Surveillance of the War Games of Finance
Some of America’s greatest economists spent World War II devising formulas for optimal bombing. Milton Friedman, for instance, had to determine whether an anti-aircraft shell should burst into 600 small pieces or 20 big pieces in order to best accomplish a mission. Many translated their work into finance’s portfolio selection theory, which was “all about balancing risk and return.”* As Friedman said, “The logical character of the problem was the same. . . . How much power do you want to sacrifice in order to have a greater probability of hitting? [Finance theory involves] exactly the same thing: How much return do you want to sacrifice in order to increase the probability that you will get what you planned for?”
Today’s finance theorists probably have not spent much time on the battlefield. But they can still have fun with ballistics trajectories, in touchscreen video games like Angry Birds. To play, you use a virtual slingshot to launch squawking birds at pigs holed up in encampments made of glass, wood, and stone. The virtual materials in the game don’t act much like real structures; that’s not the point (who really cares whether a real vaulted bluebird would displace a girder)? Rather, you gradually learn from the game itself the strategies that cause optimal destruction, blissfully unmoored from the messiness of actual materials science.
From Wars to Games to High Finance
Stock trading now appears to be similarly deracinated, concerned less with actual fundamentals than with windows of opportunity for sudden arbitrage. In “Algorithms Take Control of Wall Street,” the indispensable econoblogger Felix Salmon (and Jon Stokes) extend a line of recent articles on high frequency trading. (I collect some earlier contributions here; this piece on news-reading technology also gives the flavor of the innovations they’re describing.) They define prop trading, algorithmic trading, and predatory trading, and tell the story of a former head of American Century Ventures who built a “neural network” to optimize his picks. They also discuss the unanticipated consequences of runaway algorithmic interactions.
Salmon and Stokes develop the martial metaphor mentioned above, quoting the head of Advanced Execution Services at Credit Suisse comparing the work of algorithmic traders to submarine navigators avoiding mines. Graham Bowley also reports on the dynamic in the language of war:
Math-loving traders are using powerful computers to speed-read news reports, editorials, company Web sites, blog posts and even Twitter messages — and then letting the machines decide what it all means for the markets. . . . In some cases, the computers are actually parsing writers’ words, sentence structure, even the odd emoticon. A wink and a smile — 😉 — for instance, just might mean things are looking up for the markets. Then, often without human intervention, the programs are interpreting that news and trading on it. . . .
In a business where information is the most valuable commodity, traders with the smartest, fastest computers can outfox and outmaneuver rivals. “It is an arms race,” said Roger Ehrenberg, managing partner at IA Ventures, an investment firm specializing in young companies, speaking of some of the new technologies that help traders identify events first and interpret them.
Edward Tenner, author of “Why Things Bite Back,” throws some cold water on the boys-with-toys mindset driving these developments:
Economists and psychologists have for over a decade been analyzing information cascades, in which people’s observations of each other’s judgments may accelerate trends for worse as well as better. These systems might turn cascades into torrents. Think, too, of the ethical quandary of journalists working for the financial news agencies offering the services, knowing that any turn of phrase may nudge somebody’s machine into a decision. And to make things even livelier, speculators will be able to program banks of computers to generate and broadcast verbiage that will feed the analysis machines and move markets, as some operators have already taken advantage of the quirks of search engine algorithms.
Salmon and Stokes are also worried. They report that, at its worst, the battle of rival computerized trading strategies “is an inscrutable and uncontrollable feedback loop. Individually, these algorithms may be easy to control, but when they interact they can create unexpected behaviors—a conversation that can overwhelm the system it was built to navigate.” The flash crash of May 6 foreshadows future, more disruptive events along those lines.
Whose Conversation? Which Rationality?
But Salmon and Stokes also believe that, “At its best, this system represents an efficient and intelligent capital allocation machine, a market ruled by precision and mathematics rather than emotion and fallible judgment.” I resist the idea that we can declare the algorithms’ patter(n) of interactions “efficient and intelligent capital allocation” without much more evidence about the results of the investments it favors. After Justin Fox, Yves Smith, and John Quiggin have lain waste to various forms of the “efficient markets” hypothesis, it’s hard to see how commentators can use the term “efficient” without explaining the scope and duration of the alleged efficiency. I can certainly imagine moves that are quite helpful to certain traders during certain brief periods of time. It is harder to conjure an explanation of how the interactions Salmon and Stokes describe can generate more robust efficiencies than that.
