I'm referring to a specific part of bayesian updating, conservation of expected evidence. Specifically:

On pain of paradox, a low probability of seeing strong evidence in one direction must be balanced by a high probability of observing weak counterevidence in the other direction.

This rule did not seem respected in what little I've seen of interactions between you and Eliezer, and I was looking for external feedback and evidence (one way or another) for this hypothesis, to see if there is a valid body of evidence justifying the selection of this hypothesis for consideration or if that simply happened out of bias and inappropriate heuristics.

I suspect that, if the latter, then there was probably an erroneous pattern-matching to the examples given in the related blogpost on the subject (and other examples I have seen of this kind of erroneous thinking).

I don't know how to submit this stuff for feedback and review without using a specific "accusation" or wasting a lot of time creating (and double-checking for consistency) elaborating complex counterfactual scenarios.