Ironically, the same day as my tongue-in-cheek post about Chrome and HAL, news that United Airlines’ stock was pummeled thanks to algorithms broke.
After trying to write a short version of the story, I’m realizing it’s too complex. Here are the events in the sequence they’ve been reported by the Wall Street Journal:
- Late on September 6, Google’s crawler found a new link on the South Florida Sun-Sentinel newspaper’s website in a section called “Popular Stories: Business.”
- The link is titled “UAL Files for Bankruptcy”, and the only date context for the link is September 7, 2008.
- The link points to a Chicago Tribune website article published in December 2002.
- The page was indexed and added to Google News, but it did not appear in the headlines pages.
- The research firm Income Securities Advisors Inc. saw the article (again, without any useful date) and published a report.
- Users of Bloomberg, including many on Wall Street, saw the news (still without the date) and starting selling UAL stock.
- Many of the sell orders came from automated program that trade based on news and financial report data.
- After the Bloomberg report’s release, about 10:45 AM on September 8, UAL shares plummeted until the Nasdaq halted trading of the stock.
The New York Times also has coverage as does Techmeme, natch. The blame game has begun, with the Tribune blaming Google. Google has responded with its chronology, citing the lack of a date for its crawler to use to analyze the article’s freshness.
Regardless of fault, this is kind of a big deal. It’s not the first time that a stock has spiked or tanked based on rumors or erroneous news, but it might be the first time trusted information sources have mislead the market. After all, if you can’t believe newspapers, Google, or Bloomberg, your options get limited. Keep in mind that not all these sources provide free information, just in case you were thinking that free as in beer has its problems.
The nature of trading requires quick decisions based on information. If you can’t trust information, you can’t trade effectively. Take the quick out of the equation and how many jobs out there require accurate information? Now, how many people use “machines” to get information?
This is why it’s a big deal. We rely on machines to collect information and often to make decisions (e.g. the bots that trade based on news). If we can’t trust the machines, we’re hosed.
Skynet anyone? WOPR? HAL?
Maybe it’s a function of age or profession, but I bring a hefty does of skepticism to this game. I’m not a digital native; I grew up in the 80s and took notes on (gasp) notepaper in college. Hard to believe, but we had a black and white TV well into the 80s. Ironically, we had cable before most people, go figure.
Plus, I do software for a living, sometimes for fun. I know it’s hard, and there are lots of moving parts and failure points. This leads me to lowered expectations and skepticism, but not to outright fear.
As we embrace technology (which is good, don’t get me wrong), these issues and concerns will get worse, look at automated red light cameras and the outcry there. This discussion interests me immensely because it’s essentially an argument about control, who or what has it.
Good stuff for discussion. Sound off with your opinion in the comments.