Finally, just to make sure, the deterministics in your model are not random samples from n1 and n2, they are just the values that n1 and n2 took at each step in the sampling. In other words d2 and d3 and n2 are exactly the same as each other, and n1 and d1 are the same as each other. None of these variables is supposed to change during ppc (and they don’t, as you can see from them having the exact same mean and std, except for the offset of 1 in those that you changed).
Correct. I created this little model to isolate this weird behavior. As you point out, there are only two RVs in this model: n1 and n2. Prior to the increment, n1 and d1 have exactly the same distributions, and n2, d2, and d3 have exactly the same distributions. The trace is then modified to increment the distributions of n1 and d2. I expected the posterior predictive sample to propagate those changes to the downstream deterministics d1 and d3. It did propagate the change to d1, but somehow not to d3. And my question is why? And is this a bug or the intended behavior?