This is really interesting stuff, and I wonder how many people who buy the Kindle have an expectation going into the purchase that such data-gathering will occur. I suspect it would be a fairly high number, considering how used to demographics we've become these days, particularly folks who have the money and technophilia (or research needs) to shell out for such an item.
What I'm more curious about is how exactly we can speak of commodification in this situation, a curiosity prompted by the discussion above around the production of the discursive objects necessary for commodification and the continued circulation of capital. Comparing the work of the Kindle to Smythe's "audience commodity" seems to muddy the waters around the commodity. In Smythe, the phrase "audience commodity" makes sense insofar as audience attention is the commodity. It is literally our act of attending to advertisements that is sold to advertisers (indeed, if the TV in on but I'm not paying attention to it, it would seem that, at least in that particular case, my own attention is not commodified by the network). But it doesn't quite make sense to call "reading" the Kindle's commodity (in the way that web advertising very directly commodifies web-surfing). It is an exploited labor, to be sure, but one that goes entirely unsold and unpaid—the commodity would seem to be the information generated about the reading selection and practice (assuming that Amazon is selling that demographic info in addition to using it themselves). In the case of television, attention is sold to companies that sell goods and services. In the case of Kindle, information about human activity is sold to marketers working on behalf of those selling goods and services. This discussion does, I admit, remain within a fairly classical Marxism, and would probably be a lot more useful in dialogue with work on immaterial and affective labor.
Reading thus becomes a value-productive activity, but not itself a commodity. The upshot of this, for me, is that Kindle owners could conceivably organize a sort of demographic noise-bomb by "reading" books interminably, taking nonsensical or deceptive notes on their Kindle "books," and otherwise scrambling the data gathered by the device. Detourning the device in a way that could perhaps detourn the discursive production of reading as object, returning reading to the realm of practice and invention (even if mischievous invention). Of course, this requires buying the device and some books in the first place, but it is something one can't really do with one's relation to TV commercials—there's no talk-back with commercials, no opportunity to lie about whether or how you're watching them.