DUI Cases Thrown Out Due to Closed-Source Breathalyzer


Hundreds of cases involving breath-alcohol tests have been thrown out by Seminole County judges in the past five months because the test's manufacturer will not disclose how the machines work.

I think this is huge. (Think of the implications for voting systems, for one.) And it's the right decision. Throughout history, the government has had to make the choice: prosecute, or keep your investigative methods secret. They couldn't have both. If they wanted to keep their methods secret, they had to give up on prosecution.

People have the right to confront their accuser. And people have the right to a public trial. This is the correct decision, and we are all safer because of it.

Posted on September 16, 2005 at 6:46 AM • 72 Comments


ExcuseMeSeptember 16, 2005 8:16 AM

Wish I had something useful to say, but I just couldn't resisit being the first to respond to this post.

Tim VailSeptember 16, 2005 8:19 AM

Makes me think about if the same standards is necessary for performance enhancing drugs in sports? For some reason, I think the standards should not be the same in sports.

In any case, I think that people who get caught for DUI often are not likely to bother to try to find alcoholic beverages that fool the breathalyzer. The information getting out about how it works should not be an issue.

JosephSeptember 16, 2005 8:22 AM

With sports, the player most likely has to sign some kind of waiver about drug testing by a private process, etc. In DUIs, you are being accused by the state and you should have more rights (like the source code, for example). That's just my opinion.

Mike G.September 16, 2005 8:23 AM

The problem is it's not widespread. Most other venues don't care if the internals of the device can't be examined, be it a breathalyzer or a speed gun. (As the article points out in the last paragraph)

TanukiSeptember 16, 2005 8:23 AM

Ah, if only such openness were required for automated prosecutions in the UK; then we might at least have an escape-clause for the dreaded "Gatso" speed-cameras and the LTI-2020 speedguns which time brick walls doing 30MPH...

Chris WalshSeptember 16, 2005 8:24 AM

A couple of observations:

1. The herding behavior of the judges is interesting

2. The variation across counties is disheartening.

IANAL, so I am not sure what it would take to clarify this, even within a single state. I suppose if constitutional issues arise even in a misdemeanor case, they can be considered all the way up to the state supreme court. That would be nice to see.

The implications are potentially staggering -- a very wealthy defendant might demand access to the source for fingerprint analysis SW, forensic SW, database SW...the mind boggles.

A Devil's Advocate point:

Why does it matter *HOW* the device works? What if, as part of the normal maintenance/operation of the equipment, technicians calibrate it using known samples of both "drunk" and "sober" air (or whatever), and as long as it performs properly across a suitable range of samples, it is considered to work? Isn't this exactly what is done with, for example, radar guns?

dmcSeptember 16, 2005 8:31 AM

It would seem that the manufacturer should be able to provide a convincing amount of information about how the test works without compromising their trade secret. Or at very least provide evidence that it does not have a high false positive rate (which is all that a defense attorney should care about anyway.) If the manufacturer cannot provide such data, why would the state invest in using the test in the first place? So why not share it at trial?

Fred PageSeptember 16, 2005 8:33 AM

"Why does it matter *HOW* the device works?"
-Because if you don't know how it works, you can't defend yourself against that evidence. For example, the fact that a breathalizer works in a clean lab 99.8% of the time does not mean that it works even 90% of the time at a traffic stop with a person who just used a large amount of alcohol-based cologne (on their face) or is stopped in high winds. Simuarly, a software defect could skew the results in either direction.

The difference between athletics blood tests and breathalizer tests is that most athletes have access to the blood (they can keep a vial at the same time as they give one), for which they can give their own independant anaylsis. The time between the initial breathalizer tests and when you first can get an independant alcohol test is often large enough to skew the results.

Roy OwensSeptember 16, 2005 8:36 AM

Does it matter how it works? Absolutely. We don't know that it works until we know how it works.

The detector must be sensitive to the phenomenon in question, of course, otherwise it would be useless.

But the detector, to be put to any good use use, has to be specific to the phenomenon of interest. Otherwise, when it shows positive, we have no idea what it shows positive for.

It used to be thought that bearing false witness was evil. But now that the witnessing is automated by technology, false positives fatten the government coffers, and the judicial system shows undue unconcern for honesty.

JosephSeptember 16, 2005 8:43 AM


"Does it matter how it works? Absolutely. We don't know that it works until we know how it works."

I agree that we need to have the source code for defense purposes, but I don't agree with you here. The purpose of scientific inquiry and statistical analysis is to prove that things work without really knowing how. We have only had a glimpse how gravity works for eighty years or so, and we're still not completely sure how, but we have known it works for several thousand years.

SteveSeptember 16, 2005 8:51 AM

"People have the right to confront their accuser. And people have the right to a public trial."

Unless, of course, you're declared an "illegal combatant" and incarcerated without trial, access to counsel, or even a clear depiction of the charges against you, c.f. Jose Padilla.

DocSeptember 16, 2005 8:54 AM

This is absolute crap. So someone may be DEAD or maimed due to a drunk driver, and the drunk driver walks free because of some arcane DRM issue? Schneier, you are dead wrong on this one, bud.

Dimitris AndrakakisSeptember 16, 2005 9:01 AM


> Or at very least provide evidence that it does not have
> a high false positive rate (which is all that a defense
> attorney should care about anyway.)

