According to CNet the Minnesota Supreme Court has ruled in favour of a defendant who is contesting a drunk driving charge.
The defence may review the source code of the breath test machine, in this case the Intoxilyzer 5000EN by CMI.
Interestingly the case of who owns the source code, the state of Minnesota, or manufacturer CMI, was determined to be the state of Minnesota. And since everything the state owns is free of copyright, trademarks and patents. (except on military hardware and nuclear technology) it falls under the ‘Freedom of Information’ act.
Now, on tweakers.net (Dutch) there was some discussion on whether or not the (source) code of the machine had anything to do with the results of the breath test. People seem to think the sensor outputs a value, which represents the blood/alcohol levels (actually, what we consume is ethanol: C2H6O). This is hardly the case.
The sensor, on further inspection of the specifications, measures with infra-red light, and depending on the absorption of the beams, the value of particles C2H6O per litre (or cubic metre) can be established. From that (floating point) number, the software then has to calculate the permillage (Dutch: promilage, French: proportion pour mille, German: promillesatz) of alcohol (or rather, ethanol) in a suspects blood.
Therefore, since the results (and possible arrest) are subject to the result of the test, it is possible (though I consider the odds below 1%) that the software could have a bug that under certain condition could render a false positive.
For that reason, over here in The Netherlands, if someone fails a breath test, he is always given a blood test at the police station. Blood tests are far more accurate, and those results are valid in a court of law.
It might be interesting to see what happens in other states (over 20 of them use this exact device) in response to this ruling.