This I can answer in a pinch and I would agree. Yes, indeed if we were to base the reconstruction of any animal upon a taxon from which it diverged millions of years ago, it would be really difficult to arrive at a finished product resembling anything at all like that of the intended animal.
As I understand it the point of the post and the book as a whole is to point out that we can’t know exactly what a fossil species looked like. For example, organic material rarely preserves. We can look to skeletal morphology and to descendent taxa for clues but an exact reconstruction that we can be 100 per cent confident about is impossible. So, for example, if you decide to add strong musculature around the cheeks to, I don’t know, say a sagittal crest, based upon rugosity on the mandible, you’re making an educated guess. Or if for example you include feathers on a dinosaur taxon based upon its phylogenetic relationship to birds, you’re making an educated guess as well but that guess could be wrong. Would it have penguin-like feathers or peacock-like feathers? What aspects of the fossil can we look at to infer what sort of feathers if any feathers it had?
Basing a baboon upon a dinosaur is just an extreme example, which I believe was probably used for effect, to make a point. I haven’t read this book so I can’t be sure. The authors note in the article that they aren’t arguing that reconstructions are arbitrary but that certain aspects of certain reconstructions of certain taxa are less informed:
“What we do say in the book (and as I said in my talk) is that while the positions of muscles might be known (or reasonably inferred), their size is typically something we can’t be so sure about. Hence we now have fat-tailed theropods (Persons & Currie 2011) – as opposed to the slimmer-tailed ones produced by Bakker and Paul – and also a degree of leeway as goes how chunky and how muscular animals like Tyrannosaurus might have looked overall (Hutchinson et al. 2011)”.
In using the baboon-dinosaur as an example, they are not only being intentionally creative, as is part of the point of the book, (and hyperbolic in this instance), but are also highlighting, I believe, the need to base palaeoillustrations, that they think have become ossified in canon, upon known myology and osteology. If you don’t do your job properly, get stuck in what is assumed and/or infer phylogeny poorly and/or don’t read the fossils correctly, you could end up with something like what you legitimately took issue with—a baboon based on a dinosaur.