The Epistemic Humility of Representative Deliberation

• In Event: Ideational Contributions to State-Building

Abstract

It’s commonly thought that representative democracy is a second-best to a full, Athenian-style democracy which demanding the participation of each citizen during legislative sessions. In particular, representative democracy is commonly seen as epistemically inferior to a pure democracy, since while pure democracy draws on the wisdom of the whole, recruiting an open deliberation of all members, representative democracy draws only from a few, thus degrading the epistemic quality of the deliberative outcome. Since the proposed mechanisms for the epistemic success of democracy recruit models where deliberation produces better results when more agents are involved, it’s natural to think that, when representatives are no better judges of evidence than ordinary citizens, pure democracy is epistemically better on the merits.
We consider whether pure democracy really does have this advantage and find that, in many ways, it does not. Indeed, we show that representative deliberation is frequently epistemically superior to open deliberation, both in ideal and non-ideal circumstances. Yes, representation is better when haste matters, but surprisingly it’s also advantageous where expertise is helpful, even when the representatives themselves don’t possess any special epistemic gifts. While one might suppose that open deliberation does as well or better at handling subtle problems than a more restricted forum with fewer representatives, our results show otherwise. And to boot, these models preserve the democratic character of the system through epistemic proceduralism. We argue that representation’s bad rap is both epistemically and democratically unearned.
Directly evaluating two deliberative systems like this is a difficult task. The use of a computational model here is meant to “open the hood” of these different methods. It does so by outlining a clear procedure that incorporates important extant normative and empirical criteria. We start from a sensible departure point and see where it leads. This paper brings together epistemic considerations of democracy with expansive work on political representation and deliberation to compare the epistemic advantages of both open and representative deliberation. One important upshot of these findings is the way that representative systems can achieve better outcomes by not overfitting the decision function. Whereas open deliberation is apt to read too much into the evidence presented—thereby overfitting the deliberation to inevitably incomplete evidence—the representative schema utilizes a more epistemically humble approach and in so doing achieves superior epistemic outcomes.

Author

• William (Zev) Berger, UNC