Crime Prediction Software Is Here and It's a Very Bad Idea  

While I quite like IBM's "smarter planet" marketing campaign and the use of advanced analytics in applications like smart grids, this particular use of data analysis seems rather evil. From Gizmodo - Crime Prediction Software Is Here and It's a Very Bad Idea.

IBM clearly wants this to go big. They have spent a whooping $12 billion beefing up its analytics division. Again, here's the full quote from Deepak Advani:

Predictive analytics gives government organizations worldwide a highly-sophisticated and intelligent source to create safer communities by identifying, predicting, responding to and preventing criminal activities. It gives the criminal justice system the ability to draw upon the wealth of data available to detect patterns, make reliable projections and then take the appropriate action in real time to combat crime and protect citizens.

If that sounds scary to you, that's because it is. First it's the convicted-but-potentially-recidivistic criminals. Then it's the potential terrorists. Then it's everyone of us, in a big database, getting flagged because some combination of factors—travel patterns, credit card activity, relationships, messaging, social activity and everything else—indicate that we may be thinking about doing something against the law. Potentially, a crime prediction system can avoid murder, robbery, or a terrorist act.

It actually sounds like a good idea. For example, there are certain patterns that can identify psychopaths and potential killers or child abusers or wife beaters. It only makes sense to put a future system in place that can prevent identify potential criminals, then put them under surveillance.

The reality is that it's not such a good idea: While everything may seem driven by the desire to achieve better security, one single false positive would make the whole system unfair. And that's not even getting into the potential abuse of such a system. Like the last time IBM got into a vaguely similar business for a good cause, during the 1930s. They shipped a lot of cataloguing machines to certain government in Europe, to put together an advanced census. That was good. Census can improve societies by identifying needs and problems that the government can solve. At the end, however, that didn't end well for more than 11 million people.

And yes, this comparison is an extreme exaggeration. But one thing is clear: No matter how you look at it, cataloguing people—any kind of people—based on statistical predictive software, and then taking pre-empetive actions against them based on the results, is the wrong way to improve our society. Agreeing with this course of action will inevitably take us into a potentially fatal path.