I thank Mary Scudder for her generous comments on Deliberation Naturalized and for the opportunity to expand on them here. I cannot respond to all her specific points, but on the claim that the book neglects the need for critical reflection, see chapters 4 and 7; and on the claim that the book endorses exclusion, see the discussion of “democratic weights.” True, I exclude clone claims from my updating but only after paying them enough attention to recognize them as such. Structurally, that is no different from Scudder endorsing listening to someone just long enough to realize that they are a conspiracy theorist and then ignoring them.
Her major point is that the Bayesian framework underlying my model of internal deliberation (henceforth, BID) is a form of majority tyranny that overrides minority voices, enabling epistemic injustice in the aggregate. I agree with the weaker claim that it can fail to be fully informative—unless further specified, as I do here.
Take this simple example. A large population that is 90% white and 10% Black is assessing whether the proposition, “Police plant evidence,” is generally true or false. Assume that everyone is perfectly connected in a deliberative network with full transmission and no information loss (the book offers evidence that, with some addressable caveats, this is approximately the case). Assume police never plant evidence on whites (so their credence in the proposition is nearly zero), but they always do on Blacks (so their credence is nearly 100%). Once everybody updates in light of everyone else’s opinion, across the entire community the credence in this proposition will be around 10%. So everyone thinks the probability of police planting evidence is 10%—which, on the facts stipulated, is epistemically accurate across the entire community. But what that misses is the racialized distribution of police plants. It is the right answer to the wrong question.
Scudder’s critique of the aggregative aspect of BID is thus ultimately one about BID failing to be fully informative. This is not just an epistemic problem (white people not knowing that police plant evidence on others is a fact that washes away in collective updating). It has serious political effects as well, undermining the community’s capacity to address these issues democratically.
But through informal deliberation citizens can easily find out that the proposition they had been assessing was poorly formulated. If my Black friend tells me that police plant evidence when the rest of my white friends do not, I should take this as a sign that he could have exclusive information that neither I nor my other friends hold. Through further engagement with him, I would learn that this sort of practice targets Blacks only. Deliberation can thus bring about a respecification of the very general proposition into two more specific ones: “Police do not plant evidence on whites,” and “police do plant evidence on Blacks.” Subsequent deliberations and updating on those more specific propositions can then easily fill in important details that were originally left out. But this can happen only if democratic deliberation is an informal practice that happens anytime, all the time. The constant flux of deliberation makes it a harder object to study. Yet it also equips deliberation with important mechanisms of self-correction like the one just mentioned.
