We’ve talked about the changing nature of the data center and the critical role that even more automation – from virtual machine movement to runbook tools to auto-remediation and more – will have in trying to manage data center operations in real-time. But it’s always a balancing act. How “smart” can automated processes really be? What really should be automated versus requiring some level of human scrutiny and decision-making?

Well here’s a story where the tradeoff for speed and efficiency caused a massive stock dump erroneously.

Apparently, many traders use automation software that trolls the Web for news stories and then, depending on what it finds, executes stock trades automatically. It was United Airline’s bad luck that an old article about its 2002 bankruptcy-court filing showed up on Google’s news service and somehow made it to the list of most popular stories. In one of a series of mistakes here, the story had no date on it – which means Google’s algorithm for assessing popularity didn’t have a way to exclude it as an “old” story – OR (because there are conflicting accounts) the South Florida Sun-Sentinel actually put “today’s” date on the page that the story appeared on. This got picked up by the Income Security Advisors newsletter and sent over to Bloomberg News as a one-line brief. Plus there’s the inevitable conspiracy theory that people manipulated the web traffic for this story to adversely affect UAL. Regardless, on Monday afternoon, the stock plunged 76% in less than a day.

But the real problem here is the growing use of automated programs to trigger stock trades without any human interaction – instead based on news headlines and earnings data. According to the Wall Street Journal, these automated programs were responsible for a very surprising 25% of NYSE trades in the last week of August.

I’m sure we’ll hear more as the lawyers are now involved trying to figure out who should get the blame.