I also take issue with their curious use of the term “conversation” to describe the interaction of computerized trading programs. There is a fundamental difference between communicative actions like conversations and strategic competition. Since conversation and other basic human interaction involves “performances where an explicit definition of the problem seems beyond our capacity . . . [and] deploys skilled performances which are themselves not explicitly thematized,” the strategic modes of artificial-intelligence “thinking” can never properly mirror the communicative nature of human interaction.
Humans are receptive to the world, altering their responses to it, and their rules for altering responses, as a result of encounters with others. The algorithms are not “conversing;” rather, they are analogous to programmed weapons sent out to strategically outwit one another. Samir Chopra and Laurence F. White’s work A Legal Theory for Autonomous Artificial Agents may eventually inform our regulatory determinations about the degree of responsibility programmers should have for the negative consequences of their creations, but we should be under no illusions that anything resembling a “conversation” is taking place here.
To return to the Angry Birds analogy: I believe the most important thing to realize here is how unmoored contemporary finance markets are from meeting real human needs. Part of that is an inevitable result of wealth inequality; a sliver of the population owns most stocks. But the technology of finance is also playing a role. As McKenzie Wark puts it in Gamer Theory,
The game is what grinds. It shapes its gamers, not in its own image, but according to its algorithms. The passage from topography to topology is the passage from storyline to gamespace, from analog control of the digital to digital control of the analog, from the diachronic sequence of events to the synchronic inter-communications of space, from voice to code. . . .[emphasis added]
The final question for a gamer theory might be to move beyond the phenomena of gaming as experienced by the gamer to conceive of gaming from the point of view of the game. . . . Surely we resemble a Beckettian assemblage of abstracted functions more than we do a holistic organism connected to a great chain of being. As games players, we are merely a set of directional impulses (up, down, left, right); as mobile phone users, we take instructions from recorded, far distant voices; as users of SMS or IM, we exchange a minimalized language often communicating little beyond the fact of communication itself (txts for nothing?).” Gamespace is an end in itself.
Computerized finance is becoming an end in itself, as well. We’re the tools of our tools. And in what is perhaps the most depressing aspect of the Angry Birds analogy, it’s hard to imagine Wall Street suffering any more from societally catastrophic capital allocations than the casual gamer who drops a bird instead of flinging it. Most of the traders are already rich; the bonuses have already been booked. It’s a pure M-M’ play, to use a theoretical framework this excess is, unfortunately, making ever more relevant.
A New Focus for the Surveillance State
So what do we do next? A recent iBrief by Michael J. McGowan surveys some of the options:
Current efforts to regulate flash orders do seem to be a step in the right direction. . . . Possible solutions may [also] lie in introducing rules that eliminate the effects of pinging, or introducing certain taxes on share transactions or rules that curb the more harmful types of algo-trading across the board.
As I noted in another context, monitoring also appears to be key to any good regulation of rapidly changing, tech-driven industries. As Daniel Altman explains, the key to preventing further disruptive events is to assure that regulators have some reliable map of all trading activity:
All the talk of regulation misses a key point: If we don’t know which institutions are doing what–if we don’t actually monitor what we’ve regulated–then that regulation won’t work. . . . The new tools that researchers now envision are meant to foresee crises in financial systems that have become impossibly complicated. “You want to see the build-up to a crash,” Markus Brunnermeier, a professor of economics at Princeton, told me in a recent interview. . . .
In February, Otmar Issing and Jan Krahnen, members of a commission advising the German government, wrote in the Financial Times that a global network map was “a vital element” for preventing future crises. And the main consultative document prepared for the European Union also recommended a map of global risks. But, without cooperation from the United States, any supposedly global map will be woefully incomplete.
I hope that the Office of Financial Research is taking that recommendation seriously. Tech luminaries like Jaron Lanier have proposed methods of representing a “wide range of innovative, nonstandard transactions” in order to give central banks and “other authorities” a “full comprehension” of the risks involved. As the OFR sets standards for Legal Entity Identification for Financial Contracts, and the SEC works on its Consolidated Audit Trail, they must develop methods for real-time monitoring of troubling developments caused by computerized high frequency trading. Though there are serious First Amendment issues raised by some government surveillance programs, these programs would not raise such concerns.
*I rely here on Justin Fox’s excellent book, The Myth of the Rational Market, pp. 47-48. Philip Mirowski has a much more extensive account of how “many of the major preoccupations and much of the theoretical machinery that now dominates economics can be traced to military research carried out in cold war think tanks, especially the RAND organization,” as Kieran Healy explains.