Stratified sortition for balancing representativeness and efficiency in large participatory institutions
By Seth Frey
As a self-governing institution grows, it becomes a target for elite capture. Sortition, or governance by lottery, is a mechanism displacing elites and for preserving democratic rule. A large legislature with randomly appointed representatives, or a small board with oversight over that legislature, can foreclose nepotism, rent-seeking, and other abuses.
However, sortition in large institutions suffers from a tradeoff, particularly in resource-constrained institutions. A small body of randomly appointed representatives can make decisions quickly and efficiently but, due to the high standard error of small samples, may not be representative. Conversely, the central limit theorem can ensure that a large body is representative, but it will need much more time and structure to reach decisions.
Is it possible to ensure randomly appointed bodies that are both representative and efficient? We propose a stratified sortition, in which appointments are biased to ensure representative bodies. An institution with intentionally biased random draws may be able to ensure appointed bodies that are both small and representative. Of course, it is important that the member data defining representativeness is robust to gaming. Stratified sortition will be less effective at preventing capture if members can tune their own "metadata" to improve their chances of selection.