IANAL but I think that, even with a small rate of false positives, a case against the test can be made.

I liked very much Roy's analogy with false witnesses. Why should we accept a machine we know it's lying (even if it happens rarely), when we do not accept a human that's lying ?

paulSeptember 16, 2005 9:01 AM

If there were a large enough body of tests under a wide enough variety of conditions (performed under circumstances that everyone had reason to trust) then you might not need to know how the device works, but otherwise, well, geez.

If the device is only tested in the lab against a few samples with varying levels of faux-exhaled alcohol (or metabolites) in it, you don't even know whether it's actually measuring alcohol, or just all molecules above a certain size/mass. (Obviously that's an unrealistically extreme example, but that's the problem with black boxes: you really don't know what's going on.)

The theory behind gizmos like this is (from some literature) easily solid enough to support the results, but without solid information about how that theory is implemented, you could have pretty much anything in there.

Dimitris AndrakakisSeptember 16, 2005 9:07 AM


There other ways to provide evidence for a drunk driver. What happened to the traditional "walk-the-line" test ? Or the police officer stating that "he smelled whiskey from a mile away" ?

IMO, Bruce states clearly that the matter is very serious because of the side-effects it could produce:

> Think of the implications for voting systems, for one

Jacob AppelbaumSeptember 16, 2005 9:19 AM

I was hit by a drunk driver when I was a teenager. I was walking and *smash* I went flying. The guy was drunk and there was no doubt about it - they took blood samples. It's an open process that allows for faith in the system. I have no such faith in closed systems and could not in good faith ask for someone to be tried with such evidence. Everyone makes mistakes.

For all those supporting the maimed victims of drunk driving, you're missing the bigger picture. It's not just about drunk driving, it's about justice. This is the right choice. Let a guilty man go free rather than sending everyone that a machine convicts to jail without recourse.

While I realize that I lived, it's not just a matter of death or maiming. It's a matter of people being screwed over and breaking not just the letter of the law but the spirit.

For people to just let their bitterness and their so called compassion take over, you've missed the boat and you're just being spiteful.

Not in my name thank you very much.

Clive RobinsonSeptember 16, 2005 9:28 AM


"Schneier, you are dead wrong on this one, bud"

Fortunatly Bruce is right, there are correct tests that can be applied to check for blood alchol levels, if the Police / State are to lazy or tight fisted to do them, then it is upto them to show a judge that an offence has been commited without evidence... Otherwise what's to stop a police officer pulling you over saying you are drunk, just to get his quota of arrests?

"Wrongfull arrest" and "letting villans go" are emotive statments, and should have no place in a fair and just system.

It used to be the case in the UK that if you where pulled over and brethalized and it was positive, you where then taken to a Police station where you where given a choice submit to a blood test, or go before a judge without, in which case it was upto the Judge to decide, on the testomany of the defendent the police officer and any wittnesses their might have been.

linnenSeptember 16, 2005 9:56 AM

@ Doc

Disclaimer: IANAL

If someone dies or is maimed in an accident, why would one go for the 'cherry on top' DUI charge? The main charges should be some variant on manslaughter or homicide.

DUI charges are a preventative measures to get the 'drink and drive' people off the roads before anyone is killed. Faulty testing equipment, here breathalizers, just get in the way of doing this.

Roy OwensSeptember 16, 2005 10:12 AM


I worked at a facility with a lot of expensive equipment that got a latest-and-greatest fire detection and suppression system. The detectors were tested by squirts of Freon. However, when actual smoke -- from a burning hard drive -- wafted across the detector, it failed to notice. A human smelled smoke and used a hand-held fire extinguisher to put it out while the automated system sat there stupidly. Later a tech came and tested the detector, pronouncing it just fine. Later an engineer checked the system and pronounced it ineffective.

The detector was supposed to detect smoke. It did not.

Why would you trust a detector unless you knew how it worked? If it fails to detect what you're worried about -- a false negative -- you're screwed. If it announces it has detected something that isn't there -- a false positive -- you're fooled, unless you're smart enough not to trust it.

DavidSeptember 16, 2005 10:16 AM


When a chief judge of a county comes out with a ruling on something they generally expect the other judges to comply with that ruling. This is the case in Seminole County. A few of the judges are still not following his guidelines.

Also, please remember that their are issues here with chain of evidence and specifically with evidence collection. As we have seen over and over if you screw up collecting evidence it is (and should be) grounds for removing the evidence.

Florida has freed a large number of inmates (including death row inmates) recently through new DNA testing that outright proved that these "convicted killers" were in fact innocent. The original evidence was obviously faulty, or simply incomplete yet a jury found them guilty. These inmates are not always compensated for their wrongful incarceration by the state (an apology currently is the only requirement).

DavidSeptember 16, 2005 10:23 AM


Sorry, but if a death is involved the charge is Vehicular Homocide. Proving they were drunk is usually done via a blood test (that does not use 'software' :) if anyone is severly injured (or dead), including the victim.

The charges being thrown out are DUI charges that are only being done with a device that the defense can not independently evaluate (properly, since the software not chemical test is the one that presents a result).

