Breathalyzer Source Code

Once again, my interest in criminal law and my interest in technology have come together in a brewing legal controversy.  The issue is whether a DWI defendant who has submitted to a breath test for alcohol has a right to access the “source code” of the breath analysis machine.  The Minnesota Supreme Court, in State v. Underdahl, available here, just ordered that a particular defendant be granted such access.

Let’s start with some background.  Breath analysis machines have been around for many years.  There are actually several different companies that make such instruments.  Most of the reported cases I’ve seen involve the Intoxilyzer 5000, manufactured by CMI, Inc., though my understanding is that the machine currently in use in North Carolina is the Intox EC/IR II, manufactured by Intoximeters, Inc.  I believe that both machines work by shooting infrared rays through a breath sample and looking for the distinctive refraction caused by alcohol in the breath.  (Readers who are versed in DWI, correct me if I’m wrong.) [Update: a knowledgeable reader tells me that the Intoxilyzer 5000 works by shooting infrared rays through a breath sample and looking for absorption, not refraction, while the Intox EC/IR II measures alcohol using a “fuel cell.”]

Some defendants who have submitted to breath analysis testing and who have received readings in excess of the legal limit have argued that the machines are defective, and specifically, that the source code, or programming, of the machine is flawed.  (This is similar to the arguments some have made about electronic voting machines, as discussed here.)  It appears that the complaints about breath test machines revolve around the possibility of false positives when certain non-alcohol compounds are present in a breath sample.  The source code tells the instuments what readings to “count” as alcohol and what not to count, and the defendants argue that they need to know what the instruments are counting in order to determine whether the instruments’ readings were likely to have been accurate in their cases.

Thus, these defendants have asked to examine the source code.  Most states have argued (1) that there’s no right to examine the source code absent some preliminary showing that the source code is likely to be flawed, and (2) that the states don’t have access to the source code, which belongs to the machines’ manufacturer(s).  Most courts have accepted one or the other of these arguments. See, e.g., State v. Bastos, 985 So.2d 37 (Fla. Ct. App. 2008) (“[W]e cannot accept the proposition that simply because a piece of testing equipment is used in a criminal case, it follows that the source code for its computer must be turned over. There would need to be a particularized showing demonstrating that observed discrepancies in the operation of the machine necessitate access to the source code.”); Hills v. State, 663 S.E.2d 265 (Ga. Ct. App. 2008) (finding that the source code was not in the state’s possession and was therefore not discoverable).

In Underdahl, the Minnesota Supreme Court actually considered two cases.  In one, the defendant requested access to the source code without making any preliminary showing, and the court ruled that his request should have been denied.  In the other, the defendant submitted a number of exhibits in support of his request for the source code, including one from a computer science professor at UC Berkeley (my alma mater!) who “explained the source code in voting machines, the source code’s importance in finding defects and problems in those machines, and the issues surrounding the source code’s disclosure.”  The court ruled that this showing was sufficient, that the state owned or had control of the source code, and that the defendant was entitled to review the source code.  An interesting news story about the case is here.

What’s the scoop in North Carolina, you ask?  Well, we don’t have a reported appellate case on this issue, and it doesn’t seem to be coming up very much at the trial level.  (Again, readers, tell me if I’m wrong.)  Perhaps that’s because of G.S. 20-139.1, which provides for automatic admissibility of breath analyses conducted in keeping with rules promulgated by the Department of Health and Human Services — defense attorneys may have concluded that there’s no point to requesting source code if there’s no way to keep the test results out of evidence.  Or perhaps it’s because such requests had, prior to Underdahl, mostly failed in other states.  Regardless, Underdahl may prompt a re-examination of the issue, and I wouldn’t be surprised to see both sides gearing up on this subject in the months to come.

One comment on “Breathalyzer Source Code

  1. […] returning to a topic I blogged about here, a New Jersey case has resulted in two analyses of breathalyzer source code being conducted — […]