Agencement Exposed: when the 'black box' failed

The Dow Jones Industrial Average plunged 1,000 points on Thursday.  In that time billions of dollars was lost.

Thursday evening when the news came across Twitter, I was sitting in a make-shift press room with one of my favourite fellow journalists Chris Milton, who remarked, "I'll bet somebody tripped over a cord."  In other words, somebody momentarily unplugged the electric supply to a trading machine in some important someplace.  The next day a WSJ headline about the plunge used the words "fat fingers" to describe the cause of the incident.  While the romantic in me would like to think this was some ingenius work by supertheives with hacking skills, taking pennies off millions of trades and transferring them into a bank account, the answer is probably a lot less nefarious. 

The continuing puzzlement over the cause of the plunge reminds me of a truly excellent study of financial markets by Donald MacKenzie at the University of Edinburgh.  In Material Markets: How Economic Agents are Constructed, MacKenzie uses a discussion of agencement (Callon via Deleuze) to illustrate just how participants in financial markets shape the construction of the market, which has feedback effects, such that the market construction evolves with unintended (misunderstood?) consequences.  As MacKenzie's theory as it applies to the financial crisis suggests that as pricing models were used to predict market prices, the more those predictions shaped the market, making the predicted prices less predictable-- pricing models based on historical data, the future would no longer resemble the past.  (please don't stop reading here, that was the most complicate bit, promise)

MacKenzie quotes Callon and Caliskan (2005) to explain agencement:

"Agencement denote socio-technical arrangements when they are considered from the point of view of their capacity to act and to give meaning to action."

This is also called "actor-network" theory.  In other words, in financial markets, a stock trader with a calculator (MacKenzie's example) may be more effective than a stock trader without one, but give the trader without one a calculator and they will still not have the same knowledge capacity.  Experience aside, it is the physical surroundings, relationships to tools and to other actors that shapes the full knowledge capacity of the trader.  This is agencement.

Agencement involves (say Callon and Latour as qtd in MacKenzie) a said market participant (stock trader, economic actor, what-have-you) sitting on a ‘black box’ process-- “a device the internal structure of which is opaque or can be disregarded, and which can be treated simply as transforming given inputs into predictable, appropriate outputs” (MacKenzie, Material Markets, p34). What happened Thursday involved black-box failure and sounds like it will resulting in a re-examination of the black box bit of agencement.

What I want to convey about the incident is that it is a reminder about agencement : we depend upon financial markets which depend upon the humans and machines active in them-- but we often forget that who is acting on (using?) what can be clouded by continuous, little-noticed adjustments in our relationships that only become noticeable when something fucks up.  But at this point, it’s usually too late, the Dow has plunged 1,000 points, which spurs overseas markets to plunge as well-- billions of dollars are lost on a ‘black box’ error.