textual metanoia

(Also see the review posted on Amazon.com)

Of all the canonical books on enterprise information management, such as Rockley’s or Rosenfeld and Morville’s, Stewart’s recent book speaks with less authority but gets the message across in a more precise, terse, and actionable way. For good or not, Stewart does not get mired in conceptual quibbles and does not break any new ground. Unlike when Information Architecture and the World Wide Web was first introduced, Building Enterprise Taxonomies has not launched a revolution of practice and thought. Yet, Building Enterprise Taxonomies isn’t really an in-the-trenches technical book either, it describes the technologies but never bothers to give you exact instructions on how to implement the strategy it sets out. So, being neither a ground-breaking conceptual book nor a technical how-to, what does Building Enterprise Taxonomies offer? A lot, actually.

More than anything, Building Enterprise Taxonomies is a first-rate resource for decision making—either helping to formulate your own strategy or providing framing arguments for convincing reluctant executives and stakeholders. Stewart does not take it for granted that people in purchasing roles will understand the value of an enterprise taxonomy; he spends considerable time offering ways to frame the decision, in addition to providing current research data (both economic and process-oriented). Building Enterprise Taxonomies would make an excellent companion to a technical resource: first you frame the situation with Stewart’s data and logic, then you roadmap the process with his strategy recommendations, familiarize yourself with the technology with his jargon-free descriptions, and then find a technical manual (or hire a developer) to make the ideas come to reality. For example, in chapter 4 Stewart recommends the development of a content model, as part of an overarching content management strategy, and makes a convincing case for its development, but he never describes the exact process. It’s surely a tough balance to strike, seeing both the forest and the trees, and in general Stewart gets the balance right, at least right enough for a reasonably technically-savvy information architect to benefit.

Although Stewart eschews technical jargon, he does clearly favour a particular sort of rhetoric. Perhaps in an effort to reinvigorate the field with its roots, Stewart tends to describe the conceptual issues in terms familiar to library and information science studies. As a librarian, I think he is both historically and conceptually correct, but some translation may have to occur when presenting this information to some computer science types. For example, in chapter 3 Stewart describes the variants and components of a taxonomy and refers to the parts in what are clearly their library names: synonym rings, authority files, etc. Further, Stewart describes facets not by their logical and digital parts, but rather as a librarian would. This kind of straight-forward language is useful in an enterprise setting, however, since it highlights aspects that are of central concern to corporate issues, such as the instability of information, the pace of change, the requirement for authority, the need for speed, and so on. Stewart does bring newer (but now “traditional”) information architecture concepts and strategies into a library melee. For example, Stewart’s description of card sorting (chapter 6) is very useful, and while news to a librarian, is old hat to an information architect.

The more “technical” sections of Building Enterprise Taxonomies are precise in terms of naming and recommending technologies for strategy, but broad in terms of instruction for implementation. Because Stewart describes technologies by their precise and historically-contingent names, in 5 years Building Enterprise Taxonomies is going to look as dated as those old books with the diskette in the back cover that you find for 25 cents. The issues are sure to still be relevant, but any future peoples will have to update the names and conventions. Stewart gives ANSI/ISO names and numbers where appropriate (chapter 5, factoring and term construction), mentions software packages by name (chapter 6), and describes not merely core technologies such as XML but specific implementations of the logical models (chapter 9, RDF/OWL).

Building Enterprise Taxonomies is a fairly short book, and covers a lot of material, which means Stewart often has to gloss over important details. Stewart does a great job of conceptually linking disparate ideas (e.g., his typology of metadata in chapter 2), but it frequently seems too quick and too simple. Building Enterprise Taxonomies certainly does not read like an academic book, it’s cavalier about facts and recommendations, and while it presents a persuasive thesis, it certainly doesn’t tackle the critics or alternatives (especially in chapter 1).

The physical book is well constructed and utilitarian in design. The typography is simple and borderline inelegant, but gets the job done. There are very few editorial errors or typos, although a few references to figures and images are mixed-up or missing. References are collected in chapter-specific sections at the back of the book, and the index is decent if not sometimes a little long in the tooth for broad categories. Any slight issues with the index are not a problem if you actually familiarize yourself with the content of the book, since the chapters are clear and well organized, which allow you to easily flip to the appropriate section (after all, Building Enterprise Taxonomies is hardly a canonical reference book, it reads more like a story than a technical manual). Stewart includes many useful diagrams and examples (chapter 8 on RDF is one of the best pictorial descriptions I have ever seen).

Overall Building Enterprise Taxonomies is good book. There are some slight flaws, but it’s more a matter of balance than egregious errors. I disagree with some of the balance at times, but another person may find Stewart’s balance spot on. Stewart’s advice is straight-forward and far from revolutionary, but you’ll have a difficult time finding a better primer, and it is a useful reference for ready-to-go strategy advice. Stewart offers tried-and-true advice that is neither full of hype nor outmoded. Stewart clearly believes in certain technologies (such as XML, which is a leitmotif throughout), but I think his thesis is a sound one.

(The copy of the book for review was supplied by the author to me free of charge.)