Wednesday, November 20, 2013

Trade-offs - Diversity vs. Cohesiveness, Efficiency vs. Robustness

From The Paradox of Diverse Communities by Richard Florida, reporting on the results of a computer simulation of diversity and cohesiveness. I have long argued that our policy folk are misguided in their blind pursuit of diversity and integration, treating these as outcomes to be valued in their own right rather than as dependent variables arising from more important variables.
After 20 million-plus simulations, the authors found that the same basic answer kept coming back: The more diverse or integrated a neighborhood is, the less socially cohesive it becomes, while the more homogenous or segregated it is, the more socially cohesive. As they write, “The model suggests that when people form relationships with similar and nearby others, the contexts that offer opportunities to develop a respect for diversity are different from the contexts that foster a sense of community.”

The graph below, from the study, plots quite plainly the negative relationship between community cohesion and diversity.

These findings are sobering. Because homophily and proximity are so ingrained in the way humans interact, the models demonstrated that it was impossible to simultaneously foster diversity and cohesion “in all reasonably likely worlds.” In fact, the trends are so strong that no effective social policy could combat them, according to Neal. As he put it in a statement, “In essence, when it comes to neighborhood desegregation and social cohesion, you can't have your cake and eat it too.”
Two key issues in the field in general and reflected in this study. Diversity is not a monovariable item. While we usually speak of diversity in terms of race and then gender, it is much richer than that including religion, personality orientation, language, morbidity, class, accent, region, ethnicity, culture, profession, income, age, etc. One person can wear many hats. Our idea of diversity is terribly anemic and we fail to identify what has always been true in the US, that the important thing is not diversity per se but tolerance.

It has long been known that segregation arises primarily from free people making individual choices informed by preferences. Aversive racism is only a miniscule issue, swamped by issues such as religion and class and education valuation, etc.

In any major city, the diversity of neighborhoods is usually immense. There are clusterings by race and class but also by age and religion and profession and hobby, etc. With everyone wearing so many identity hats there is also an awful lot of graduation between segregated clusters.

The second key issue has long been noted in international economic development - that trust and mutual reciprocity are characteristics shared in most highly productive OECD nations. Think of the Scandinavian countries, the Netherlands, Germany, South Korea, Japan, etc. All notable for highly productive economies over long durations. All different cultures and histories but all having developed high levels of intra-cultural levels of trust and reciprocity. And all notably homogenous. The US is the only large OECD country with a high degree of heterogeneity (on all vectors of diversity) which also has a high degree of trust and mutual reciprocity - one of its many aspects of exceptionalism.

High degrees of trust and mutual reciprocity foster a high level of efficiency. Sharing the same cultural construct and language makes communication easier and more precise, makes collaborative efforts more efficient, makes risk management much easier, reduces transaction costs and insurance costs, etc.

So what Florida is reporting in this article is merely a conformation of what has been long known in different fields. We know there are trade-offs between diversity and tactical efficiency and social cohesion.

The really interesting questions are somewhat different. Increased diversity comes at the cost of reduced efficiency in the short term. But every system needs diversity in order to evolve over the long term. A purely efficient system will run at maximal productivity until circumstances change which make the system obsolete. It tends to binary - highly productive or non-productive. If you introduce some variation to the system, the system tends to evolve. New ideas are introduced, an error in the production process turns-up a better way of doing things, a botched batch of materials yields a new product.

The net result is that you need at least some diversity in the system in order to foster evolution and robustness at the expense of pure efficiency. In other words, you have to sacrifice some short term efficiency (by increasing variation) in order to achieve long term robustness (by evolving the system).

To me, the interesting question is: How much diversity is necessary to optimize short-term efficiency while ensuring long term robustness and evolution? Too little diversity and you suffer episodic painful adjustments. Too much diversity and you lose social cohesion before the long term arrives. How much is too little? I don't know. Based on anecdote and experience, my sense is that less than 5% diversity is, barring special circumstances, easily ignored and marginalized. Upper boundary? I am guessing somewhere around the 15% level is the boundary. Pure speculation though.

Florida and his ilk focus too much on ideological concerns such as diversity as a desirable trait in itself, instead of looking at the dynamic system as whole and over time. It is what leads to so many government policies that end up having negative unintended consequences and not achieving their stated goals.

No comments:

Post a Comment