Bad breathalyzer code means countless DUI convictions are unsafe

Originally published at:


At some point your cost in lawyers will exceed the cost of good coders. This is not only an example of a badly made device but also of very poor risk management by the manufacturer.


Here in Denmark, for what it’s worth, you can’t be convicted on the basis of a breathalyzer test.
If you’re stopped and a breathalyzer test is positive, they need to take you to the police station for a blood sample. I don’t know how the laws are regarding that in other places, but it makes sense. Breathalyzers can also be tricked, I’m told, by drinking plenty of water before driving.


I believe that’s also the case in the UK.


Same in the States. The portable breathalyzer can be used to establish probable cause so they can bring you to the station, but can’t be used as evidence for a DUI conviction because they are known to be unreliable. The fancy calibrated model at the station has to be used for evidence purposes.

At least that was the rule a few years ago when I last researched this after the last panic about unreliable portable breathalyzers.


It is sad that my least cynical take is: You need to pay good coders to make the product (i.e. before you have income). They lawyers you can wait until you have a product, and in fact one popular enough that it is in use enough for someone to find out how crappy it is.

My cynical take is: the bad code pops up “DUI!” more frequently, so of corse cops want that product! The bad code here isn’t actually bad, it is juicing sales! Everything is great! (as long as you don’t care about innocent people arrested for DUI, and guilty people who were convicted based on evidence from this machine that can get a rightful conviction expunged and then go out and do it again…)


This was my first thought as well. The company probably didn’t maliciously code for higher DUIs, but it was the “fittest” product in the environment, where selection pressure is for machines that pop higher DUI rates.

As a corollary, this would mean that any software/firmware fixes later on would be undesirable, cementing in the bad code as a permanent feature. While, as I said, I don’t think they initially planned to have bad code, they did probably eventually come to the conscious awareness that the device had a high false positive rate and actively chose to do nothing about it, because it could drive down sales if they fixed it. And that’s probably where criminal investigations and charges should start coming into play.


So, very much like the insurance company business model in the USA; the less effective a company is at doing the nominal job (i.e. paying valid claims) the more profitable and successful it is. Over time the least efficient and effective insurance companies have bought out those with better customer service.


As if you need additional proof that the DUI system is a for profit scam.

1 Like

How reliable is the blood test? What equipment and standards does it use and how competent and reliable are the humans involved?

I’m not suggesting blood tests should be treated with the same skepticism as hokey breathalyzers but when I read your comment I realized that we do seem to automatically trust lab results for some reason (maybe simply because we need something). But since my home state of Massachusetts USA is in the process of overturning tens of thousands of drug convictions because of one crooked lab worker (look up Annie Dookhan if you’re curious) I can’t help wondering about whether we can trust blood test results.


I heard a presentation from an attorney who specializes in DUI convictions, and he recommended you respectfully decline all field sobriety tests. As this article points out, they can be subjective and unreliable and only stand to bolster the case for the police. Of course, you will then be taken in for a blood alcohol test, but at least it will be an accurate measure of your condition. The best solution is don’t drink and drive.


Wouldn’t that vary on a state by state basis though.

1 Like

most medical tech stuff is so poorly engineered it’s obscene. And everything costs $10k for just about anything. Interfaces are hideously bad.

1 Like

I think that refusal varies state to state. Originally I thought refusing the test in Illinois was an automatic forfeiture of license but it turns out it’s not.

For a stop

If a driver refuses to take a breath test or if an officer believes that a blood test may disclose
the presence of drugs, the driver may be held financially liable up to $500 for the costs of
the blood tests if found guilty of DUI.

For being involved in an accident

If a driver refuses to submit to chemical testing after being involved in a crash where serious
personal injury or death was involved, driving privileges will be revoked for a minimum of
one year.

Good. Breathalyzers have an even more serious problem–they are inherently unreliable as there isn’t a consistent relationship between blood alcohol and breath alcohol.

Only the blood test is accurate enough for courtroom use but many legislators don’t care.

1 Like

The analysis is done by gas chromatography, analyzing the alcohol concentration in the air above the sample, keeping the sample in a sealed container.

The machine is calibrated with a number of samples with a known alcohol concentration. The analysis must be done on two different machines to ensure reproducibility. To correct for the uncertainty and for analytical variations, 5% of the measured alcohol blood level is deducted, for lower levels always at least 0.1 g/kg. In Copenhagen at least, these measurements are done at the Institute for Forensic Medicine at the University of Copenhagen.

Provided the people working there are legit, I should say the method is probably pretty reliable. I don’t know how they measure concentrations of other drugs.

Fun fact: Until recently, we had a completely no tolerance law regarding cannabis and driving. If any traces of cannabis was found in your blood at all, you’d immediately lose your license and get a huge fine. This led to a number of absurd situations where people were busted weeks after smoking, or (as they claimed) after being in the company of others who’d been smoking. People on weight loss programs were warned that traces of THC in their fat storage might be released and they might lose their driver’s license.

Many politicians were stubborn on hearing of those cases, saying: “But nobody should be smoking cannabis anyway!”, but in the end common sense prevailed - it was decided that traffic regulations should only cover blood levels of any drug where they’re liable to put people at risk, not trivially low levels. But the old rules were in force for several years before it came to that.


In Massachusetts, according to pages 57-58 of the Massachusetts Driver’s Manual:

According to the chart on page 59 if you’re over 21, have no prior OUI offenses, and refuse a chemical test it’s 180 days suspension.

While I suppose you could refuse the breathalyzer but ask to be taken to the police station to have a blood test taken, I don’t know if the officer could insist that you use the breathalyzer or be treated as refusing the test. [One reason to make that offer would be if you think you’re just over the legal limit and the travel time to get to the station would allow you to metabolize enough of the alcohol to drop just below the limit.]

Of course the problem with this approach is that once you have been exposed for creating an unreliable product, you have eliminated yourself from the market through the ill will and mistrust you’ve created which pretty much flushes not only your investment in the product down the drain but all that money spent on lawyers becomes wasted effort as well.

1 Like

Precisely, yes. The (incredibly difficult) trick is designing a system to select for “traits” that are desirable to the general public. “Raw” capitalism only selects for profitability, which is only rarely-at-best directly aligned with the average human’s interests.


You are supposed to calibrate it before each use, using a test gas with a known alcohol level. I am not sure that police do that. At my job, it is a requirement to calibrate and log the test before each actual use, and monthly if it is not needed.