SUG 9.5 suggests a degenerate model

In section 9.5, the full Bayes for the correlated topic model uses a mvnormal variate transformed by a softmax as probabilities. The problem is that the inverse softmax doesn't work as a link function because it isn't unique. My impression from a brief perusal of Aitchison & Shen 1980 is that the solution is to fix one element of the link-scale parameter vector to zero, which identifies which unique inverse-softmax we're dealing with as our link function.

The model of Blei & Lafferty (also cited in the SUG) appears to circumvent the problem by using the maximum a posteriori estimate for the link-scale mvnormal. Although the posterior is degenerate, it still has a maximum thanks to the prior, and once the estimate is fixed to that maximum the degeneracy disappears.

Start by reading section 9.5 and the cited Aitchison & Shen and Blei & Lafferty references. Verify whether the inverse-softmax parameterization is identifiable and determine the correct documentation change; done means the section no longer presents a degenerate model without an explicit resolution.