We study several natural problems in distributed decision-making from
the standpoint of competitive analysis; in these problems incomplete i
nformation is a result of the distributed nature of the problem, as op
posed to the on-line mode of decision making that was heretofore preva
lent in this area In several simple situations of distributed scheduli
ng, the competitive ratio can be computed exactly, and the different r
atios can be used as a measure of the value of information and communi
cation between decision-makers. In a more general distributed scheduli
ng situation, we give tight upper and lower bounds on the competitive
ratio achievable in the deterministic case, and give an optimal random
ized algorithm with a much better competitive ratio.