Scientists are gamblers.  The gamble grant money (and their reputations) on
experiments in the hopes of making scientific discoveries. At some of the
tables, the rules of the game (experiment) require the scientist to place
equal
bets on a large number of outcomes (e.g., gene expression microarrays). Games
can have a more sequential flavor at other tables, with bets being
re-considered
after each round of experiments (e.g., virus replication experiments). 
Sequential experiments of this nature can be mathematically posed as
multi-armed bandit problems.  I'll discuss some of our recent results and open
questions.