The lesson here may be: Fight the machine — or at least don’t trust it without question.
At 1:07 p.m. on Tuesday, April 23, 2013, a tweet went out over the Associated Press’s Twitter feed:
@AP: Breaking: Two Explosions in the White House and Barack Obama is injured.
The news quickly ricocheted across the Web, garnering more than 4,000 retweets. The Dow Jones Industrial Average plummeted 143.5 points in response, and the Standard & Poor’s index saw a loss of nearly $135.5 billion.
It turned out, of course, that the tweet wasn’t true.
The AP’s Twitter feed had been hacked, but for a few short minutes, the Internet — and the markets — were in free fall.
The event, which came to be known as the “hack crash,” was recently analyzed by Tero Karppi, an assistant professor of Media Theory at State University of New York at Buffalo, and Kate Crawford, who works with the Microsoft Research group and MIT Center for Civic Media. Together, they published a paper in the journal Theory Culture and Society looking at how the interplay between Twitter and the financial industry can lead to huge market shifts.
“Twitter is analyzed by algorithms, and financial markets are analyzed by algorithms,” Karppi said in an interview. “When these two fields connect over false information or a malicious tweet it can cause surprising consequences.”
In the paper, Karppi and Crawford argue that not enough attention has been paid to the role that Twitter plays in the stock market, particularly when you account for the fact that a lot of high-speed trading tools are actually automated “speculative systems” that use media sources like Twitter to determine how to make almost instant financial plays.
“Big data techniques, by which we mean large-scale data mining, predictive analytics, and machine learning, are being deployed in attempts to understand everything from human behavior to stock market tendencies,” they write, and note that this dependency on algorithms has led to a growing sense of trust in the machine or a “data fundamentalism.” The Twitter incident clearly illustrates what can happen when “massive data sets and predictive analytics are taken for objective truth.”
Karppi says that while it’s nearly impossible to know exactly what parameters are being set by the financial sector’s machines it might be time for the federal government to start thinking about regulations to help avoid future hack crashes from happening.
“You never know who’s reading your tweets,” he said, “and you never know where that information can actually be used.
The Social Media column appears every other Saturday in the Living Arts section of the Boston Globe.
Janelle Nanos can be reached at email@example.com. Follow her on Twitter @janellenanos.
Follow Janelle on Twitter