[shiandow]

> our entropy calculation will now have to use the posterior means

Now hang on for a bit, you can't just plug in averages.

At least that's what I initially thought, but in this particular instance it works out correctly because you're calculating an expected value of the entropy from the two possible outcomes and there the posterior mean is indeed the correct probability to use.

You do have to take the prior into account when calculating the posterior distributions for B, but that formula is in the article.