On Being Skeptical of 'Textalyzer' Technology to Detect Smartphone Use Before Accidents

There's a bill in the works in New York State that would allow police to examine smartphones at the scene of an accident for "signs of use" through so-called "Textalyzer" technology developed by Israeli forensics firm Cellebrite. The idea is to identity people who cause accidents because they were texting and otherwise using their mobile devices.

Police asking for papers

"Your papers and smartphone, please."

To say that I'm awash in a sea of mixed emotions is an understatement, but there's almost nothing I like about what we know so far. My biggest concern is that the makers of the Textalyzer technology itself—whatever the heck it is—are backing the bill, and the company is draping its initial promotion in the shroud of pain from the families of victims killed by idiots texting and driving.

That's a dangerous combo, in my opinion, the kind of thing I am always leery about.

The problem at hand is texting and driving or otherwise using our iPhones and Android devices while driving. Maps, browsing, reading, gaming, looking up a phone number—when using your hands, these are all things that take your attention from the road, and that leads to accidents and people being killed.

Distracted Operators Risk Casualties (DORCS) is a victims' group seeking to raise awareness and affect policy changes of the risks of distracted driving. DORCS cofounder Ben Lieberman lost his son to a person texting and driving, and he found that out after subpoenaing phone records of the driver through a civil suit he filed. Those records showed the driver was texting in the moments before the accident.

"The general public knows distracted driving is a problem, but if people knew the extent of the damage caused by this behavior, they would be amazed," Mr. Lieberman said in Cellebrite's press release [via Patently Apple]. "With our current laws, we're not getting accurate information because the issue is not being addressed at the heart of the problem—with the people causing the collisions."

Enter Textalyzer. We don't know much about this technology, but Cellebrite said, "A key part of the legislation involves new 'Textalyzer' technology that will allow officers to detect whether or not the device was being used around the time of a crash, but will not provide access to any content—keeping conversations, contacts, numbers, photos, and application data private."

That wording sends up all kinds of red flags to me. "Whether or not the device was being used around the time of a crash" is shockingly vague. Used how? Hands free? Background processes? Automated responses whose purpose is to actually keep you from using your device? Being used by malware without the device owner's knowledge?

Also, used by whom? By the driver? By a passenger? In the case of a driver without a passenger, that might be clear, but show me the technology that can identify who was tapping a screen 30 minutes ago and I suspect the late Arthur C. Clarke might be willing to label it magic.

Furthermore, does this technology work when used with X iteration of hardware or Y iteration of iOS or Android? How about X iteration of hardware when combined with Y iteration of software? The number of combination possible in the Android world is effectively limitless, and there's no way all combinations can be tested.

Does Textalyzer itself need to be updated? If so, by whom?

Even if the heart of Textalyzer is truly foolproof, what factors could generate a false positive? How would a cracked screen affect readings? Water damage on the targeted device? Internal drop damage? Very hot or very cold weather?

Another area of concern for me is that this technology would be used by the police at the scene of an accident, and not by the forensics experts at Cellebrite itself. I'm beyond skeptical of the potential results from such usage.

The irony to me is that Ben Lieberman's own crusade was sparked by information he learned through a court subpoena, not by magical technology working who-knows-how. That is a far more appropriate way of determining device-use related to an accident than what we know about Textalyzer so far.

Dave Hamilton noted in Tuesday's Daily Observations that telephone records wouldn't reveal app usage (aside from general data transmission), but that doesn't necessarily make Textalyzer the answer.

The problem of distracted driving is real. Stopping it is not only laudable, it's vital. But I'm not willing to sign off on either New York's legislation—not that anyone is asking me to—or Textalyzer itself.