Drunk Until Proven Sober

In the future, apparently, every car will be equipped with a device that passively tests the driver's blood alcohol concentration, allowing the vehicle to start only if his BAC is below the government-specified level. The first step, according to The New York Times, is to require less cool ignition devices, the kind with a tube the driver has to blow into, in the cars of anyone arrested for driving under the influence, including first-time offenders and those only slightly over the 0.08 percent line. The policy has been strikingly effective in New Mexico, the Times suggests, but it immediately undermines the claim of success (emphasis added):

With that tactic and others, the state saw an 11.3 percent drop in alcohol-related fatalities last year. New Mexico was not the only state to record a decline in alcohol-related motoring deaths, and several states showed even bigger drops. For example, from 2004 to 2005, Maryland showed a decrease to 235 from 286, or 17.8 percent. In New Mexico, which has had a chronic problem with drunken driving, state officials cited the new rule on interlocks as a significant factor in their campaign to cut the fatality rate. The rule did not take effect until June 17, 2005.

Undeterred by the lack of evidence to support this expansion in the use of BAC-keyed ignition locks, Mothers Against Drunk Driving looks forward to the day when everyone has to prove his sobriety before starting his car. MADD Executive Director Chuck Hurley suggests insurers will begin offering discounts to drivers whose cars are equipped with the devices. I've got no problem with that in principle, except that the cutoff is established by legislators in response to political pressure from groups like MADD. The argument behind MADD's push to lower the DUI threshold from 0.1 percent to 0.08 percent—drivers with BACs between those levels were getting into accidents—leads inexorably to a zero tolerance policy that forbids driving with any amount of alcohol in your bloodstream. Drivers with BACs between 0 and 0.08 percent, after all, account for a significant number of "alcohol-related" accidents.

Speaking of which, the Times suggests that progress in reducing alcohol-related traffic deaths has stalled during the last decade because the total number has remained more or less steady at around 13,000 a year. But it also notes (in a clause that for some reason appears only in the print version of the article) that "the rates of deaths per car and per mile traveled have declined," which sounds like progress to me. A more fundamental problem with the numbers is that the definition of an "alcohol-related" accident does not require any evidence that drinking actually contributed to the crash—just a BAC above zero in one of the drivers. By the same logic, we could conclude that sobriety is responsible for more accidents than drinking is.