VickiSeptember 16, 2005 10:35 AM

We do know how gravity works: we know what will happen when something is dropped in a gravitational field, how it diminishes with distance, the ballistics of spaceships, and how gravity affects Mercury's orbit. We don't know "why" in some philosophical sense, and it's true we don't need to. The problem isn't that we don't know what makes the detector sensitive to alcohol. It's that we don't know how reliably it is sensitive to alcohol, or how often it will say "alcohol" when there is no alcohol there. And we don't know whether it will say "lots of alcohol" when the person drank only orange juice, but kissed a beer-drinker right before getting in the car.

The problem with DUI deaths isn't primarily the lack of breathalyzer tests: it's that we're far too willing to say "yes, this person drove recklessly and killed someone, but we can't prove they were drunk so it's not a crime." If someone runs over a pedestrian, it doesn't really matter whether they were drunk, over-tired, or just stupid. It matters that they operated a dangerous tool in a way that hurt another human being.

Mike SherwoodSeptember 16, 2005 10:52 AM


Walking a line, saying the alphabet backwards, touching your nose, etc are all subjective tests. These are worse than the breathalyzer because they're nothing more than a justification for an officer's opinion. Being able to close your eyes and touch your nose is also a test to diagnose nerve damage.

Smelling like alcohol is also a poor test. I had to drive home smelling strongly of beer because someone who was drunk spilled it on my jacket. I hadn't had anything to drink, but I made a point of avoiding cops anyway.

The laws against DUI generally refer to a specific BAC. That is the Blood Alcohol Content. The blood test is considered clear enough to be the basis for many, if not all DUI laws in this country.

If arrested for DUI, insist on the blood test if you're innocent. If you're guilty, hope the officer is using a NabU 2000 that can be challenged for technical reasons long after the fact.

I suspect there were a large number of cases thrown out that were legitimate. However, it's the duty of the prosecution to demonstrate the validity of their evidence. Independent testing and certification are likely to be required for breathalyzers in the future. Hopefully the testing is done by people who are intentionally trying to get false positives/negatives.

Davi OttenheimerSeptember 16, 2005 11:01 AM

I thought this was an important note in the article: "conviction rate has dropped to 50 percent or less"

So DUI convictions are obviously still possible, without this particular piece of technology.

"as long as it performs properly across a suitable range of samples, it is considered to work? Isn't this exactly what is done with, for example, radar guns?"

This is definitely an approach that was taken with radar guns, but only because they were not closed-source. In other words the accuracy had to improve because it was widely revealed how innacurate they were on the whole. The developments actually led to laser and lidar and other more sophisticated technology with far higher accuracy.

More to the point of the article, however, this indicates (at least some) judges are getting smarter about snake-oil in the courtroom and demanding real proof of vendor claims.

docSeptember 16, 2005 11:10 AM

"Unless, of course, you're declared an "illegal combatant" and incarcerated without trial, access to counsel, or even a clear depiction of the charges against you, c.f. Jose Padilla."
Good. For once this country is doing it right. Kill 'em all, then let Alla sort 'em out.

Bruce SchneierSeptember 16, 2005 11:11 AM

"This is absolute crap. So someone may be DEAD or maimed due to a drunk driver, and the drunk driver walks free because of some arcane DRM issue? Schneier, you are dead wrong on this one, bud."

You're kidding, right? I believe in the principle of "innocent until proven guilty." There isn't a drunk driver until he is proven drunk, and an ethical justice system gives him the right to defend himself. Being able to examine the evidence against him is part of that right.

We are all more secure because of these rules. A government who can prosecute people based on secret evidence approaches a police state.

Chris WalshSeptember 16, 2005 11:13 AM


I didn't see that the article identified the judge whose example was being followed as higher-ranking, although of course that makes perfect sense.

W.r.t. chain of evidence, I am in total agreement with you. Living in Illinois, I have some local familiarity with the issue of folks being railroaded onto death row, unfortunately.

On the knowing that it works, but not how it works -- I personally believe that the public welfare is better served when defendants (and the public) can see "how it works". Indeed, the dishonesty of various crime lab scientists in the news recently (including the FBI, IIRC) suggests to me that the maximum amount of transparency should be required as a control against outright lying by experts as well as accidental errors by equipment designers/coders.

Nonetheless, I think that one could make a somewhat persuasive case that as long as the devices have been tested in *realistic* environments, etc., one needn't know exactly what they are doing in order to have an extremely high confidence that the results they report are accurate.

DocSeptember 16, 2005 11:13 AM

Why should it matter if the details of the innerworkings are not revealed, when there is an ENORMOUS statistical basis that the technology works correctly?

Tim VailSeptember 16, 2005 11:34 AM


I'd think because you don't know how it fails. If you don't know how it fails, then how do you know it isn't likely to fail in this particular case?

Jonas GrumbySeptember 16, 2005 11:36 AM

> Why should it matter if the details of the innerworkings are not revealed, when there is an ENORMOUS statistical basis that the technology works correctly?

