Breathalyzer tests may be blowing smoke

Nobody likes a driver who gets behind the wheel of a car when he or she is three sheets to the wind and puts innocent people in danger on the road. But even less should we like unsubstantiated accusations of drunk driving based on fallible devices -- commonly called "breathalyzers" -- that test for blood-alcohol content using technology that might as well be so many rituals and incantations because it's kept a secret from anybody who might want to challenge the results.

And that's exactly why more than 100 DUI cases have been tossed out in recent months in Tucson, Arizona -- with a whopping 19 dismissed just yesterday.

Judge Margarita Bernal threw the cases out because CMI Inc., the manufacturer of the Intoxilyzer 8000 used in Tucson, refuses to release the software source code that runs the machines and plays a large part in the reliability -- or lack thereof -- of the devices. That's a matter of concern, because CMI President Toby Hall has himself conceded that there are certain technical vulnerabilities in his company's products.

Bernal based Monday's ruling on testimony in an unrelated Pima County Superior Court case by Toby Hall, president of the company that makes the machines.

"The testimony of Mr. Hall clearly raises anomalies, errors and issues which can impact (the machine's) reliability and credibility," Bernal wrote.

In fact, problems with the Intoxilyzer 8000 date back at least two years. The Sarasota Herald Tribune reported in 2006, "The new machine's first glitch was discovered last month, and now breath-test machines across the state need a software upgrade because state officials realized the Intoxilyzer 8000 failed in certain situations. Prosecutors in Sarasota and Manatee counties have asked judges to stop accepting plea deals in DUI cases until the issue can be resolved."

The Intoxilyzer 8000 had been adopted after earlier problems with the Intoxilyzer 5000.

Turning to CMI's competitors doesn't necessarily solve the problem either. At the beginning of 2007, retired New Jersey  Judge Michael King reported to the state Supreme Court that the widely used Alcotest 7110 was unreliable, with the result, as the New York Times put it, "that thousands of convictions could be overturned."

Once again, the manufactureer, Draeger, Inc., refused to release the device's source code so that the machine's reliability could be assessed.

The manufacturerers' reticence is understandable for any company that bases its success on its technology and doesn't want that technology pirated by the competition. Keeping source code proprietary is a way of holding on to an edge. But as defense attorney Evan Levow commented after Judge King's report, "They say the source code is proprietary. Well, constitutional rights are not proprietary.”

In the absence of some knowledge of how breathalyzers work -- or whether they work -- challenging their secrecy and their reliability has become something of a cottage industry among lawyers. A simple Google search on the words "breathalyzer" and "reliability" turn up endless lawyers' Websites and blogs -- some trawling for clients, others sharing information about successful challenges among attorneys.

Those challenges are going to continue, and they'll deserve continued success, so long as drunk-driving convictions are dependent on a magic machine, the reliability of which has to be taken on faith.

Of course, not everybody got the memo about the breathalyzer's unreliability. In Massachusetts, Melanie's Law punishes drivers with automatic license suspensions for refusing to take the test -- even if they're later cleared of drunk driving.