Today’s Interconnected Systems Magnify Simple Human Errors at Frightening Speed And Scale

Brad Smith monitors stock prices at the Nasdaq MarketSite, Tuesday, April 25, 2017, in New York.

Brad Smith monitors stock prices at the Nasdaq MarketSite, Tuesday, April 25, 2017, in New York. Mark Lennihan/AP

A glitch in after-hours trading July 3 caused some Nasdaq-listed prices to go haywire.

A glitch in after-hours trading July 3 caused some Nasdaq-listed prices to go haywire: Some stocks jumped more than 3,000 percent, with Microsoft very briefly worth nearly $1 trillion. The episode highlights how today’s complex, interconnected digital systems can magnify seemingly small mistakes, impacting people around the world.

Take British Airways, which stranded 75,000 people last month after a contractor reportedly switched off a power supply by accident, knocking out the company’s computer systems. Or the Bank of the Philippines Islands earlier this year. A data processing error affected some 1.5 million clients, which some initially suspected was the work of cyber criminals. The reality was much more mundane: A specialist was blamed for entering the wrong file in a system.

There are also the periodic “fat finger” trades, where a trader mistakenly keys in an extra zero (or two or three) in an order that sends ripples through financial markets. A junior banker at Deutsche Bank accidentally transferred $6 billion to a hedge fund client in 2015. The year before, an erroneous transaction worth $622 billion shocked global trading desks. (These trades are generally cancelled, unwound, or otherwise reimbursed.)

A report by the Ponemon Institute in 12 countries found that negligent employees were responsible for about 25 percent of data breaches the researchers studied.

In Nasdaq’s case, the stock exchange says yesterday’s glitch stemmed from a routine daily data test that was moved up by several hours because trading closed early on July 3. Erroneous prices apparently based on test data showed up on Bloomberg terminals used by professional traders, as well as on websites like Google used by non-pros.

The root of the error can probably be found somewhere along the chain between Nasdaq and a small number of third-party vendors who distribute market data. Perhaps someone failed to heed a notice of the early data test, or didn’t receive it in the first place. So far, it doesn’t seem like anyone lost much—if any—money, so the main harm is reputational. But it is increasingly a fact of modern life that mundane, simple human errors now have the potential to spiral rapidly and cause problems for people all over the world.