Heh, sorry, kind of skipped the preamble there.

Yes, the post was in agreement with you, and attempting to visualize / illustrate / imagine a potential way the model could be shown to be flawed.

As for feeling "anti-bayesian", the idea that a set amount of responsibility exists to be distributed over actors for any event seems completely uncorrelated with reality and independent of any evidence. It feels just like an arbitrary system of categorization, like using "golborf" as a new term for "LessWrong users that own a house, don't brush their teeth daily, drink milk daily, enjoy classical music and don't work in IT-related fields".

That little feeling somewhere that "This thing doesn't belong here in my model.", that there are freeloading nodes that need to be purged.