It's the principle. If you were charged with a crime, and the main piece of government evidence were a black-box "99% reliable (in laboratory conditions)" guilt detector, how could you possibly mount a defense?

jammitSeptember 16, 2005 11:57 AM

All these cases are being thrown out because of the breathalyzer? A cop wasn't doing his job. You never use one thing to nail someone with. Every cop knows to use multiple things to test with. If you've ever watched a reality TV series involving street cops, you know when they pull over a guy who's weaving really bad, they pull him over, talk with the guy using small talk, breathalyze him, make him do the drunk dance, and then slap cuffs on him. The weaving is caught on camera, the slurred speach is caught on audio, the breath is smelled, the drunk dance is caught on film, and lastly the breathalyzer results are used. A radar (or other) gun isn't used to convict you, it only gives a reason to pull you over. The cop dash board speedometer, mile markers and a watch, and radar gun are used to convict you. I'm personally glad these were thrown out. Maybe with this wake up will slap the cops into shape and really keep us law abiding citizens safe.

Davi OttenheimerSeptember 16, 2005 12:22 PM

@ jammit

Yes, you're right that things are *supposed* to work that way, and perhaps you are a lawyer/judge making a plea for better justice, but I have personally seen judges place the burden of proof for accuracy of a new technology squarely on the defense.

"A radar (or other) gun isn't used to convict you, it only gives a reason to pull you over. The cop dash board speedometer, mile markers and a watch, and radar gun are used to convict you."

We hope, but I am familiar with a judge who has believes laser-based speed traps are 100% accurate and stand on their own.

I agree that an officer actually starts with a "hunch" or observation and uses whatever evidence available to prove their case, but their behavior changes dramatically when they are led to believe that they only need one test.

Fred PageSeptember 16, 2005 12:27 PM

"I agree that we need to have the source code for defense purposes, but I don't agree with you here. The purpose of scientific inquiry and statistical analysis is to prove that things work without really knowing how..."

I think your gravity analogy is poor. Very little software is as reliable as gravity.

In an example from a previous job, we had a gambling game that had an X% payout (where X is a fixed number less than 100%). We did a paytable anlysis, analysis of the random number generator, had an independant contractor do independant analysis and code review, and (in this case) months of running on hundreds of machines in the field, all of which pointed to this game having a X% payout.

Once we found a defect in the game, we realised that this was wrong. Anyone who (accidentally or deliberately) hit the defect could get an overall payout percentage of 100+ %. Needless to say, this was illegal in said juristiction, and the game was pulled later that same day.

ProbitasSeptember 16, 2005 12:47 PM

Once per year, I must provide the report of a qualified, independent technician who has tested the smoke detection and fire alarm systems at all of my facilites, and found them to be reliable. this is on top of the burden to show that the system gets tested and fire drills are run regularly. This tech needs to test not only that the device does what it is supposed to do, but that it is operating within design specifications.

I think that this is an example of the same standard being applied. If the company will not disclose what is to be tested, i.e. the methodology used to determine "drunkeness", the technician cannot determine if the equipment is functinoing properly.

Rhetoric about individuals killing innocent families aside for a moment, what is a breathalyzer but a form of an expert witness for the prosecution. We are routinely given the opportunity to question witnesses to establish their proficiency in the subject at hand, I don't see why it should be any different for the machine.

AdaSeptember 16, 2005 1:12 PM

Wow. If the gov't has the burden to prove software reliability, I think I'd like to become a criminal defense att'y.

David ThornleySeptember 16, 2005 1:13 PM

Doc said:
Why should it matter if the details of the innerworkings are not revealed, when there is an ENORMOUS statistical basis that the technology works correctly?

Given a closed-source system, you don't know that there is any statistical basis at all, unless somebody has conducted a study of many examples of the exact same device. Simply, you don't know exactly what technology is being used, and you don't know if it's being used correctly. Even with massive tests, it's really hard to be sure that there aren't conditions where it will normally be wrong.

I'm in software myself. Testing a closed-source system is the ultimate in black-box testing of a potentially extremely complicated system. There is no end to the questions you can bring up (consider some of the questions here: high winds, exposure to another person, beer-soaked jacket, confusion with other molecules), and each of them has to be addressed to gain confidence. Open up the system and it's possible to test parts of it and ask intelligent questions, and it's also possible to reason from systems you know are similar.

For these reasons, it is virtually impossible to trust a closed-source system containing both hardware and software. It can serve as evidence that warrants collecting further evidence, in this case a blood test, but it cannot be relied on as the only evidence.

EvanSeptember 16, 2005 1:23 PM

Bruce: Innocent until proven guilty is the statement that is claimed, but remember there is the addition of reasonable doubt used in cases.

I think the prime example is with cable theft (see: http://www.pirateboxblues.com/#BURDEN). Owning it doesn't make the person guilty, but what reasonable person would think that they aren't using it for something they don't own?

There aris one other factor: Driving is not a right, it is a priveldge (this is what I'm told by people when I say I think sobriety checkpoints are illegal). People don't have the right to drive, they have to have the skill set and responsibility to drive.

jblSeptember 16, 2005 1:44 PM

"If someone dies or is maimed in an accident, why would one go for the 'cherry on top' DUI charge? The main charges should be some variant on manslaughter or homicide."

I(also)ANAL. But I believe that in some circumstances, or in some states, the presence of a blood alcohol level above a certain amount is an element of the crime of homicide and at the least would affect what charges were brought. It might also affect the determination of fault.

Bruce SchneierSeptember 16, 2005 1:56 PM

"This is absolute crap. So someone may be DEAD or maimed due to a drunk driver, and the drunk driver walks free because of some arcane DRM issue? Schneier, you are dead wrong on this one, bud."

