A few months ago I got a speeding ticket from a robot camera and was ambivalent. On one hand, I knew I was speeding and was glad not to get stopped and have a notch taken off my license. On the other hand, it sucks to get ticketed by a camera.

Red light cameras are probably more common than automatic speed traps though. theNewspaper.com has a great collection of articles examining how longer yellow light cycle times reduce accidents and violations.

For example, in Virginia in 1999 traffic engineers determined and set a safe yellow light time for an intersection on Rt. 50 at 5.5 seconds. A year later a contract for red light cameras was signed with Lockheed Martin and suddenly the county reduced the cycle time to 4 seconds. In the year it took to install the cameras there was a dramatic increase in crashes. When the cameras were eventually installed they generated steady revenue and Lockheed Martin was able to compare accident rates after installation against the artificially inflated rates of the previous year.

When Virginia increased the cycle time back to 5.5 seconds, however, average monthly violations dropped from 250 to 20. The intersections were safer, but Fairfax County wouldn’t collect nearly as much revenue.

Recently Norcross, Georgia canceled a contract with the red light company LaserCraft because revenue from tickets couldn’t support the cost of the cameras used to issue the tickets. The Georgia General Assembly had issued a bill requiring intersections with cameras to increase their yellow light times by one second. Two cameras in Norcross went from issuing fifteen tickets a day to three. Instead of generating $260k in annual citations, the city would be sinking $145k into costs associated with running the cameras.

The financial incentives for municipalities to game the system must be tremendous. At what point does the increase in safety override the benefit to the bottom line? For yellow light cameras, it’s an imperceptible fraction of a second.