02 April 2019

The Random Ten Percent

Universities have standards they use to judge whether applicants deserve to join.
We have elections to decide who should serve on city council or in state legislatures.
Venture capitalists and banks have criteria for deciding whether or not to fund a startup or business.

Given that all of those processes can be gamed and all of those processes depend on judgement against a set of criteria that may turn out to be a really poor predictor of success, we should set aside 10 percent of all slots for a randomly chosen group.

This would have a host of benefits.

One, given these people would be randomly chosen from the population, they would be more representative. Right now, more than 90% of money venture capitalists give is given to men. Blacks and Hispanics are under-represented at universities. The folks who win elections tend to be good looking. The processes we use are skewed and a random process would help to partially offset that.

Two, it would be a chance to test the null hypothesis. All of these institutions have theories about what predicts success. Inevitably those theories have flaws. One way to discover those flaws is to compare the success rates of those randomly chosen with those intentionally chosen. To the extent that unpredictable things happen, it will be a chance to update the choice algorithm, throwing out criteria that turn out to be unrelated to success and adding criteria that - surprisingly - turns out to be very predictive.

Three, it would be an experiment to test the notion that access to these institutions will improve anyone's life, and not just the lives of those carefully chosen. Perhaps 4 years of Harvard will improve the life of a kid with a 2.4 GPA at least as much as it will a kid with a 4.2 GPA. Perhaps the community will be improved as much by the startled bartender or civil engineer who learns she or he has been chosen to serve in the state legislature by lottery as any politician capable of getting big financial donations. Institutions are powerful and it is important to better understand how much of the benefit they yield comes from their careful selection of who gets access to them vs. who those people are. That is to say, we need to understand when the goal should be to widen access to these institutions as a means to improve everyone's life or when we should narrow access to make sure that only people with a certain potential should be given access.

No comments: