Yes, though I’ll flag that we don’t have robustness with respect to approximation on the agreement condition (though we do have other ways around that to some extent, e.g. using the Solomonoff version of natural latents), and those sorts of updates are the kind of thing which I’d expect to run into that robustness problem.
...how is that a way around it? It looks to me like, there, you both for sure observe the same datapoints, even if you’re allowed to condition on extra stuff by adding the additional length required to the epsilons.
Yes, though I’ll flag that we don’t have robustness with respect to approximation on the agreement condition (though we do have other ways around that to some extent, e.g. using the Solomonoff version of natural latents), and those sorts of updates are the kind of thing which I’d expect to run into that robustness problem.
...how is that a way around it? It looks to me like, there, you both for sure observe the same datapoints, even if you’re allowed to condition on extra stuff by adding the additional length required to the epsilons.