And, with enough legal prowess — and financial resources — defendants can line up parades of experts to try to undermine a prosecution witness’s interpretation of forensic data, from the timing of a Google search to the movement of a human body.

“As technology advances at such a rapid pace, the things that we used to think were black and white aren’t black and white anymore,” said Christina Miller, a professor at Suffolk Law who previously focused on cases that involved digital forensics as a Suffolk County Prosecutor.

She noted two recent Massachusetts Supreme Judicial Court rulings that each upheld decisions to disallow certain data from being used as evidence because of questions about their accuracy: In 2021, the courts disallowed the calculation of a defendant’s speed by a GPS device, and earlier this year , the courts prevented evidence of a defendant’s cellphone location history from being introduced in a criminal trial.

In the latter case, the analysts for the prosecution had used a different version of an iPhone’s operating system as they sought to replicate the data. That underlined one of Miller’s main points: “The forensic examiner is only as good as the tools they use, and the tools are only as good as the data.”

Expect to see more court challenges, she said.

Michael Kendall, a former federal prosecutor who’s now a defense attorney, added that judges have to be “much more demanding” in determining the validity and credibility of someone claiming to be an expert — as well as what science and processes are rigorous enough to constitute presentable evidence.

“There has been so much phony scientific evidence that has railroaded people over the years,” he said. “There has to be some validation of the expert. The court needs to police the quality of the experts and the quality of the science.”

The reliability of certain digital forensic data varies with the nature of the technology at issue. Programs were developed to complete specific functions, not, for example, to serve as an official time-stamped record of events that could constitute irrefutable evidence, said Seth P. Berman, a defense attorney and former prosecutor. So, while emails, Google searches, or phone calls may include a time stamp, that doesn’t mean the time stamp itself is accurate.

“This entire field of computer forensics is essentially an accident,” said Berman, who leads the privacy and data security practice group at law firm Nutter and previously worked for a firm that specialized in digital forensics. “Nobody created computers with the goal of using them to create evidence.”

So, he added, “As a result, the data is not that clear. There are a bunch of things that just go wrong,” and can lead to different expert interpretations.

Take, for example, the Read case. She was charged with backing into her boyfriend, Boston Police Officer John O’Keefe, with her Lexus SUV after a night of heavy drinking in 2022 and leaving him to die outside the Canton home of a fellow Boston Police officer during a blizzard. Her defense team claims she is being framed, and that O’Keefe was actually beaten by people who had been attending a gathering inside the home, and then dumped outside. Read found O’Keefe’s body hours later in a snowbank, after returning to look for him.

The case ended in a mistrial in July, and a new trial is slated for January.

At the core of the defense’s theory is the timing of a Google search for “hos (sic) long to die in cold” by Jen McCabe, a woman who was at the gathering inside the Canton home. According to an expert hired by the defense, data shows she Googled the inquiry on her phone at 2:27 am, hours before O’Keefe’s body was found. Many among the crowds of Read supporters who gathered regularly outside the courthouse cited the testimony as a crucial indicator of her innocence.

But prosecution experts said the testimony was wrong, and that the search occurred after Read and McCabe found O’Keefe’s body shortly after 6 a.m. The discrepancy, prosecution witnesses said, stemmed from confusion around what the time stamp was referring to; they said the 2:27 am stamp simply referred to when the web page that was later used for the search was first opened.

There were similarly differing claims over other evidence: calls that were deleted from the phone, or not; how fast Read’s car accelerated while in reverse; O’Keefe’s movements, based on data from his phone and watch.

Read’s team of lawyers mounted an aggressive defense, sharply cross-examining most of the government’s witnesses and also producing some of their own.

A judge declared a mistrial after the jury reported it was deadlocked and could not reach a verdict. Read maintains her innocence.

Berman noted that most defendants don’t have the financial means of Read, a financial analyst and adjunct professor who also benefited from the donations of ardent supporters. A defense effort that had less time, labor force, and money probably would not have been able to push back so forcefully on inconsistencies in the data, he said.

Ultimately, judges are the arbiters in determining the credibility of an expert witness or the validity of a science, guided by appeals court decisions, including precedents set by the US Supreme Court. The goal is to “winnow science from junk science,” said Rosanna Cavallaro, a Suffolk Law professor who teaches about evidence. But that can be difficult, she said, as new technologies and expertise in those technologies evolve.

Cavallaro also said a “battle of experts” can be detrimental to a case when the process devolves into each side simply hunting for the most favorable expert they can find — someone who not only will come to the conclusion they seek, but who will communicate it engagingly and effectively.

At times, she said, “you do become concerned that the person’s opinion is up for sale. The problem has been pervasive across the sciences.”

---

Sean Cotter can be reached at [email protected]. Follow him @cotterreporter.