We study a model of a competing population of N adaptive agents, with simil
ar capabilities, repeatedly deciding whether to attend a bar with an arbitr
ary cutoff L. Decisions are based upon past outcomes. The agents are only t
old whether the actual attendance is above or below L. For L similar to N/2
, the game reproduces the main features of Challet and Zhang's minority gam
e. As L is lowered, however, the mean attendances in different runs tend to
divide into two groups. The corresponding standard deviations for these tw
o groups are very different. This grouping effect results from the dynamica
l feedback governing the game's time-evolution, and is not reproduced if th
e agents are fed a random history, (C) 1999 Elsevier Science B.V. All right
s reserved.