And another thing, in which jurisdiction does a driver who kills or maims someone get convicted if drunk, but walk free if sober?

ARLSeptember 16, 2005 2:04 PM

If the question was asked about the formula of the steel used to make the case of the device, and the answer could not be produced, should the DUI be thrown out?

If I take a picture of a crime with a digital camera, and there is no source code available, should the picture be thrown out?

Police in Florida already have to have a resonable belief that the person is under the influence before they can ask for a BAC test.

The breath test is quick, cheap and painless. If negative the suspect is cleared and can quickly leave. My guess is they will start to ask for blood draws (permitted under Florida law) wich will take longer to process and result in innocent people getting to spend the night in jail.

Either the device works, (produces an accurate result) or it does not. This looks like a case where the judges have decided that they don't like the DUI laws and little about rules of evidence or peoples rights.

tcSeptember 16, 2005 2:12 PM

I was a jurist on a DUI case a few years ago. I was amazed at what gets by for scientific acceptance of the breathalyzer machine's accurracy. The machine in the trial I was a jurist on had a "supposed" test sample attached to it that is supposed to be a known concentration (0.08). The problem is the state lab that produces this sample tests the entire batch that they pull the concentration from and not the individual samples themselves. Thus, they have the classic error of batch sampling (see writings by Deming and others on the problems with batch sampling). They assume that the individual draws of the concentration match the entire batch, but never statistically test this assumption.

In the case I was on, it didn't matter however, due to the fact that the participant admitted to the officer that his driving may have been a little impaired. By the state law, this combined with the field soberiety test was enough to find the alleged driving under the influence.

Personnally, I think what should happen is that a blood draw should be taken and tested (from certified hospital/lab), and not rely on the breathalyzer machine. Some states do use blood draws instead of breathalyzers.

Bruce CleaverSeptember 16, 2005 2:48 PM

To those who recommend blood draws vice breathalyzers - doesn't that just push back the debate one level? Whatever code/assumptions go into the mass spectrometer (say) used for checking blood are then fair game for the open source/closed source debate.

Davi OttenheimerSeptember 16, 2005 3:00 PM

@ Bruce

"in which jurisdiction does a driver who kills or maims someone get convicted if drunk, but walk free if sober"

Good question. It reminds me of a tragic twist of fate that was briefly mentioned in the news earlier this year:


"An unlicensed driver who fell asleep at the wheel of a car that crashed, killing a passenger, has walked free from court."

That might shock some at first glance, but it's even more unusual when you review the circumstances and judgement:

"In his sentencing remarks, Judge Punshon said Fosio had not wanted to drive his friend's car.

But he did so because his friend had insisted, saying he was too drunk to drive.

Judge Punshon said Fosio's motivation for driving was to prevent the deceased man from getting behind the wheel."

Fred PageSeptember 16, 2005 3:13 PM

@Bruce Cleaver

The typical thing to use for blood alcohol content would be a clinical chemistry analyzer, not a mass spectrometer. I agree with your point, however - particuarly in cases where the person has no opportunity to gain an equivalent sample and have it independantly verified.

Note that even with a blood test, there are many ways to mess up the results, including preparing the area to draw blood from with alcohol.

Ari HeikkinenSeptember 16, 2005 3:15 PM

I think this is simply stupidity. Everyone knows that breath-alcohol tests are only good for checking if there's a reason to believe someone's drunk. You need a blood test to determine it precisely. In court, it would be reasonable to ask how precise the machines are and perhaps even have an expert to testify about the technical details, but asking for source code is plain silly.

I personally know someone who was breath-alcohol tested and the meter showed "over the limit" and was taken to a blood test. When the results of the blood test arrived they barely showed anything (he had one beer before driving).

P.S. I'm always amazed about Bruce's position how about everything should be disclosed. I can't wait for someone suing Bruce and demanding source code for counterpane's custom monitoring systems.

Bruce SchneierSeptember 16, 2005 3:35 PM

"I'm always amazed about Bruce's position how about everything should be disclosed. I can't wait for someone suing Bruce and demanding source code for counterpane's custom monitoring systems."

I don't believe everything should be disclosed. On the contrary, I think of myself as very pro privacy.

I don't even say that open-source software is obviously more secure than proprietary software.

Secrecy's relationship with security is very complicated, and I talk about it some here:

With regards to government, I believe that transparency is one of our most important security measures. And that's the principle that's in play here. Secret police procedures simply aren't healthy for society.

Davi OttenheimerSeptember 16, 2005 3:38 PM

"I can't wait for someone suing Bruce and demanding source code for counterpane's custom monitoring systems."

That's funny, but not really comparable since it Bruce offers a paid service that people opt into voluntarily. As far as I know it is not something used by law enforcement to judge compliance/guilt. Note that the judge ruled "Florida cannot contract away the statutory rights of its citizens".

Davi OttenheimerSeptember 16, 2005 3:49 PM

@ Bruce

"I don't even say that open-source software is obviously more secure than proprietary software."

But you have advocated for public security over proprietary in the Cryptogram:


"Public security is always more secure than proprietary security."

Maybe that is the source of confusion?

Brian ThomasSeptember 16, 2005 3:54 PM

The tradition in law that started with the "presumption of innocence" clause in the Bill of Rights, and led to the "reasonable doubt" standard, also motivated the "prior restraint" test, particularly in reviewing laws restricting certain kinds of speech. In the past, laws against certain activities, or possession of certain kinds of items, because of their association with criminal acts, have been struck down on the basis that they constituted prior restraint of criminal activities.

It is interesting (and disappointing) to note that there is so much fear of crime and terrorism that an unhealthy emphasis on prevention has grown, even to the point of prior restraint, at the cost of the precious and hard-won Constitutional guarantee of presumption of innocence. Bruce's earlier link to the article on crime-facilitating speech is an example; this and the attempt to shut down file-sharing networks, the DMCA itself, Sarbanes-Oxley, and all kinds of juvenile restrictions like forbidding hats in banks plus all the security "theatre" we like to bash all point to a childish obsession with avoiding risk that is not only ineffectual but risks losing all the freedoms we claim to value.

Many of the things I mention can (and should) be used as evidence of intent and/or capacity to commit a crime, when a crime has actually been committed, but in themselves they do not constitute crimes, nor are they acceptable as the only evidence of a crime having been committed.

In the case of the breathalyzer, the question is whether reasonable doubt does indeed exist that its indications are accurate. As others have intimated, establishing reasonable doubt of the accuracy of a single reading of a device (whose internal workings are not as simple or well-characterized as, say, gravity) does not require proof that the device has ever malfunctioned. Further, it is perfectly reasonable that a defense attorney should make use of the unwillingness of the manufacturer to reveal enough detail to convince a jury that no reasonable doubt of the device's reliability can be entertained.

The question is not whether drunk drivers were inappropriately exonerated; I have little doubt that they were. Indeed (having not read the article yet), I don't really entertain any serious doubts about the device. But because of the manufacturer's secretiveness, reasonable doubt does exist, and we can never know.

I also don't believe that emotional argument is inappropriate; it's extremely important that crimes that cause great harm should be prosecuted with great ardor. But remember that the greater the harm and the injustice of releasing the perpetrator, the greater the injustice of convicting an innocent party. Our justice system has had, at least until now, a strong constitutional bias against condemning the innocent, with the full knowledge that we thereby risk excusing the guilty. I believe that this is right.

It is quite true that justice has miscarried. But the fault lies neither with the judge who made this decision nor with the attorney who raised the issue, sleazy though he may have been. Rather, the fault was with the judges who relied solely on evidence that could be so challenged, and the resultant arrest policies of the police forces. And ultimately, with the attitudes of a nation that is less willing to be properly diligent at seeking justice than to fight to stop something they fear without fully considering the consequences.

Ari HeikkinenSeptember 16, 2005 3:56 PM

"As far as I know it is not something used by law enforcement to judge compliance/guilt."

Ok, here's another analogy: Counterpane detects an attack on one of its client's networks and provides the evidence needed to take the case to court. The attacker demands to see Counterpane's monitoring system's cource code because he argues the evidence is not reliable. What's the difference?

Ari HeikkinenSeptember 16, 2005 4:08 PM

I totally agree that transparency is one of the most important security measures when it comes to our governments, but to what extent to allow nitpicking in courts? I mean, following those source code disclosure demands why not also demand exact PCB printouts with proofs of correctness? Where to draw the line?

Ashley GlsselSeptember 16, 2005 5:41 PM

Here in Victoria, Australia, breath test devices are only used as an indicator. If you fail you a breath test via a hand hel device you are then lead into a bus and they the police take a blood sample for testing later. The police will prosecute based on the results of that blood sample. (We call the buses "booze buses" :-) )
This system seems fair and just, and we don't really need to know how the breath tester works, only that it is reliable because a lot of people could get upset about fivolous blood tests.

Roy OwensSeptember 16, 2005 6:15 PM

Another issue of the fairness of automated DUI detection comes thanks to the police policy of pre-test screening: when the cop stops you, he first has to find out who you are before he can know whether to (1) wish you a nice day, (2) give you a police escort home, or (3) nail you on a DUI. Note that in cases 1 and 2 the arrest is run off the books; only case 3 shows in the records. The police generally run most of their business off the books, a dishonest practice.

The people who matter are not subjected to the the testing or to any attendant injustices: the testing is not democratic.

Worse, traffic stops are increasingly treated as opportunities for revenue enhancement, finding every possible excuse to issue citations and make (on-the-record) arrests.

Moreover,the gizmo in use was likely selected not on the basis of performance but simply because it was cheapest.

another_bruceSeptember 16, 2005 11:56 PM

the bill of rights gives criminal defendants in america the right to confront witnesses against them. when a test is offered into evidence, defense counsel has the right to ask everything about how that test was done and how the result was computed. the proprietary interest of a corporation cannot trump a constitutional right under our system. as thomas more observed, if you cut down all the laws to get at the devil, what are you going to do when the last law is down and the devil turns round on you?l

DarkFireSeptember 17, 2005 4:12 AM

I believe that the judges in these cases made absolutely the correct decission.

Sometimes it is really is more important to maintain the secrecy of a method of intelligence gathering than it is to stop the actions of those being watched.

This is a delicate call to make and the above is usually only justifiable in the most serious of situations, with the most extreme of necessities.

