I’ve been reading John Seely Brown and Paul Duguid’s The Social Life of Information. Ironically enough, for a book that extols sensitivity to the way people actually encounter and use information, it’s not especially good at providing information in a fashion I can use.

First, the book is vague about its terms and about its arguments. The chapter on bots, after making several quite good (and by no means obvious as of 2000) points about the limitations of software agents, moves to a discussion of cryptography. Some people claim that better crypto will solve the problems caused by rogue bots. Oh, really? And who, exactly, says that? Even for 2000, that’s an odd claim. Seely and Duguid do a reasonable job explaining why crypto is not a panacea, but it seems like a non sequitur to move from the one to the other. Crypto and bots are different problem domains.

The critical transition sentence is “Faced with the rebellion of these fallen agents, people often suggest that another cycle of Moore’s Law will produce suitably powerful encryption to resist them.” There’s a footnote call at the end of the sentence, but the footnote turns out to read, in its entirety, “See Chapter 1 for our discussion of Moore’s Law solutions.” That’s not a citation. That’s not evidence. It’s not even an excuse for the transition. The book is chock-full of odd shifts from one argument to another like this, and the footnotes do little to make sense of them.

Remarkably enough, they also fail to footnote some matters that ought to have easy and obvious citations. They cite the “famous case against American Airlines’ SAABRE system” as an example of bot bias. Now, that case is famous enough that I’d heard of it (although I thought it was spelled SABRE and most authorities seem to agree with me). But I doubt that most circa 2000 readers of this book had heard of a mid-1980s computer mini-scandal. Me, I wouldn’t have minded a pointer to some more useful source. Do they footnote it? No. Many such anecdotes go unsourced.

Finally, there’s a more fundamental social problem with their footnotes. Or rather, I should say, with their endnotes. When you flip to the notes section of the book, you find a bunch of calls of the form “Smith, 1992” and “Jones, 1997.” Which Smith? What piece by Jones? To find out, you need to flip a second time, to the bibliography section. The result of this citation “system” is that I’m reading the book with three bookmarks: one in the text, one in the notes, and one to the start of the bibliography.

Every major citation system I know does one of two things. Sometimes, the main text contains references directly into a (usually alphabetical) list of works cited. This approach has the virtue of pulling all the works cited into one easily skimmed list. Sometimes, the main text contains references to notes (foot- or end-) that contain the bibliographic information. This approach has the virtue of making it comparatively easy to work backwards from a note to the accompanying text. Both approaches get by with only one set of non-textual material. Using either, you can go from a passage to the bibliographic information of the source that supports it in one step.

By comparison, under the Seely-Duguid “system,” it takes two steps: one hop to the notes and then another to the bibliography. When I read nonfiction, I like to be able to set one work in the context of others, to be able to chase tangents and lines of argument out through the universe of recorded human knowledge. This two-step shuffle inhibits the movement; it partially seals The Social Life of Information off from other books.

Ironic, no?