In New Jersey, a conviction for drunk driving can lead to serious consequences—loss of one's driving license, a significant fine, and even jail time. Since such a conviction can be adjudicated based on data produced by a machine standing alone, it is imperative that machine-generated data be reliable.

The machine in question is the Alcotest, used in New Jersey since 2000. In 2016, the person within the State Police who was responsible for performing semiannual calibrations of the Alcotest instrument was indicted for falsely certifying that he had performed the required periodic calibration procedures when in fact he had not done so on machines used in five New Jersey counties: Middlesex, Ocean, Somerset, Monmouth and Union. Specifically, he had not used a thermometer that meets the standards set by the National Institute of Standards and Technology (NIST) to measure the temperature of simulator solutions used to calibrate the Alcotest devices.

The number of individuals whose breath samples had been obtained using these faultily calibrated instruments was large—20,667. When the police official was indicted, New Jersey's Attorney General's Office notified the Administrative Office of the Courts about the problem.

On Nov. 13, 2018, the New Jersey Supreme Court decided a case filed by one of these 20,667 persons, who was seeking to withdraw a guilty plea she had decided to enter based solely on her Alcotest reading. State of New Jersey v. Eileen Cassidy. Although the movant had since died, the court held that the case was not moot because it involved “an issue of significant public importance [that] is likely to recur.” It affirmed and adopted the findings and 198-page report of a special master, retired judge, Joseph F. Lisa, whom the court praised “for his diligence and insightfulness so evident in his extensive and thorough report.”

While a summary cannot do justice to the special master's detailed, careful and lengthy analysis of the evidence, including his credibility findings of the experts who testified, suffice it to say that the temperature of the solution used for calibration is key to the accuracy of the instrument's readings and must be at the generally accepted temperature for human breath, creating a vapor that, the court said, “is a proxy for human breath.” In short, accurate temperature readings of the simulator solutions are “critical to the accuracy of the Alcotest.” As the Supreme Court held, failure to use a thermometer that produced NIST-traceable temperature readings for calibration “undermines the reliability of the Alcotest.” Use of such thermometer is especially important because the NIST-traceable thermometer was the only temperature measuring device used in the calibration procedure that was not manufactured by the company that made the machines themselves. Thus, if there were bias or error in the manufacturer's laboratory, that could affect the calibration and be undetected if no independently manufactured thermometer were used.

Since the Alcotest machines in the five counties had been improperly calibrated, the Supreme Court (a) ordered the state to notify all affected defendants that the Alcotest results in their cases are inadmissible in evidence so that they can “take appropriate action;” (b) relaxed the five-year time bar for post-conviction relief under Rule 7:10-2(b)(2) in the interests of justice; and (c) vacated defendant Cassidy's drunk driving conviction.

The careful scientific analysis undertaken by the special master and adopted by the court is especially important in this state where this machine-generated data, standing alone, effectively amounts to a per se conviction because a machine cannot be cross-examined. It also reminds us of the importance of science and scientific analysis in today's world where scientific evidence is too often ignored.