Books to Read While the Algae Grow in Your Fur, March 2021

Attention conservation notice: I have no taste, and no qualifications to opine on the sociology of radio and the music industry, or on movies.

(I didn't finish a lot of books this month, since I'm not counting re-reading bits and pieces of arcane tomes on golem-making as needed for my own shambling creation.)

K. C. Constantine, Sunshine Enemies

Mind candy from 1990: the nth in a series of mystery novels set in the fictional western Pennsylvania town of Rockford, PA, somewhere in the environs of Pittsburgh — what I've heard called the yinzerlands. It's a good mystery novel, but what really sets it apart is the dialogue. Constantine has an incredible ear for the way locals of that generation spoke, and turns it into riveting dialogue. The depiction of the life-ways of these communities also feels authentic, but that's harder for me to judge. Strongly recommended if you like well-written detective novels, or are interested in fiction set around here.

Gabriel Rossman, Climbing the Charts: What Radio Airplay Tells Us about the Diffusion of Innovation

This is a short sociological treatise about, primarily, how songs become hits on commercial American radio, or fail to do so. It's well written (not just "well written for sociology"), and has a number of very interesting points to make about topics like the diffusion of innovation, corruption, the role of genres in popular culture, and more besides. The points which most interest me are the diffusion ones.

Rossman's starting point is to look at curves of cumulative adoption over time --- how many radio stations have, by a given date, ever played such-and-such a song? His main methodological tool is to distinguish between two types of adoption curves. One is the classic elongated-S curve, looking roughly like $\frac{e^{t\lambda}}{1+e^{t\lambda}}$, which one would expect to be produced by contagion, whether mediated by a network or by some more mean-field-ish process (like a best-seller list). The other ideal type of curve is "concave", indicating a constant probability of adoption per unit time, so looking like $1-e^{-t\lambda}$. The latter he interprets as indicating some shared external forcing. Most songs which become hits follow the latter pattern (though he has illuminating things to say about the exceptional endogenous hits). The obvious question is the identity of the external force. Rossman makes a compelling case that this is, in fact, the record companies, and not (e.g.) radio station chains; on this basis he goes in to an examination of the history and theory of payola. (Basically: radio "moves product" for the record companies, so you don't want to be the only record company which is not bribing radio stations to play your music.) He also has a less compelling but still fairly persuasive analysis showing that radio stations don't really decide what to play by imitating other radio stations (at least for one "format" of radio station, during one time period). I could go on --- Rossman packs a lot into only ~200 pages --- but forbear.

The central distinction here, between curves due to external forcing and curves due to endogenous contagion, is one that's persuasive in context, but isn't necessarily either airtight or generalizable. That promotional efforts by a record company would translate into a constant hazard for adoption seems plausible enough, but one could imagine a record company whose promotional efforts start small, ramp up rapidly when one song or another takes off, and which tapers when it becomes clear that the pool of new adoptees is almost exhausted, imitating a logistic, "endogenous" diffusion curve. (It doesn't seem like good business strategy, and I take Rossman's word for it that that's not, in fact, how record promotion works.) My efforts to come up with a "just so" story in which contagion produces a constant hazard are less convincing even to me, but I only gave five minutes to the effort. Returning to my perpetual hobbyhorse of the difficulty of establishing social contagion, I would say that this is an example of using subject-matter knowledge (i.e., actual science) to rule out alternatives, which couldn't be done on purely statistical grounds.

Recommended if you have any interest in the diffusion of innovations, or in social contagion. (Probably good if you're interested in the sociology of music, too.) Finally finished, 8 years (!) after I started it, because of reading a more recent paper by the