skyhooks of the amazons

by Jacob Lee

One of the interesting things about the modern human environment is the extent to which autonomous processes and artificial intelligent agents of various kinds (and intelligences) not only figure in determining the situations in which we navigate, but figure in determining the situations in which *they* (the artificial intelligent agents ) navigate as well. For example, many retailers use automated pricing bots on sites like Amazon. Frequently these bots base pricing judgments  upon the prices of similar items being sold by their competitors. As might be expected, this can lead to various interactions between bots as they adjust to changes in other retailers prices. Sometimes the result can be amusing, even fascinating, as blogger Machael Eisen relates in his investigation of two absurdly priced books at Amazon daily ratcheting up in price:

What’s fascinating about all this is both the seemingly endless possibilities for both chaos and mischief. It seems impossible that we stumbled onto the only example of this kind of upward pricing spiral – all it took were two sellers adjusting their prices in response to each other by factors whose products were greater than 1. And while it might have been more difficult to deconstruct, one can easily see how even more bizarre things could happen when more than two sellers are in the game.



3 Comments to “skyhooks of the amazons”

  1. This is fascinating. It reminds me of an article I just read for the sociology of culture seminar, “Constructing a Market, Performing Theory” (pdf) by MacKenzie and Millo. M&M are interested in testing Callon’s argument that sociology “is wrong to try to enrich economics’s calculative, self-interested agents. Such agents do exist, he suggests; sociology’s goal should be to understand how they are produced” as economic performances. Their case is the Chicago Board Options Exchange and the development of the options market. Famously this market was described in a Nobel-winning equation by Black, Scholes, and Merton (1973), which became “the central paradigm… of financial economics.” M&M show that the equation actually described empirical option trading rather poorly, until its public acceptance fed back on those markets which reshaped themselves to fit it. This is just one of the delightful ideological ironies in the piece. Point being, though, the construction of ‘natural’ realities through the unintended feedback dynamics of interventions understood more modestly as descriptions or management systems.

  2. Yes, that is an interesting point Carl. Economics might have it that it is at once a prescriptive and a descriptive science; it is certainly the former, and I suppose that if people take the prescriptions of economics seriously, then economics might warrant being called the latter! Perhaps more so if our economic activities are guided and/or performed by artificial proxies (decision agents, price-bots, software fast trading, etc.) designed explicitly with various economic theories in mind.

    Still, I would not want to over-emphasize the point. I actually have a fair bit of respect for the economics scholarly enterprise, when properly interpreted (it has, for example, contributed a great deal of theory to artificial intelligence (of the GODFAI variety, at least). But more than that: its pretty obvious in the case at Amazon I pointed to that software agents don’t always work out as expected. Whether that is economic theory’s fault, of just the fault of lazy programmers, or “Acts of God” aka Law of Unintended Consequences…or all three (four), it is hard to say. But I will say that the effect described in the blog post is predictable in a game-theoretic way (two agents pursuing different strategies the outcomes of which depend on the actions of the other).

  3. I am reminded again of two observations in Miller and Page’s _Complex Adaptive Systems_: (1) that conventional analytic models tend to assume either a single agent, e.g., the economist’s household, or an indefinite number of similar agents, e.g., the economist’s market and (2) that the agents in question tend to be either too stupid, relying on a small number of fixed heuristics, or too smart, relying on game theoretic calculations, to be good simulations of human behavior, most of which seems to consider only a limited number of others and to involve fluid heuristics that change over time.

    The mention of game theory makes me wonder about the relevance of The Destruction of Economic Facts” described by Hernando de Soto in an article for Businessweek. How does one do game-theoretic calculations if the “facts” on which they depend become undependable?

Leave a Reply!

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: