Betting, bribery, and bankruptcy - A simulated economy that learns to predict

Derby is a collection of independent computational entities that exchange money for work and information in the course of solving some problem. The key to building an agoric system is to design a monetary incentive structure that forces the individual entities to cooperate and work on the user´s problem. While building Derby the authors discovered that it is very useful to think of the entities as being opportunistic and uncooperative. In Derby, information is traded in marketplaces, with sellers issuing predictions and placing bets on their correctness at predicting incoming data streams. Buyers submit bids of how much they are willing to pay for each dollar of bet placed. A sealed-bid second-price double auction determines which bidders are accepted. Later, the buyers report how happy they were with the information they bought, and this determines each seller´s winnings in the parimutuel betting pool.<>