A six-year battle in Minnesota regarding the reliability of breath test results in impaired driving cases in light of alleged defects in the testing instrument’s source code ended last summer. The State mostly won, though the Minnesota Supreme Court determined that machine-generated reports of a deficient breath sample were unreliable absent corroborating evidence. The case is In re Source Code Evidentiary Hearings in Implied Consent Matters, 816 N.W.2d 525 (Minn. 2012), and here is the backstory:
In 2006, a driver charged with DWI in Minnesota filed a motion for discovery of the source code for the instrument used to test his breath, the Intoxilyzer 5000EN. (The breath testing instrument currently used in North Carolina is the Intoximeter, Model Intox EC/IR II.) After his motion was granted and the state’s challenges were rejected on appeal, discovery of Intoxilyzer source code became standard litigation strategy in criminal DWI and implied consent proceedings in Minnesota. (Discovery of source code has not been so liberally granted in other jurisdictions, as Jeff noted in this earlier post.) In light of the numerous discovery requests, Minnesota sought to have assigned to a single judge or panel of judges all pending and future implied consent cases, both civil and criminal, challenging the reliability of the Intoxilyzer 5000EN based on allegedly defective source code. The Minnesota supreme court assigned the cases to a single district court judge.
Obtaining the source code was, of course, only the first step toward challenging the breath tests. Defendants and other petitioners (hereinafter “petititioners”) retained experts to study the source code. Those experts identified several deficiencies. Chief among them were criticisms that (1) the instrument did not conduct self-testing to determine if, for example, it was unplugged or if there was a change in voltage, (2) the instrument did not adequately detect radio frequency interference from cell phones, (3) the instrument’s failure to report the margin of error rendered the results unreliable, and (4) the range for breath volume required for an accurate reading was too narrow, resulting in reports of a “deficient sample” for breath samples that were, in fact, adequate.
Petitioners moved to exclude all test results in their trials or hearings, arguing that the defective code affected the reliability of both the instruments that reported a breath test result and instruments that reported a deficient breath sample. In the alternative, the petitioners moved for a pretrial order allowing them to present expert testimony in their individual trials or hearings that the alleged defects in the source code affected the precision of the test results.
After a two-week long evidentiary hearing, the district court denied both requests. The court determined that the alleged defects did not affect the reliability of numerical breath test results. The court determined that the self-tests were not necessary because the machine was operated by a person, the possibility of radio frequency interference caused by cell phones was remote, and that any requirement for reporting of a margin of error with every breath test result was beyond the scope of the issue before the court and was instead in the realm of policy decision making. As to machine-generated reports of deficient samples, however, the court determined that shortcomings in the source code did affect the reliability of these reports. Thus, the court ruled such results inadmissible absent other evidence that supported the sample being deficient.
The petitioners appealed, arguing, among other things, that the trial court’s ruling precluding them from introducing evidence of problems with the source code violated their right to due process. A majority of the supreme court disagreed on the grounds that the petitioners had been afforded a hearing on the issue and the source code defects the petitioners had identified were not relevant to the validity of the test results. Indeed, the petitioners’ primary expert had concluded the Intoxilyzer 5000EN provided valid breath alcohol measurements, a fact that, in the words of the district court, cast “‘a large shadow over the hearing.’” Three justices dissented, expressing the view that the source code would be probative of accuracy in certain cases and that the rulings deprived defendants of the opportunity to cross-examine officers about the reasons why certain procedures must be followed in administering a test, such as ensuring that a cell phone is kept sufficiently far from a testing device.
In North Carolina, a chemical analysis of a person’s breath administered pursuant to the implied-c0nsent law is, by statute, admissible in any court or administrative proceeding. See G.S. 20-139.1(b). I’m sure defendants regularly challenge the reliability of those results. I don’t know whether any have successfully obtained the source code or mounted a challenge on that ground. My sense is that while the battle in Minnesota may be over, the North Carolina skirmish has not yet begun in earnest. And, to complete the metaphor, I suspect the war over the accuracy of chemical testing will rage on.