1: \begin{abstract}

2: We consider a version of large population games whose agents compete for

3: resources using strategies with adaptable preferences. 

4: The games can be used to model economic markets, ecosystems 

5: or distributed control. 

6: Diversity of initial preferences of strategies is introduced by randomly 

7: assigning biases to the strategies of different agents. 

8: We find that diversity among the agents reduces their maladaptive behavior. 

9: We find interesting scaling relations with diversity 

10: for the variance and other parameters such as the convergence time, 

11: the fraction of fickle agents, and the variance of wealth, illustrating 

12: their dynamical origin. 

13: When diversity increases, the scaling dynamics is modified 

14: by kinetic sampling and waiting effects. 

15: Analyses yield excellent agreement with simulations.

16: \end{abstract}

17: