Slashdot Mirror
Automated News Crawling Evaporates $1.14B
cmd writes "The Wall Street Journal reports that Google News crawled an obscure reprint of an article from 2002 when United Airlines was on the brink of bankruptcy. United Airlines has since recovered but due to a missing dateline, Google News ran the story as today's news. The story was then picked up by other news aggregators and eventually headlined as a news flash on Bloomberg. This triggered automated trading programs to dump UAL, cratering the stock from $12 to $3 and evaporating 1.14 billion dollars (nearly United's total market cap today) in shareholder wealth. The stock recovered within the day to $10 and is now trading at $9.62, a market cap of $300M less than before Google ran the story." The article makes clear that Google's news bot only noticed the old story because it has been voted up in popularity on the site of the South Florida Sun-Sentinel newspaper. The original thought was that stock manipulation may have been behind the incident, but this suspicion seems to be fading.
When you're talking about numbers like that then there is definitely a responsibility somewhere to try to prevent it happening again.
Never happen.
Every brokerage house in the country is increasing their presence in "Automated Trading". The majority of these companies income is now made on small percentage point fluctuations brokered very fast. With the automated systems pulling in that much more money than the traditional ones, and with more algorithms being designed (and held as proprietary info), what would YOU do?
Ridiculous? Yes. Not going away? Also Yes.
Might as well vote to bring back the horse drawn carriage since the price of gas has risen so much lately.
This space for rent. All reasonable inquiries will be entertained at proprietors discretion.
After reading TFA, it sounds like google is LEAST to blame out of the many many automated systems involved. First of all, the damn story should have been dated. That's the tribune's fault. Google doesn't seem to have claimed it as today's news, only ranked it high up. No one should have ever reprinted the story without actually CHECKING WITH UNITED AIR FIRST. That's neither google nor the tribune's fault. That's every service that reprinted the story as new without verifying its fault. Google and tribune seem least at fault because neither ever gave any indication it was a new story.
If an officer ever threatens to taze you, say you have a pacemaker.
No, that wouldn't be awesome at all. It's amazing how ignorant slashdot users can obviously be if you take a second to think about the consequences of what they suggest. This is, in fact, the second time today I've responded to someone calling for the spectacular failure of a major company, in the middle of a major economic downturn. Ignoring the huge impact Google falling would have on our economy, maybe you can imagine something a little more practical, that hits home: What search engine do you use?
Whale
I am by no means a lawyer but it sounds like the automated trading software has the majority of the blame since it is the one that actually intiated the trade. So what if google descided to reprint old news...nothing wrong with that.
Trying to install linux on my microwave, but keep getting a kernel panic...
Maybe if the industry you were in wasn't on the brink of an economic disaster for the past 7 years an old story being dug up by Google News wouldn't have had such a drastic impact.
I wonder who gets sued for this. Obviously Google's Crawler was at fault, somewhat
How do you figure that? It's doing what it's supposed to. The problem wasn't the crawler, it's the people who thought the crawler was some kind of magic AI that could find relevant stories for them.
It's gambling, it isn't investing.
Deleted
This smells like a case of the Frankenstein complex to me. Although Google News may have linked the article in its recent results because of the fresh link on the Sun Sentinel home page, both the WSJ article and the Forbes investigation make it very clear that the problem was a human editor who misinterpreted the original article and posted it as new information (with a freshly written headline) in a by-subscription-only investor information service that is carried on Bloomberg trading terminals.
A human saw the story, failed to check the date (there was no date line at the top of the article), refreshed it with a new headline, and republished it on a trading service that was believed to be a source of credible journalism by its readers.
...die by the sword.
If you're trading based on news, minutes or seconds of advantage, and arbitrage, then you'll occasionally get nailed by something unforseen. Like a two-or three year old news story that 'looks' new all of a sudden.
So is someone watching over this, ready to pull the plug if something doesn't smell right? Nope. When the timing is so short, you're at the mercy of the system.
Besides, most of these programs are pure arbitrage. Which is sort of the 'greater fool' theory in play. You make money at someone else's expense. Like they didn't know about something, and you sell out a minute before they find out.
In this case, justice, if there is any, is won by those who saw the activity, realized the mistake, bought low and sold back high. Too bad, big institutional investor. You lost this round.
I can't shed a tear.
deleting the extra space after periods so i can stay relevant, yeah.
Not only is it unnecessary, but such a law would be completely unenforceable. Look at how much trouble online gaming servers have in preventing people from running bots which play their characters for them.
If I can be modded down for being a troll, can I be modded up for being an orc, or a balrog?
... it will be now
If you can read this... 01110101 01110010 00100000 01100001 00100000 01100111 01100101 01100101 01101011
Exactly. People who watch the news all day going BUY SELL BUY SELL are just asking to lose money on mistakes like this.
I think it is reasonable for google news to ignore articles without dates.
http://michaelsmith.id.au
Automatic trading applications take their inputs from Beta software?
Except the article did have a date on it. There was a box on the page for today's date. I certainly hope you don't expect google's indexing system to be able to figure out what the context of the date on the page was.