Robin Hanson's Work In Progress with Abstracts

Social Science

A ``bar bet'' is a simple bet on a risk-irrelevant topic negotiated in an informal but non-private social context. We might explain such bets as costly signals in a signaling game, made either to credibly signal to an audience containing potential associates (employers, lovers, friends, etc.) that one is informed about a topic, or to persuade a decision-maker regarding the probability of some outcome.

A simple mechanism is presented to allow one to buy a verifiable secret known by more than one person. In the unique sequential equilibrium, the secret can be bought for an arbitrarily small sum, even if the secret getting out would hurt each holder by an arbitrarily high amount.

Consider a two-stage game. 1) players simultaneously choose whether or not to verifiably reveal their value. 2) hold a sealed-bid first-price auction. Assuming i.i.d. private values, it is an equilibrium for no one to reveal in stage 1. That is, it doesn't pay to reveal your type before the auction, if no one else does.

The possibility of replacing current monopoly public law enforcers with competing private enforcers has been explored by Becker and Stigler, who suggest offering fixed bounties per conviction, and David Friedman, who suggests that the law declare functions relating punishment levels to enforcer-chosen probabilities of conviction, and that victims own the rights to enforce their complaints. This paper suggests instead that the law fix victim restitution and appoint criminal agents to choose enforcers. Not only should this institution require political oversight to know the least about law enforcement, but it can also lower costs by offering criminals better incentives to work to aid, rather than avoid, convictions.

Half-Baked Ideas