The New York Times describes what happened to United Airlines stock value on September 8th, How a Series of Mistakes Hurt Shares of United
Investors wiped out $1 billion of the market value of UAL, United’s parent, within minutes of an erroneous news flash on Bloomberg screens about a United bankruptcy. Google and the Tribune Company, the owner of The South Florida Sun-Sentinel, whose Web site was the source of the article that led to the headline, soon blamed each other for causing the fiasco.
A chain of mistakes and vulnerabilities led to United Airlines six year old bankruptcy being reported as fresh news. Investors reacted, destroying value to the point that trading of UA stock had to be stopped until the situation could be cleared up.
Two features on a Tribune run website started the chain. One allowed an old article to appear in the most viewed box. The implementation apparently doesn’t prevent obscure articles from filtering to the top in off hours.
The second displayed the old article on the Sun-Sentinel site with today’s date but no original publish date.
Google crawled the article. It’s age and lack of original publish date confused the automated news search into interpreting the article as current.
Then a private analyst published it without independent verification. Then Bloomberg included that analyst’s report in their feed.
This led to the panic selling. Trading was resumed and the stock recovered much of its value but don’t mistake that many people lost alot of money.
The Times focuses on the relationship between newspapers and search engines. You can also focus on the pressure of news agencies to keep up with “breaking news” on various platforms. You can focus on a loss of discipline among editors.
I’d like to highlight how two casually implemented features on a website indirectly led to serious harm. I can’t imagine a print editor allowing an old article to appear without an original publish date. So how was it acceptable to allow online content to appear that way?
Given the relevancy algorithms Google is known to use, how much was the behavior of Tribune’s “most viewed” area intended to create exactly the behavior that backfired in this case, i.e. to create referential links back to Tribune for old content.
There’s plenty of blame to go around but how much sits with those who defined, accepted and implemented this behavior?