“Digilantes” find that facial recognition still sucks

When a group of digilantes formed a Google Group last week dedicated to applying facial recognition technology to photos from the London riots to identify culprits, it caused quite a stir in the media. But the group’s organizer contacted Kashmir Hill at Forbes over the weekend to say that they have abandoned the project.

They created an experimental app using tools from Face.com and tested it with 30 of their friends. Their plan had been to release a Facebook app to the public so that people in the UK could volunteer to scan riot photos to see if any of the ne’er-do-wells were friends of theirs. (Not good friends, I’d have to assume.) They also gave me access to the app to give it a try. The results were too disappointing for the digilantes to actually release it. It wasn’t identifying people it should (friends of the guinea pigs) with high degrees of confidence, and it was saying with relatively high degrees of confidence that rioters were people who they were not.

Meanwhile, Scotland Yard says it is putting facial recognition technology into use, according to the AP, using a face-recognizing tool that was being developed in preparation for hosting the Olympics in 2012.

And Conor Friedersdorf writes in The Atlantic

[P]erhaps the British surveillance state won’t make much of a difference in preventing future riots or prosecuting people in the aftermath of this one. If law enforcement there isn’t any better at punishing rioters than their analogues in other countries, that’s a strong argument for rethinking their whole system: if surveillance doesn’t prevent wanton street violence and property destruction, the notion that its benefits outweigh its costs (loss of privacy and potential abuse by authorities) is all the more dubious.