Algorithm VS Human

By Panjandrum / 28 October 2012 / Intel Analysis, Technology

There’s an interesting bit in the week-before-last’s economist (yes, things are getting a little bit like that around here) about the Canadian naval officer who was flogging intelligence to the Russians.

The whole thing is ripe for comment, but a couple of sentences really stick out:

Sophisticated algorithms meant to identify suspicious activity did not pick up on the searches Lieutenant Delisle made over a period of almost five years for information pertaining to Russia…

It was a human, not a machine, that brought the naval spy down. A border security officer, suspicious about the amount of cash and pre-paid credit cards that the naval officer was carrying after a short trip to Brazil, alerted the national police in late 2011.”

That ‘sophisticated’ algorithms let us down is nothing new. Neither is the fact that well trained and well-motivated humans are pretty good at spotting anomalies and oddities and other causes for concern.

This goes to the heart of what IMSL is all about. We’re not looking for the perfect algorithm that will detect fraud or predict where the next attack is going to happen. But we can provide a first rate analyst backed up by some of the best analytical tools in the world – a combination that is proven to work.

 

Please insert your own gag here about naval officers, golden rivets, prison showers, dropped soap, etc. etc.

Leave a Reply

Your email address will not be published.