Researchers Say a Breathalyzer Has Flaws, Casting Doubt On Countless Convictions (zdnet.com)
An anonymous reader writes: The source code behind a police breathalyzer widely used in multiple states -- and millions of drunk driving arrests -- is under fire. It's the latest case of technology and the real world colliding -- one that revolves around source code, calibration of equipment, two researchers and legal maneuvering, state law enforcement agencies, and Draeger, the breathalyzer's manufacturer. This most recent skirmish began a decade ago when Washington state police sought to replace its aging fleet of breathalyzers. When the Washington police opened solicitations, the only bidder, Draeger, a German medical technology maker, won the contract to sell its flagship device, the Alcotest 9510, across the state. But defense attorneys have long believed the breathalyzer is faulty. Jason Lantz, a Washington-based defense lawyer, enlisted a software engineer and a security researcher to examine its source code. The two experts wrote in a preliminary report that they found flaws capable of producing incorrect breath test results. The defense hailed the results as a breakthrough, believing the findings could cast doubt on countless drunk-driving prosecutions.
They take blood now, and refusal is an automatic conviction. The breath thing is just to let them know if they need to go to the trouble of calling a phlebotomist to the scene.
In .nl at least, these breathalyzers are used by police to do a quick test only. If the result is positive, the suspect is carted to a police station where a much more accurate machine is used to determine the blood alcohol levels. Only that ladt number is used as evidence.
I was assuming all countries had similar methods.
If an experiment works, something has gone wrong.
For charges so serious, it doesn't make much sense to me. It's a test that in nature seems like it is just made for a quick rough assessment that is fast and easy to do in the field. Great for weeding out who is worth the time to take to the station for a real test.
OK, I get it, the user interface etc. should be computerized for convenience.
But the part that "holds up in court" should be as close to the raw data as possible, which can and probably should be analog or at the very least a very simple, relatively-easy-to-audit-for-correctness digital system.
This web site describes a breathalyzer which appears to be analog. It also describes an intoxilyzer, which uses a microprocessor. If the electrical pulse being fed into the microprocessor can be captured for later playback in court, then the defense team's experts can interpret it using their own algorithms if they wish.
Knowledge is how to play a game, intelligence is how to win, wisdom is knowing what game to play.
A Drunk Driving conviction is big business in California.
Of course they would tweak the test to get more and more money.
You can easily spend $15K if you get one.
The neurological deficits of drinking (and using many other drugs) persist long after blood levels have dropped to zero. So either develop tests that measure actual impairment (and bust a lot more people) or just admit that so many people are impaired but sliding under the limits that we may as well just let them drive.
Have gnu, will travel.
So in most cases they use the breathalyzer along with a field sobriety test to determine if they should arrest a driver, after which they normally draw a blood sample for a more precise test. (In most states, drivers can refuse the tests and receive an automatic license suspension, though that's often not as bad as a conviction, but it doesn't preclude a conviction based on the sobriety test alone.)
So we have lots of cases where there is both breathalyzer test data and blood test data. This gives a huge amount of data that can be correlated. Also, the times of the test should be recorded, so expected declines due to delays between the two tests can be computed. So all we need to do is gather up the data from a few years of use, and then we can see what the reliability of the breathalyzer is in comparison to the blood test.
We know that any test can fail (equipment failure, cross contamination, operator error, etc.), but this will give solid data on the expected error.
check your six.
Having worked at Draeger I am somewhat biased but I will also be able to provide a more in-depth description of the company itself. The company manufactures loads of equipment for diving (tanks, breathing apparatus), fire prevention, fire detection, infra-red cameras, alarm systems and more. Everything is obviously thoroughly documented, tested and accounted for due to risk of life. The R&D floor at the office I worked housed proper chemical and electronics labs containing millions of dollars worth of inventory. I find it hard to believe that there are "serious" issues with these breathalysers, especially considering they are sold not just in the US but dozens of other countries.
The real issue is with police departments not taking proper care of these machines. Not following calibration procedures and using them in out of spec environments (eg. near an air conditioner set to full blast). I know for a fact officers using this equipment can't be bothered to read the "MOUTHPIECES ONLY" sign on the returns barrels as I usually found at least 20 USED needles, condoms, pills and other trash in every barrel.
Regardless, this is just a bunch of convicted drunk drivers that want to argue numbers to get out of jail and refuse to drive sober. Even if the device was off by 6% (as claimed in the article), zero will still be zero.
Lets be real, police officers are not fond of paperwork and they will only test you if they have good reason... When they see you swerve all over the road being at least twice over the legal limit, the test itself is just to prove objectively that you are drunk.
If only these people would invest as much effort making sure people don't drink and drive, drunk drivers wouldn't be the cause of for the vast majority of accidents.
