Three-Toed Sloth

Books to Read While the Algae Grow in Your Fur, December 2020

Attention conservation notice: I have no taste, and no qualifications to opine on the business side of online advertising.

(Left almost-finished in 2020, because I got interrupted, and posted in 2026, because I wanted to procrastinate about half-a-dozen research projects.)

A. R. Larkwood, The Unspoken Name

Mind candy fantasy. This obviously owes a lot to Le Guin's The Tombs of Atuan (supposedly-reincarnated priestess of the chthonic gods is broken out of her temple/death-trap by questing foreign wizard), but isn't just a re-telling of a classic. Reading this led me to go back and re-read Atuan, which is unquestionably better --- to use a Le Guinian distinction, Larkwood's language and ideation belongs to Poughkeepsie rather than Elfland --- but this was enjoyable enough, for what it is, that I will probably read the inevitable sequel. §

Philippe Riche, Bad Break

Declan Shalvey, Bog Bodies

Mind candy comic-book crime fiction. §

Tim Hwang, Suprime Attention Crisis: Advertising and the Time Bomb at the Heart of the Internet

I have mixed feelings about this. On the one hand, I enjoyed, and learned from, the rather caustic presentation of how online advertising works as a business, and how ineffective such advertising is. (I'd actually be sort of interested to read a debate on this point between Hwang and, say, Shoshana Zuboff.)

On the other hand, the central conceit here just isn't very persuasive to me. It's an analogy between what such advertising does --- deliver the attention of a group of people who have such-and-such specified characteristics --- and the way subprime mortgage were financed, at least pre-2008. Back then, they got bundled up into "mortgage backed securities", where (to simplify a little *) the buyer of the security got a share of the mortgage payments of some collection of subprime loans. The idea was that while each individual loan had a non-trivial risk of default (that's what made them "subprime"), the collection would offer steady payments, because, after all, what were the odds of a lot of mortgage borrowers all going bust at once?

The only part of Hwang's analogy which I see as holding any water is that both advertising and mortgage-back securities involve some sort of aggregation. But, in finance, lots of things involve aggregation. Back in the old days (#include <itsawonderfulife.h>), when banks or savings-and-loans held on to the mortgages they made, the bank still counted on aggregation, on averaging over its whole portfolio of loans, in order to have any sort of reliable cash-flow. Buying stock in a bank which holds on to a portfolio of loans is buying a claim on the flow of income from those loan payments. Heck, opening a savings account was lending the bank money, in exchange for interest income, which comes from their portfolio of loans. (**) Similarly, with a good old fashioned newspaper or national TV channel, the people who bought ads were buying the aggregated attention of huge populations --- or at least the chance at getting their attention. So I don't think saying "aggregation, phooey!" is really a good criticism. If anything, the promise of more targeted advertising is that there is less aggregation than in the good old days.

One could reply that in both the cases of mortgage finance and advertising, the contemporary forms (1) replace aggregation over some spatial region with aggregating over more abstractly-defined classes, and (2) both are accompanied by false promises of precision. Point (1) is interesting but not decisive --- if anything, you could argue that people only settled for spatial aggregation because nothing else was feasible with weak communications technology. Point (2), however, might be a better ground for drawing an analogy, but surely a very weak and loose one.

To his credit, Hwang does try to suggest some things which might be better than our current, advertising-supported Internet. But those suggestions are, well, curious in their own right. In place of social media, for instance, he suggests a website which would look to users like a huge canvas where they could write whatever they want --- but anonymously, and with the constant prospect of being over-written by other users. How this would replace, or pre-empt, the uses which lead people to Facebook, or Twitter, or any of the others, Hwang does not say. He also does not remark on the way his proposal resembles nothing in existence so much as 4chan.

Trying to draw these threads of reaction together into something resembling a judgment, I think that Hwang does a convincing job of arguing that online advertising, as it exists, has some very bad consequences for just about everyone except stockholders in Google and Facebook, and is not nearly as effective as both its advocates and detractors like to claim. But the central analogy is just too weak to be any use, let alone to suggest that the current system is Inevitably Doomed. §

*: Among other things, I am omitting details about "tranching", i.e., prioritizing paying off those who accepted lower interest rates. ^

**: Regular readers will be aware that I have a lot of opinions on subjects where I don't really know enough to be entitled to an opinion. (New readers will not exactly be surprised by this news [though maybe by my being aware of it].) One of those opinions-I-am-not-entitled-to is that, for the reasons just given, the official stories about securitization being about risk allocation, and letting investors choose their levels of exposure to risk, do not make much sense. But securitization does make sense through a combination of regulatory arbitrage, and wanting to shed the costs associated with trying to make good loans. But this is something I've ranted about elsewhere. ^

Books to Read While the Algae Grow in Your Fur; Scientifiction and Fantastica; Pleasures of Detection, Portraits of Crime; Actually, "Dr. Internet" Is the Name of the Monsters' Creator; Linkage; The Dismal Science