By Richard Scarlett
The term predictive search got a whole new meaning this week, with news that our online activities could potentially be used to alert the powers that be about everything from flu pandemics, surges in unemployment, stock market crashes and serious criminal activity – all before they've even happened.
No, it's not a re-release of that Tom Cruise movie Minority Report, nor is it an excerpt from the plot of a new JJ Abrams TV series, nor have we introduced October’s answer to April Fools' Day. Instead, a US computer scientist has apparently found a virtually infallible correlation between the Dow Jones and Twitter, with the suggestion that the general mood on the social networking site could potentially predict the stock market three days in advance.
If this wasn’t weird enough, it's also been claimed that search engine queries can be used to predict flu epidemics as accurately as government organisations, and can offer an indication of unemployment trends even sooner than those who are paid to analyse these things. It made me think back to a simpler time, when online behaviour and tweets were used simply to assess the popularity of Beyonce or Justin Bieber during a particular week. Now it seems that the questions have taken a turn for the serious and become 'are you likely to commit murder' or 'are you patient zero in a global bird flu crisis'?
Bear in mind that this all comes in the same week that the Defence Advanced Research Projects Agency (DARPA) announced the plausibility of plans to use algorithms, online behaviour and computers to enable the US government to predict when someone is about to commit a serious crime, and you'll see why I'm finding this more than a bit peculiar.
Don't get me wrong, if someone I knew was plotting a violent crime, or a devastating global flu pandemic was imminent, a little heads up would be nice – but at the risk of sounding cynical, I have my doubts about the prophetic abilities of Twitter and Google. That said, you only have to check out the Facebook ads to see how far user targeting technology has come (though it's probably best not to think about how much personal information must've been collected for advertisers to know that I might be interested in an annual Men's Health subscription or the latest flight prices to New York). It's no secret that the internet is getting creepy, and that invisible, easily-forgotten trail we leave behind with each website visited is becoming increasingly (and uncomfortably) apparent.
However, the (main) problem with relying too much on this data to predict something as serious as a crime is that internet users are fickle (I'm allowed to say that, because I am one). If I were caught short some day and searched for information on picking a lock or getting into a car without a key, would that place me on some strange burglar-in-waiting watch list somewhere? If I were having an off day and conducted a rigorous search for fail-safe remedies for flu symptoms, would I be on the way to sparking a global panic?
Reading all of this, it seems to me that we're just three shaven-headed precogs away from being caught up in a truly bizarre science fiction movie. Which is pretty sad, and very worrying, when you really think about it.
Comments