file in Geocities and use it as the entry point, but it is too late to do that tonight.

ת Revolutions and Reputation. My experience with Chancellor Brehm made me think of a new reason why revolutions might be so sudden and unanimous. The best-known idea currently is the one that, I think, Timur Kuran uses in his book: that it is so dangerous to be the only person to start a revolution that nobody will do so until it is obvious that the revolution will be successful--at which point it usually will have massive support. Very similar is if you hope to get a reward from the new regime (rather than fearing a punishment from the old). The new reason I've thought of is not fear, but the desire to look powerful. If you know that the revolution is going to succeed, you should back it, so people will increase their estimate of your power.

Here's how a formal model might work. There are N people in an organization, each of whom is either powerful or not powerful. Whether someone is powerful is not common knowledge, and each person has prior probability alpha of being powerful. Simultaneously, each person decides whether to attack the current leader (who is not one of the N) at cost C>0 or do nothing. Attacks are publicly observed. If even one powerful person attacks the leader, the leader falls, but attacks by unpowerful people have no effect.

A person's payoff is decreasing in expenditure and increasing in other people's perception of his power-- the posterior level after it is observed whether he attacks or not.

Assume that if one powerful person decides to attack for exogenous reasons, each of the other people discovers this with independent probabilities beta (they observe the exogenous reason and can deduce the attack). These exogenous reasons occur with probability gamma. What will the other powerful and unpowerful people do if they discover an attack or do not discover one?

My guess is that in equilibrium if C is low enough anyone who discovers that a powerful attack is occurring will also wish to attack. Anyone who attacks will have a posterior greater than alpha, and anyone who does not attack will have a posterior less than alpha.

There is also an effect that can go the other way: that a person has independent reasons for wishing the leader to be defeated. If this is so, then a powerful person might wish to launch an attack if he thought no other powerful person was attacking, but might not wish to duplicate the effort if another powerful attack was going to be made.

[ permalink, http://php.indiana.edu/~erasmuse/w/03.11.20a.htm ]

To return to Eric Rasmusen's weblog, click http://php.indiana.edu/~erasmuse/w/0.rasmusen.htm.