The machine pictured above (often referred to as “an instrument” by police in lab coats) is used by police to test the amount of alcohol present in a drunk driving suspect’s blood. How do you test blood alcohol taking a breath sample? Well, this machine attempts to do this by taking a sample of a subject’s deep lung breath, “analyzing” the sample, and then applying some assumptions and general formulas that purport to translate the amount of alcohol found in the breath, to an equivalent blood alcohol ratio. The key words in that previous sentence are “some assumptions” and “general formulas”. Plenty of both are employed to derive a result in a specific case, and, as is always the case in any such generalization, Your Mileage May Vary.

For example, the machine assumes a constant 2100:1 ratio of units of alcohol in the blood to units of alcohol in the breath. This assumption is usually referred to as the “blood/breath partition level”, and is based on the average partition level found in most persons. That’s right, it’s an average. That means that for some people the actual ratio is 1700:1 and for some people it’s 3500:1. But that doesn’t matter to the machine, because anyone who blows into it is going to be assumed to have a 2100:1 ratio, with no account taken whatsoever for the individual variances known to exist. This means that you might actually have a blood alcohol level of .06% (below the legal limit) and still blow a .10% (over the legal limit) because your body chemistry is not “average”. Sorry about that.

This issue’s been litigated to death, and courts under the sway of Madd Mothers realized long ago that the politics of DUI don’t permit such fine distinctions to be made. Your objection is noted, Counselor, but the Court bases its decision on the DUI Exception to the Constitution. Next case.

Well, ok. Putting aside the issue of how valid it is to apply quasi-scientific assumptions and formulas to derive a reading for a particular test subject, how does this machine actually apply these assumptions in the first place? If I lift the hood to watch what’s going on inside the machine as the breath travels through it, will I see 2100 units over here, and 1 unit over there? Not exactly. You see, this machine is like any other computer: it has to be programmed to do what it does. And that programming is done with software code. The code is low level machine code, programmed into a proprietary EPROM (Erasable Programmable Read Only Memory) chip and plugged into a circuit board on the machine. Somewhere deep inside that proprietary code the instructions, we are told, use the 2100:1 ratio to come up with their result. And we all know that software is never wrong, and never has bugs, right?

No problem. In a criminal case, the defendant has a right to confront the evidence used against him and may compel discovery in order to verify that things are in fact what they seem. Right? So, let’s open up that machine and let me take a look at that EPROM to see what the code is actually doing, so that I can verify not only that it’s applying the proper ratios and assumptions in the first place, but also that it’s not doing anything else fishy. Unfortuanately, that’s not possible in this case because the manufacturers of these machines aren’t willing to open up the kimono and reveal their proprietary source code. It’s trade secret, you see.

This leaves judges, juries and lawyers in the position of having to trust the machine manufacturer’s word that the code really is doing what they say it’s doing. In other words, trust us, the software has no bugs.

This is NUTS and completely unacceptable in virtually every other context, but here, once again, the DUI Exception to the Constitution applies, so no, you can’t have the source code. Trust us, you’re guilty.

Not so fast, said a Florida court this week. On November 2, 2005, three courageous judges in Sarasota County, Florida ORDERED the manufacturer of the Intoxylizer 5000 (pictured above) to turn over the EPROM source code to an expert hired by the DUI defense lawyers. The Court acknowledged that the code was indeed a trade secret, but held that it would be contrary to law to allow a trade secret privilege to trump the defendant’s right to confront the evidence being used to convict him. According to the Order, the manufacturer has 15 days to comply. Calls to the manufacturer were not returned and the company took a “no comment” position when asked if it would comply. The Court’s order specifically holds out criminal contempt as a potential sanction for violation, so an interesting showdown looms ahead and I look forward to reporting the result here.

The implications for the defense bar are enormous. Even though this case is only controlling in the applicable Florida jurisdiction, now that there’s actual appellate legal precdent on the point, lawyers everywhere will cite this case for its precedential and persuasive value, and the issue is bound to come up in more jurisdictions. And based on the usual record of showdowns between the rights of criminal defendants and various asserted privileges (just ask Judith Miller of the New York Times), I’m betting on the criminal defendant.

(With thanks to the venerable Lawrence Taylor for first discovering the “DUI Exception to the Constitution”).

RP

This entry was posted on Saturday, November 5th, 2005 at 11:28 pm and is filed under Software, News, Junk Science, DUI. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.