Observations of Society
Main Menu Previous Topic Next Topic
Turnover can Contribute to Diversity
So far, our society has been composed of a fixed set of individuals. We now consider a slightly more realistic situation. We suppose that agents are not necessarily fixed, but can be replaced in the organization by outsiders. We assume that in each time interval, each agent has an equal and independent probability of quitting or being replaced with a new agent having random beliefs. We call this probability the "turnover rate." Turnover brings in new agents, increasing the diversity of beliefs. Thus it increases organizational exploration. What effect will turnover have on the growth of knowledge? We will see that turnover can potentially improve the amount of knowledge in society. In the graph below, we plot the code knowledge in time period 20 under different levels of turnover. The code learning rate has been fixed at 0.5.(1)
We see that for low agent learning rates, turnover will not improve code knowledge. However, for moderate socialization rates, some amount of turnover will improve the amount of knowledge in period twenty. Turnover introduces some level of exploration and improves long-run knowledge. Note that the new agents have zero knowledge on the average. The improvement in code knowledge comes not from the sheer knowledge of the new recruits, but solely from their diversity. Newcomers are more different from the code, and so more likely to contribute new information. Note that we have assumed that the replacements are selected with random beliefs. If an organization selects recruits that already share a lot of beliefs in common with the societal code, the gains to accuracy will not be as large. If it selects for diversity, gains could be greater. Clearly, when an organization chooses its hiring strategy, this decision influences the levels of exploration and exploitation that occur.
1. Each data point is based on 500 trial societies with N = 50, M = 30.
Previous Slide Next Slide