Slashdot Mirror
Breathalyzer Source Code Revealed
Nonillion writes "New Jersey attorney Evan M. Levow was finally able to get an order from the Supreme Court of New Jersey forcing the manufacturer of the popular Draeger AlcoTest 7110 to reveal the source code. Levow turned the code over to experts, Base One Technologies, to analyze. Initially, Base One found that, contrary to Draeger's protestations that the code was proprietary, the code consisted mostly of general algorithms: 'That is, the code is not really unique or proprietary.' In other words, the 'trade secrets' claim which manufacturers were hiding behind was completely without merit." Following up an earlier discussion here, the state of Minnesota has (without explanation) missed a deadline to turn over the code for a different breathalyzer.
The "code" probably digests an 8 bit unsigned char output of a A/D converter, a signal from the "alcohol detecterizer chip", the innards of which are probably proprietary. Then, if [quantized signal] is greater than X, then light the yellow light, if greater than [X+Y], light the red light and make a beep sound.
I want to delete my account but Slashdot doesn't allow it.
I've been thinking about DUI laws in the US and how the laws are just the second coming of prohibition of liquor. Why else would they take two legal activities like drinking alcohol and driving, and make it criminal. Yes, I understand people get hurt by drivers under influence. But lets be real and compare it to teens getting into accidents, senior citizens getting into accidents, sleep deprived individuals getting into accidents, etc. Think about it... You don't see people being tested for reaction speeds when taking driver tests!! You don't see people being tested for intelligence when taking driver tests!! You don't see people being tested for decision making ability when taking driver tests!! If people had to pass these types of tests we wouldn't have so many traffics jams. Think about it, why have some states in the US that use whisky plates (plates for cars owned by individuals convicted of a DUI) run out and had to expand the letters used.
PS People who drink and get into accidents should be prosecuted as if they had reckless intent.
The reason why no source code has been released in MN is that the manufacturer of the breathalyzer in that case, CMI, refuses to hand it over to anyone. They are asserting that it is a trade secret, and are resting on the fact that there is little a court in MN can do to force them, a Kentucky corporation, to hand it over.
I represent three clients in Phoenix, AZ, who have been trying to get the code from CMI for the same reasons, and have been met with nothing but frustration. Fortunately, a couple judges here have agreed with the defense that examination of the code is necessary to mount a defense, under due process grounds. We (myself and a number of other attorneys) have had dismissals in a total of about 11 cases in the City of Phoenix, all of which are being appealed. There are a few cases in superior court that will be appealed shortly as well. It's been a busy time in the world of DUI litigation.
Unfortunately, many judges here do not see the relevance. Further, they have enacted legislation to prevent the preclusion of breathalyzer results, despite the inability to examine the "schematics or source code" of the machines.
Believe me when I tell you - these machines are unreliable, and subject to many errors, most glaringly the result of RFI screwing up the results. I've read the findings of the independent lab on the NJ case, and it does raise many concerns. My biggest problem is that law enforcement can essentially hide behind a foreign corporation, and a jury never hears about many of the problems at hand.
That they don't have any funky algorithms does not mean that the firmware is not a trade secret.
Then what DOES make something a trade secret? The mere fact the software is compiled and/or programmed onto a chip? An EULA? An"anti-circumvention device" as defined by the DMCA? Seriously, where should we draw the line with "trade secrets" when it comes to protective legislation? The only "trade secret" revealed here is the fact that the manufacturer in question embedded alpha-quality software in a product released to production. That sort of a "trade secret" is generally considered willful negligence or fraud.
It still takes significant engineering/test/validation effort to get to a working device.
It is apparent that little to no such QA was done on this particular device, which to me sounds like a grave mistake considering the device is trusted to keep drunk drivers off the road. Keep in mind that this device is theoretically able to report just as many false negatives as false positives, do not only would it be possible for a sober driver to be falsely charged with a DUI (as this lawyer claims) it is also possible that countless drunk drivers falsely blew UNDER the limit and were allowed to continue on their way and put others in harms way. That could be considered criminal negligence on the part of those who engineered this device.
Just because it takes effort (in time and money) "to get a working device" even when there is nothing novel in its functionality does not mean that those putting forth the effort should be able to hide from scrutiny behind a "trade secret". The systems I work on are sometimes involve safety interlocks. My employer subjects their software division's development practices to audits from government agencies. Our clients often stipulate that they must have access to source code (though since we are a closed-source shop we never grant redistribution rights). Even if there are novel implementations or "trade secrets" there are legal instruments to accommodate for them and still remain accountable.
These "breathalyzer" devices used in the field are far from trade secrets--I remember plans for one in Radio Electronics years ago that was said to be quite reliable as a preliminary measurement device (didn't report a specific value, but had a "traffic-light-interface" of 3 LEDs). The "trade secret" excuse is flimsy and shameful. It is worse than the whole Diebold voting machine debacle because it can directly affect a person's safety and well-being.
