Another reason to distrust computers (Photo RADAR/Photo RED LIGHT)

The story below once again illustrates that towns adopting Photo RADAR or Red Lights just don’t understand the limitations of computing challenges int he real world (or, that they are greedy and lazy and don’t care about these limitation or the effect of their citizens). There are real and effective solutions available through long-tested traffic engineering principles (85th percentile speed limits, yellow-light timing, light coordination, etc.) that most of these communities seem to ignore in favor of a little more revenue and another way to extract dollars from motorists.

(reposted from

Review finds Baltimore speed cameras errors led to hundreds of incorrect tickets

Like most major cities, Baltimore has installed a net of speed cameras across its streets to reduce traffic accidents and provide a new source of cash. Last year, those cameras provided 904,000 tickets generating some $36 million — and several complaints that the cameras were inaccurate, including one from a man who got a ticket for speeding in his Mazda 5 even though it was clearly stopped. After a city probe, the camera operator found that it had issued tickets wrongly in 5 percent of cases, a statistic that’s only raised more questions about how off other systems might be.

The review was handled by Xerox, which bought the company that installed the 83 speed cameras in Baltimore. When those cameras capture a speeder in the act, the company has its technicians review the evidence before forwarding to Baltimore police, who issue the actual ticket. That $40 ticket carries the same legal weight as one given by an officer standing by your car; it’s a sworn statement that law enforcement saw you commit the act.

Except in Daniel Doty’s case. As the Baltimore Sun reported, Doty received a ticket for going 38 mph in a 20-mph zone even though the video captured by the camera shows his car stopped at an intersection, with its brake lights on. After his complaint and several others spurred a review by Baltimore officials, Xerox did its own investigation and told the city on Dec. 11 that at five of the 83 cameras, an “unusually high rate of occurrence of radar effects” had caused the cameras to tag speeders when they shouldn’t have been. Because the processors didn’t catch the errors, Xerox estimated 5.2 percent of the several thousand tickets launched by those five cameras were wrong; outside of those cameras, Xerox says its systems were incorrect just 0.05 percent of the time.

The company says it updated its training to handle errors from two of the five sites, but in three other sites it hasn’t been able to turn the cameras back on. Because the cameras use radar to sense speeders, they can be thrown off by large trucks and other surfaces that scramble waves, and Baltimore officials say they will need to void hundreds of tickets from those five sites.

Such stories only power opposition in the 13 states that use speed cameras over the past few years. In Ohio, a computer failure forced the dismissal of hundreds of tickets; in southern Maryland, a city has been sued for allegedly issuing tickets without a review and letting a speed camera system issue hundreds of inaccurate tickets. Camera equipment companies say their devices are calibrated daily, but as the Baltimore cases show, just because a camera says you’re speeding doesn’t mean it’s being candid.

2 replies on “Another reason to distrust computers (Photo RADAR/Photo RED LIGHT)”

  1. Rich, you advocate throwing out the baby with the bathwater. Note that because of one technician’s drug-testing misconduct in Boston, hundreds of cases (where people were sent to jail, not just fined) are now under review. Should we therefore do away with chemical testing of evidence? If there are sites where radar speed detection is difficult and error-prone, don’t use it there. 78 cameras worked as expected; the failure was in adequately vetting the detectors’ output and not removing the ones in bad sites. There may be corner-cutting by the contractor; cities and towns that install radar detectors need to hold the vendors accountable, with contractually specified error limits on the detectors.

    Speeding’s a problem; widespread speeding on freeways undercuts respect for the law (*), and even small amounts of speeding in the 20-30mph range causes an increase in risk to pedestrians in more congested areas (risk of pedestrian death in a collision rises from about 5% at 20mph to 40% at 30mph; even 1mph over makes a difference — source for this is NHTSA, UK DOT, Australian Federal Office of Road Safety).

    (*) Ever heard anyone complain about bicyclists breaking traffic laws? Where do you think they got the idea that it’s okay to pick and choose which traffic laws to obey? Note that most of them also drive cars.

  2. Hello David – glad to see we both survived the Apocalypse.

    I think you missed my point about the post. In short, there are more effective, tested, tried and true engineering solutions to most of the problems that purport to be “solved” by automated enforcement. In the worst cases, automated solutions result in less safe environments. Why? Because there’s money on the table and the goal of private companies is to get that money. It’s not about safety. It’s about the case (Econ 101, see “Capitalism”).

    Let’s start with red light cameras — the issue is how to make intersections safe. The number one solution to most problem intersections is simply to increase the length of the yellow to the recommended US Federal Manual on Uniform Traffic Control Devices. ( If you want to run a Red Light Camera (RLC) system, however, the manufacturer requires that the system be “productive” (i.e. that it makes money). In order to do so, the city must shorten the yellow light ( so more people get “caught”. As an added bonus, the city can then expect more rear-end collisions. The USDOT study on RLC’s acknowledged this fat “The bulk of the results appear to support a conclusion that red light cameras reduce right-angle crashes and could increase rear end crashes”, and concluded that (at best) the economic benefits (which they equated to safety) were “modest” (and these guys are pro-RLC).

    Regarding Speed Cameras (SC), the challenge is to set appropriate speed limits according to the well accepted 85th percentile system. I.e. let traffic engineers do their jobs. Here too, the profit motive comes into play. Instead of setting unreasonably low limits and then enforcing them for profit, we should set reasonable limits are let our Police do their jobs.

    Here’s my beef with RLC and SC’s:
    A. They are not 100% accurate and if they fail, it’s up to you to prove they have failed. This should be the other way around — any challenge to a SC or RLC should immediately, and automatically, dismissed.
    B. The use of these devices is often “sold” as a “safety measure”, even though studies show the safety effects are marginal, but operated as a for-profit institution. Towns too often use their power to set a fund-raising agenda instead of a safety agenda (See the cautionary tale of New Rome, Ohio,_Ohio) – while this was not an SC issue, it’s illustrative of the legislative attitude that Motorists can safely be treated as “Cash Cows”.)

    David, I’m certainly not arguing that we should not encourage safety and conversely that those who flaunt safety should not be punished, but let’s have a basic agreement on what “safety rules” make sense. This conversation becomes impossible when the economic factor of RLC/SC company profit and/or town greed comes into play in these situations. If we start with the basics — get the appropriate engineering methodologies in place and guard them from legislative tampering — then if there’s still a problem (which I doubt there will be), let’s turn the RLC/SC infrastructure over to a non-profit and have them run it.

    Regarding (*) Ever heard anyone complain about bicyclists breaking traffic laws?

    I’m really unclear of your point here and uneasy that we should start our traditional “It’s all the cyclists fault/It’s all the motorists fault” argument. Jeeze, we’re not even into 2013 yet 🙂

Comments are closed.