How well can we assign jobs to machines, if these machines are the
property of agents who pursue their own goals and who may give us false
information about the machines? And how bad can a schedule become, if
the jobs are controlled by selfish agents, and the agents all seek out a
machine where their jobs will be completed the fastest?
In this talk we concern ourselves with these questions. We discuss the
question of how a mechanism can optimize a goal function, even if it has
to deal with agents who have their own and possibly conflicting
interests. Furthermore we give lower and upper bounds for the quality of
schedules that are constructed in a selfish manner, the so-called Price
of Anarchy.