I've never heard of Base One, and I never would have guessed from their home page http://www.base-one.com/ that they could do what they did, but that has to be one of the more impressive "code reviews" I've seen. The code review encompasses hardware, software, testing, architecture, design - in short, a rather thorough analysis. Seeing all of those skills come together for an embedded system project is pretty impressive. (Consider that, at Draeger, it appears that almost none of those skills were bought to bear...)
So kudos to Base One. Great work.
This is interesting.
The effect is that if your levels are dropping, it magnifies the drop, and if your levels are rising, it magnifies the rise. I'm not certain that this is a problem, as such. It might even be a benefit, on some level (especially if your levels are dropping.) Read on to see what I mean.
If you were to take ten readings, and they were rising from "1" to "10", then the correct average according to the report would be the sum of 1 to 10 divided by 10, or 5.5. That is, you would have ten readings and the average of them would be 5.5 even though your present reading was 10. This artificialy minimises your reading. Using the machine's algorithm, your "average" reading would be 9.001953.
If, on the other hand, your levels were descending from 10 to 1 in the ten readings, then the "correct" average is 5.5 still, but the machine would say your reading was 1.998047. Which is closer to your final reading.
OK, so what does this mean efffectively? Well, I guess the biggest concern is a residual alcohol effect, where you ahve alcohol in your mouth either from the beer you had between your knees, or from the mouthwash you quaffed at the last second. In this case, you would expect to spike the first reading. The machine's algorithm takes the spike into account, but minimizes its effect on your overall reading.
It should be noted that both algorithms have the same result for one or two blows. To maximise your results, you would need to insist on more than 4 blows, and only if the average reading was dropping each time.
10001001111001110110011000011101110
It would be wise not to rely on your logic when you are drunk.
The cheaper solution is to use logic earlier.
Yes, 0.08 is a bit low. You can get a DUI/DWI with lower than that, too.
Yes, it's a bit of a witch hunt. Banning giant parking lots at sports bars would more effective.
Besides, if it's deaths you are concerned with, a governor limiting car speed to less than 75 would save many lives, and be much cheaper than random tickets of patrol cars. But that may impact the sale of certain cars. Corporate profits are more important than lives of individuals of course.
I am still licensed to be a medical technologist, I spent a lot of time working in hospital chemistry labs with computerized equipment, and that software fails all kinds of reasonable criteria for calibrating and operating any equipment.
There's plenty of studies showing that talking on the phone impairs drivers more than a 0.08 BAC. If we accept the premise that drivers should be criminally liable for driving while impaired that means one of two things: a) the laws for driving while phoning are too lenient or b) the legal limit of blood alcohol content is too low.
I pull imapired drivers over all the time. Yeah I'm frequently called a dickhead for charging a cellphone user with failure to control or assured clear distance, but so be it. "Ma'am, I stopped you because you appear to be impaired. You've put your entire car over the line at least seven times since I've been following you in the last mile and a half. Is everything OK?" "Oh yeah, just fine, I haven't been drinking or anything, I'm just talking on my phone." (More than half the time the driver is still on the phone when you approach the vehicle).
Do I think it sucks that a person pulled over for DUI faces significantly higher penalties even if they don't cause an accident versus someone simply not paying attention? Yeah, but I don't write the laws. Now, there is a difference between distracted driving and driving drunk. A distracted driver who is on the phone, putting on makeup, etc, can turn it off and become 100% focused on driving again. A drunk can't do this. They're impaired until the alcohol wears off. Thus the exposure time for a drunk driver is 100% of the time that they're behind the wheel. If traffic gets hairy or the weather turns bad, a distracted driver can hang up and drive. Am I implying that all distracted drivers do this when the situation requires more attention or am I saying that when traffic is light you're free to goof off? Absolutely not, but at least the ability to focus again is there while drunks don't have the option. My policy is if you're driving like you're impaired you probably are and I will make contact to determine what the issue is.
Personally I like non-DUI impaired driving cases. It's much easier to write someone a citation or a warning and send them on their way than to have to go through the rigmarole associated with a full-blown DUI arrest. In my mind both groups are equally dangerous on the road, but the former is much easier to deal with and get back to patrol. A drunk might take two hours from violation to calling 10-8 (back in service).
You cannot be compelled to testify against yourself.
You can refuse any and ALL tests which may incriminate
you. Easy.
Besides, cheap ass whore cops will hold you for 30 minutes
to allow the alcohol you drank a hour ago , to reach limits that
WILL make you over the mark. They are cowards and criminals. Simple !
The GNNA should be involved in this discussion!!!