Skip to main content
Jan 19, 2009

What happens when Googlebot gets it wrong

News crawler contributed to United Airlines stock collapse as an old bankruptcy story was presented as new by Google.

It sounds like the plot of a science fiction film: highly intelligent robots prowl the land, wreaking havoc and leaving a trail of destruction in their wake. IROs quiver beneath their bedsheets.

Okay, so things haven’t really got this bad. But there are computers out there that can do serious damage to a company’s investor communications. A case in point is United Airline’s stock price collapse in September 2008.

That month, Google’s news crawler, Googlebot, which trawls the web building a searchable index of documents, presented as new an old story about the airline’s 2002 bankruptcy. The next day, a journalist working for investment advisory firm Income Securities Advisors (ISA) found the story on the search engine and wrote a summary. ISA supplies stories to Bloomberg, and when Bloomberg picked up the story, investors started bailing out of United. Within five minutes, the company’s shares lost 75 percent of their value. 

It took a press release from United stating that the story was ‘completely untrue’ and a brief suspension of trading on NASDAQ to calm the situation. News sites also removed the offending story from the web. When trading resumed, United’s stock price climbed, but didn’t recover completely.

Much wrangling followed about who, or what, was to blame. It transpired the original story had been archived without a date on the website of Florida’s Sun-Sentinel. As a result, Googlebot searched for another date on the web page where the story appeared. The only one it could find was that day’s date, at the top of the page, so it stamped the old story with the new date.

To be fair to the bots, some human error was involved. If anyone had read the story properly on its journey from Google to Bloomberg, he or she would have spotted its age pretty quickly. There is, however, no disguising the fact that news bots present a danger to IROs. The rapid dissemination of stories online, coupled with the high speed of today’s automated trading machines, means erroneous news can wipe out a stock’s value in minutes.

So what protection is available for IROs against such news on the web? The SEC has an internet enforcement program that investigates cases of online market manipulation. An SEC spokesperson would not confirm or deny the presence of any ongoing investigations but it is understood the regulator is looking into recent incidences at United and Apple (see Human error, below).

Rapid response
On the corporate side, responding quickly is vital, says Michael Pranikoff, director of emerging media at PR Newswire, who spends a lot of his time educating staff at his firm and other companies about the opportunities and risks of new media. ‘The best thing to do is have a communications plan in place that gives power to one person, allowing things to move very quickly,’ he comments.

Some service providers advocate putting out a consistent stream of information to prevent a ‘news vacuum’ appearing on the web, which could be filled with negative news or rumors. This sounds like the equivalent of a jet pilot announcing over the tannoy every 10 minutes: ‘There’s absolutely nothing to worry about’ – which may not produce the desired effect.

However, Darin Wolter, vice president of sales at Marketwire, views the analogy from another angle. ‘We don’t think the pilot needs to come on every 10 minutes and say there’s nothing to worry about,’ he says. ‘But it’s nice when you get an update on speed, altitude, whether you are arriving on time or early, or if turbulence is expected and for how long. While consistent communications and broad levels of transparency are advocated, it’s important the content is material and newsworthy.’

Of course, it would be better to prevent erroneous news getting out in the first place. Companies and media outlets can help by making sure all information is dated, commented Neil Hershberg of Business Wire at the time of the United incident, which he referred to as a ‘clarion call for action’.

Even so, it’s inevitable news bots will throw up problems for companies every now and again. ‘Unfortunately, nothing is perfect when it comes to Botland,’ concludes Pranikoff.

Human error

Apple has also suffered recently from erroneous news appearing on the web, although – in this case – humans must take most of the blame.

In August 2008 a Bloomberg reporter accidentally published a draft obituary of Apple’s charismatic founder, Steve Jobs. It was quickly removed and Apple’s shares didn’t suffer, but the incident breathed new life into speculation over Jobs’ health.

There was more trouble for Apple in October: the company’s shares took a hit when a member of the public posted a false report on a CNN website for so-called citizen journalists stating that Jobs had been rushed to hospital after suffering a massive heart attack. The stock lost around 5 percent before bouncing back.