In building a breathalizer to detect wayyyyy too much drinking, one would think the first thing they did was extensively test the happy path.
(-1: Post disagrees with my already-settled worldview) is not a valid mod option.
Why did you call him an asshole? He didn't say anything about driving drunk, and it's obviously assumed that everyone is talking about a situation where the driver is sober. But what do you do when a cop wants to test the sobriety? Don't we want to have DWI/DUI laws? If you have these laws, you need some way to investigate suspects. But impaling innocent people (yes, I'm using totally loaded language here, on purpose) is something most of us don't want.
So a whole game develops in the middle ground of uncertainty. Go too far one way, and you don't really have effective laws against drunk driving. Go too far the other way, and you're poking holes in innocent people to make them bleed. Breathalysers were a great compromise .. if only they hadn't tried to keep the inner workings a secret from the very society that judges the accused, thereby making them bullshit "evidence." FAIL.
Open up the breathalysers to auditing and maintenance and you'll have a useful technology. Just like we eventually determine with everything else that's important (e.g. the software in your desktop computer). Anything you can't audit, is bullshit.
Comment removed based on user account deletion
All the chubby chasers would fail....
first, I'm just a generally clumsy person
second, I have a bad knee, actually one really bad knee and another knee that isn't great because of decades of compensating for the other one
those two things together mean that I could be as sober as the proverbial deacon and fail all that mess about walking a straight line toe to heel, standing on one leg, whatever
Jeez, this diverges into rantings about various DUI laws when IMHO the real issue is "Can we prove it's accurate without looking at the source?". IMHO, yeah. You specify the list of requirements, neither knowing nor caring how the magic box does it's magic, and if it passes the tests it's good to go.
If you have half a brain you automate these tests and retest the things every once in a while, like maybe while your taking the blood/urine sample and booking your suspect into jail.
And yes, I've written a few fully automated and some semi-automated test suites.
From the article, the findings from the defense experts were scenarios that could result in false positives. But, what's the false positive rate as a function of the reading? That's what's important. Readings are just samples that estimate the true characteristic. The experts found that readings near the legal threshold could be inaccurate. However, that should have been obvious even without examining the system. The real question is what the confidence levels as the readings get near the legal threshold. That crucial information may not be knowable by only examining the software. However, inaccuracy from a measurement device doesn't mean that the device is useless. All measurement devices are inaccurate. The key to practical usage is understanding the expected confidence in the measurements.
You already failed.
If you are pulled over for something stupid, and you have the slightest wiff of alcohol on you, you are guilty of DUI in most jurisdictions. The breathalyzer and the limit is just to take subjectivity out of it.
The defend you a second time and get paid, win or lose. YAY! I can pay off my student loans.
The Alcotest is a roadside screening device. People who fail the road side test are then taken to a station (or mobile truck during RIDE type programs) for a more precise test.
The roadside screening only has to give the police officer reasonable and probable grounds to arrest the person and continue with the investigation and subsequent test. Even if this result proved 100% that the Alcotest produces a lot of false positives, it casts zero doubt on any convictions or even matters currently going through the courts. The cops would have had RPG on a good-faith basis and every right to have proceeded.
Going forward, it may be possible to argue that with the alcotest proven unreliable it can no longer be used to establish RPG, but even that won't be automatic because the roadside screening device isn't really required if the cop can say their were sufficient signs of impairment to arrest on impaired driving rather than over 80. The over 80 can then be laid at the station if the person blows over.
TL;DR nothing to see here.
Yet another thing self driving cars will fix. Once those are standard, there will be no more DUIs. There will also be no more tickets for speeding, running red lights, failure to maintain control, reckless driving, or any other moving violations.
Sadly, I think cops will hate this. They want to be able to have a reason to pull someone over, especially if they are black (and this is coming from a white male who reeks white privilege). They will also fear this will eliminate a need for a major part of their job. I don't know the number, but suspect a sizable percentage of police activity involves traffic enforcement. If cars are driving themselves, their need will greatly diminish. At some point, municipalities will decide they don't need to pay someone to enforce laws that aren't being broken and there is no longer a threat to society by drunk and reckless drivers. I'm not sure what the final outcome will be, but it doesn't look good for the men in blue.
Ninjas don't carry tic tacs
but should they be driving in the first place.
ban them.
The opinions of lawyers and paid code readers are hardly probative. If there is an objective accuracy or reliability issue testing with many devices under various conditions is needed. Alternatively a formal performance standard and test procedure is reasonable. The FDA or even NIST are well qualified to do this but of course very few are manufacturer or users are willing to cover the costs.
All the chubby chasers would fail....
I am chubby you insensitive clod!