In the case of a blood-alcohol measuring system this plainly and obviously just doesn't apply. I could understand if the company manufacturing the equipment wishes to maintain industrial secrecy, but surely all this would require is for a company employee to apply to give his evidence in closed court?

What adds to the stupidity is that preventing drink drivers from driving is a very necessary thing - it prevents them from killing people by their actions.

As I said, sometimes total secrecy is necessary, but this is very definitely NOT one of those occasions.

Bruce - with all due respect, this is a nonsense:

"Secret police procedures simply aren't healthy for society"

Yes I understand people who are suspicious of the Police and all they do, but to divulge everything would have serious implications not only for public safety but also for the personal safety of Officers and their families. Let's rememebr that once a copper takes off his uniform he/she also wants to go home in one piece.

By all means disclose everything that would not impact on effectiveness & safety, but some things should and must remain confidential.

VWMSeptember 17, 2005 10:07 AM

"(Think of the implications for voting systems, for one.)"

Funny you mention this. The German Federal Ministry of the Interior rejects to publish a Test report about the voting machines that a going to be used at some polling stations during the parliamentary elections (Bundestagswahl) on 18-09-2005 [1].

Furthermore, the manufacturer (a Dutch company called Nedap [2]) of those software controlled machines rejects do disclosure the source code [3]. So besides Nedap nobody knows how that thing actually works. Not even the officials.

There is an English Report from the Irish Comission on Electronic Voting about the voting system [4].

[1] http://www.heise.de/newsticker/meldung/print/...
[2] http://www.nedap.nl/
[3] http://www.spiegel.de/politik/deutschland/...
[4] http://www.cev.ie/htm/report/first_report.htm

Paul OSeptember 18, 2005 1:50 PM

@Bruce, Sep16, "in which jurisdiction does a driver who kills or maims someone get convicted if drunk, but walk free if sober?"

IANAL, but in many jurisictions, including Ontario (Canada), there is a "presumption of fault" when alcohol is involved. That is, the other driver may have caused the accident, but if you have a BAC over the legal limit, the entire legal burden falls on you instead. The scenario you outline is entirely possible, although details of particular cases are not widely available here.

On the other hand, it is reasonable to test the evidence at trial; it is not reasonable to put the evidence itself on trial at every instance. The device certification procedure should be public, and should be where it is determined whether the device produces fact or fiction (and exactly what facts it indeed produces).

If a defendant wishes to challenge the certification procedure (and full disclosure rules say that you know what evidence is coming your way at trial), the judge should defer the prosecution on the charge until that legal challenge is settled in a separate action. (Perhaps handling it as he would other motions during trial.) A judge should not confuse the trial of the accused with a trial of the certification procedure.

ChrisSeptember 21, 2005 2:07 PM

Several other posters have noted that police often use digital photographs, videotapes, and the like as evidence at trial.

In US courts, these things don't have that much force. What does have merit is that one side will always have an expert or witness recount the events and testify under oath that the photograph, video, whatever, is an accurate account of what they observed. (I'll leave out "administrative proceedings" that have become so popular in the last few years)

In Texas, when you are issued a citation for speeding, at trial the officer will usually testify (often in a rapid-fire exchange between him and the prosecutor) to the following:

1) That your high rate of speed attracted his attention to you,
2) With his police training and years of experience, he is a trained observer, and
3) He estimated your speed as X mp/h,
4) he used his radar gun/laser gun/stopwatch/etc. to verify his estimate, and
5) the posted limit was Y mp/h, and
6) your speed was above what was reasonable and prudent.

The readout on the speed-measuring device is only used to confirm what the officer directly observed.

Now, police departments love numbers. It's a line in the sand, and if you can convince a jury that line was crossed you're pretty much assured of a conviction. Numbers remove the opinion out of measuring compliance with a law; inviting opinion about compliance invites opinion about the law itself and that just complicates matters. Numbers, however, are simple.

When we get to numbers in this case, we're talking about blood alcohol content and how you measure it. In the speeding example , a trained observer can usually make an accurate prediction about the speed of a large object. You can test his accuracy and he'll often be quite good.

But the human eyes, ears, and nose cannot measure the level of alcohol in the blood of human being; a device is required. If the device is a blood sample, there will be testimony about how the sample was collected, handled, and tested. A lab technician will testify, under oath, at trial how he tested the alcohol content and that it exceeded the level specified in the appropriate statute. He will be available for cross-examination.

This issue isn't about drinking and driving.

It's about how do you put a machine under oath, and how do you cross-examine it? I am guaranteed the right to confront any witnesses that testify against me at a trial by jury of my peers. Will those peers one day include a simple handheld device powered by a 9-volt battery?

Skipping the under-oath part, the cross-examination starts with looking at the device's internals and examining its source code.

Benji JohnsonSeptember 27, 2005 5:00 PM

I am 19 and just received Driving While Under a Unlaful Blood Alchol Level. I was in a parking lot moving my car from senior to regular student parking at UF. The cop came to harrass me about my tag being exspired wich it wasn't. He then did a brethalizer and I blew a .03. And my liscence was suspended for 6 months. I say if there is ANY way to fight back against these cops ruining peoples lives then more power to it.

AnjaApril 6, 2006 10:40 AM

"I say if there is ANY way to fight back against these cops ruining peoples lives then more power to it."

