Peter Morville wrote the books Ambient Findability and was/is an author or co-author for various editions of Information Architecture for the World Wide Web. He's smart.
This book is a mixed bag. At bottom it's a defense of the role of the information architect and the idea that the study of cultural information patterns in institutions and organizations is necessary and urgent in order to create solid representations, pathways, and taxonomies of information. And he's very good at helping us understand the provisional nature of taxonomies and how they have to balance a lot of forces. The role of the IA needs defense because it's being challenged by roles in marketing, product development, design, and engineering. Morville, like many IA professionals, comes out of library/information science. So he's defending a very particular version of IA.
The constant rhetorical strategy here is to note how companies don't really take the time to understand their own information patterns -- and in many cases can't understand them, owing to cultural blinders. This is a pretty simple idea. So why make it? The reason is that there are emergent practices such as "agile" development that tend to reduce the emphasis on documents and deliverables in favor of a more iterative explorative model that doesn't leave a lot of time for IA in the fashion Morville describes it (he would probably disagree with this formulation). In many places, the book emphasizes planning and thoughtfulness both of which get swept under the rug all too frequently.
The good bits: Chapter 3 is about "Connections" and talks about alternatives to the web's idea of links. This is welcome because the pre-history of the web is often forgotten. Having said that, he's a little too eager in his allowance that the web is good because it's essentially all we could afford in the early 90s -- i.e., we could never do the full Ted Nelson version of Hypertext because it was too complicated (that's a counterfactual if I ever heard one). He takes up alternative conceptions of intertwined data - paths, maps, loops, forks. An odd misstep is a quotation from Eric Raymond to the extent that "forking" "is a taboo of open source culture" (p. 104). That was true in 1999, when Raymond's book came out, but with the advent of Git and GitHub, forking is all we do in open source culture. (Raymond was almost surely talking about "big forks" such as in the evolution of Emacs -- but we have moved beyond the era of heroically large open source projects, which is something else worth taking up.)
Chapter 4 is a solid explanation of the importance of the idea of culture in IA, and convincingly presents ethnographic study as the master tool for IA. Having said that . . . you'd get this from a good undergraduate liberal arts education and Anthro 101. As he cites Clifford Geertz, James Spradley, and others, I'm thinking: Really? People don't get this in college anymore? I guess not. If those names mean nothing to you, read that chapter.
The rest is pretty basic. Chapter 2 on "Categories" reminds us that classification is about choices. To it's credit, this is the first treatment I've read about cultural categories that doesn't quote from the opening of Foucault's The Order of Things. There are some oddities here: There's a bit on Facebook's "Like" mechanism with an illustration of a "Dislike" (p. 51) -- but of course Facebook doesn't have a "Dislike" operation, which is surely the most salient fact regarding Facebook's ontological mechanisms.
Chapters 1 and 5 both talk about our embeddedness in ecosystems, but . . . I think there is a rupture in the argument here. It goes something like this. Morville tells us that cultural ecosystems (and natural ones) are very complex and we have to negotiate them on our own, and not give in to limits. That's what a great information architect does. Are you with me? So a great IA navigates through authorities and information sources.
But then we get this: He talks about how messed up medical institutions are (pp. 153-155) and tells us that "our trust in doctors is even more misplaced, since malpractice isn't as random as a butterfly flapping its wings" (p. 155). Meanwhile, medicine today manufactures its own consent (p. 156). What do do? Morville says that he himself mostly avoids that institution in order to find his own path: "That's why I've gone rogue. I haven't been to a doctor in years. I don't see my dentist either. If I have a serious problem, I'll consult a professional, but I believe checkups are dangerous" (p. 156).
So here we have a book that is staking out a kind of expertise; and yet it rejects expertise in another discipline. This is really faulty. The problem here is that just as businesses should not reject the fundamentals of IA because they're hard, just so Morville should not reject a whole institution because of its large-scale malfunctions. Really, he needs to find a new doctor, one more attuned to the arguments of his book!
Insofar as this is really a book about professional roles and "doing the right thing," it is highly problematic that he would be so unembarrassed to pretend to go out on his own for his personal health. In other words, just as he says for himself: "That's why I pay more attention to my environment, economics, and behavior" (156), I would say that he's making a profound categorical mistake to not see that these exact renovations of everyday practice are happening in other fields.