I thought this was interesting. The cop was just there to harrass. Hum...If you were not under the influence then why didn't you ask for a blood test. If you were under the influence then you did break the law and you should accept responsability for your actions. Granted the "test" may not be totally reliable and that needs to be remedied but society as a whole needs to accept responsibility for their actions (I would like to think that everybody is honest and tells the truth; but, you know as well as I do that is not the case & that does go for law enforcement as well).

If a case gets thrown out of court because of lack of disclosure of the inner workings of the machine then the cop did not do their job in documenting what caused him to be suspicious in the first place. In Texas alone it has been proven that no less than 2% of people on the road at any give time of the day are under the influence. Between the hours of 9PM to 4AM that number jumps to 47%. There does need to be a way to try and keep these people off the road to try to make it safer for the rest of us. Apparently a lot of people feel a need to break the law and don't care that their actions can not only endanger themselves but others as well.

If you are innocent then consent to all the tests available to prove your innocence beyond the shadow of a doubt until there is a fool proof system in place. I know, you shouldn't have to, but if it means a conviction vs. dropped charges which would you rather live with? If you are guilty, then suck it up and know that you got caught breaking the law and be a responsible upstanding citizen and accept responsibility for it.

Someone involved in the Florida CasesJuly 8, 2006 8:26 PM

I just wanted to make a comment in response to all of you who say that the reason the source code should be disclosed is to determine "how the machine works", and to be able to effectively cross examine the operator and/or maintenance inspector. I have had the dubious honor of sitting in court to hear many of the intoxilyzer motions being argued, both as a prosecutor and as a defense attorney, and I can tell you that the reason that the defense bar wanted (and I say wanted because the rulings by the county court judges have been recently overturned)the source code was because they knew that the manufacturer would not give it up, and they would be able to get rid of breath test results in hundreds of DUI cases because the State of Florida was not ABLE to provide the source code, not that they chose not too. The very same arguments were made by the very same defense attorneys about 18 months prior in regards to mantinence manuals and schematics (The cases were Muldowny and Pitts), and the Appellate Court ruled that the State must provide those documents to the defense (which they were able to do because they actually had them in their possession). However, after almost 2 years of fighting, once the manuals and schematics were handed over, not one motion was filed based on those "mysterious documents" from the defense bar. Why? The short answer is that there was nothing in those documents that would indicate there was a problem with the Intoxilyzer, and the Defense bar knew that there wouldn't be, but they still knocked out hundreds of cases in the meantime. Just like in the Source code issue. The Defendants don't want to know if the instrument is accurate or not, that's too dangerous a road to travel down. What if their high-priced expert says: Nope, sorry, it's working 100% accurately, your clients were over the legal limit. No, they don't want to risk that, they just want the cases to be thrown out. Just wait, in a few months, there will be a new "issue" with the intoxilyzer 8000 ( the new one in Florida), and all the same arguments will start again. Hopefully the judges in Florida will be able to weed through the crap and recognize if there are any real "issues" to rule on.

issieSeptember 11, 2006 8:56 AM

my husband is on federal prison in williamburg sc for 1994 dui 12 year this happen he serving 1 year in prison now they want to revoked is license for 1 year is not a felony charge it was on the navy base in pensacola the thing is why they send him to sc for his dui when he can do it here no accident or killing anyone now when he gets out he has to go with all enchilada dui classes also they want him to use an ignition interlock device installed upon all the vehicles they treat him like this his 3rd dui this only his 1 dui in 1994. i understand 6 month in jail but 1 year and so far from home when here the locals get dui's they out again in short time and also they get 1, 2 3, dui

jeepjeepAugust 6, 2007 5:14 PM

Q? is there any current feedback on the nabu 2000 test results/procedures as of july 2007?

gkOctober 30, 2008 10:49 PM

i don't think it matters why...the court will believe the breathalizer 'has done it's job anyway' -so i'm going through the process and i'll let ya know the outcome

TSAugust 24, 2010 9:56 AM

I am wondering how to get a court order to launch an audit and investigation of breathalyzers in my county?

Clive RobinsonAugust 24, 2010 3:16 PM

@ TS,

You would first have to say what country you are in.

Also you will have to accept the odds are very very much against you. From the little I know of it the companies try every trick and wrigle to avoid being held accountable and there is little or no will in the judiciary to understand the issues involved. Finally there is considerable political will to stop such attempts for a number of reasons which basicaly boil down to revenue and votes.

One trick you will come up against is that various press people will interview victims relatives about the loss of their loved ones. Worse they will almost certainly find some non photogenic low life scum bag with little or nothing but a sense of grevence about how they lost their licence for having flatened a baby in a pram whilst either driving at excessive speed, largered up or both. And these same press people will do their best to transfer the justified villification of the scum bag onto you.

You basicaly lose to the "for the children" mentality that pervades politics these days and there is little or nothing you can do about it.

Leave a comment

Allowed HTML: <a href="URL"> • <em> <cite> <i> • <strong> <b> • <sub> <sup> • <ul> <ol> <li> • <blockquote> <pre>

Photo of Bruce Schneier by Per Ervland.

Schneier on Security is a personal website. Opinions expressed are not necessarily those of Co3 Systems, Inc..