The Tiny Detail that Is Still Bothering Us About 'Serial'
The show glosses over the unreliability of cell-tower data during a key moment in episode five.
Image: David Schott/Flickr
Warning: This post contains light spoilers.
I'm not one of those Serial listeners who was disappointed by the ending. Serial was always less of a straight-up whodunnit than it was just great investigative reporting. But there was one aspect of the final episode that struck me. What about that Washington Post article mentioned in episode five, "Route Talk"?
The conviction of Adnan Syed, Serial's central character, came down to evidence corroborated by additional evidence gathered from cell phone towers, and it was during a key moment in "Route Talk"—host Sarah Koenig citing the above Washington Post article—that the scientific accuracy of cell-tower evidence was first brought into question.
In the episode, Koenig mentions the article's headline, "Experts say law enforcement's use of cellphone records can be inaccurate," but then doesn't quite unpack the piece itself or call the experts it cites. Instead, she contacts outside experts, none of whom are among the sources from that Washington Post article.
It's a nitpick, sure, but why mention the article at all without digging into it, or calling experts directly from the piece itself?
During the podcast's finale, even in Koenig's last fell swoop to map out the day's events, the cell-tower evidence continues to be taken at face value, with no reference to lingering questions posed earlier about tower accuracy.
'Serial' dug into the minutiae of every detail, and that's what made this oversight stick out
This is the sole aspect of the finale that caught me by surprise, not the series' ambiguous conclusion. Serial's 12-hour, no-stones-left-unturned runtime dug into the minutiae of every last detail, and that's exactly what makes this small detail stick out.
So, I decided to call up Edward J. Imwinkelried myself, a law professor at the University of California, Davis, expert on the admissibility of cell-tower evidence, and, most importantly, one of the sources from the very Washington Post article referenced in episode five, "Route Talk."
I cold-called him and left a voicemail. He called me back an hour later:
MOTHERBOARD: Hey, Professor Imwinkelried. The reason I'm calling you is because I saw your name cited as a source in a Washington Post article written by Tom Jackman this year, and in turn that article was referenced in Serial. Have you heard of Serial? The podcast?
Imwinkelried: No, I'm afraid I haven't. No.
Oh OK, never mind, I'll just jump into it: Is the use of evidence gathered from cell-phone towers reliable and accurate when trying to pinpoint a person's location, say, in a murder case?
The important thing to realize is that cell phone technology was not designed as a technology for locating people. It's a communications technology. It's one thing when using a technology specifically designed for locating—GPS—it's another thing when you try to adapt a technology which was never designed for that purpose.
GPS is the technology to locate people. The problem is, in the past, many of the cases have involved cell phones without GPS, and there they've been relying on these supposed assumptions that the phone connects to the closest tower, or the tower with the strongest signal.
So they're not reliable at all? Location evidence gathered using cell-phone towers isn't good science?
They were never intended to serve that function.
The decision as to which tower to connect to isn't made by the cell tower, it isn't made by the phone, it's made by the network computers. And what are the network computers interested in? Balancing the load, using all the towers in the network.
And that's why you can sit in your room in a 10-minute period, make three cell phone calls, and connect to three different towers. You haven't moved at all, but you've connected to three different towers.
GPS in smartphones notwithstanding, have cell phone towers themselves changed much since, say, 1999? They're just as unreliable today as they were 15 years ago when used to pinpoint a person's exact location?
That's right. I mean, when you're trying to determine a person's location based upon call records—they were never intended to serve that function. They're intended to give customer satisfaction—to virtually all the time connect you. And in order to do that, the network computers have to balance the load.
What was the precedent for courts accepting this evidence as being admissible in court? What's the status of its admissibility today?
They've been accepting cell phone evidence for a long time, well over a decade. It's only recently that they've begun to question it.
One of the most important cases was a case decided by the Northern District of Illinois in Federal District Court in 2012—United States of America vs. Antonio Evans. It was this case that, perhaps for the first time, the courts really questioned the simplistic assumptions witnesses used in the past.
So cell-tower evidence is still admissible today?
Even today a prosecutor would be inclined to use cell-tower evidence, but now there is a body of law questioning the use of cell-tower evidence.
And the other thing I've seen is that the prosecution witnesses are becoming much more circumscribed in the testimony they give. They won't say it always connects to the closest tower, or always connects to the strongest tower—they'll use generalization, but even those generalizations are suspect. Because, again, of the need to balance the load, a need fulfilled by the network computers.
Thanks for taking my call, I think that covers it.
OK, have a nice afternoon.
So, according to this expert, at least, the idea that cell phones connect to the closest or strongest tower is, if not an outright myth, an inconsistent data point. And admissibility of this evidence for determining someone's location now has a body of law that disputes it.
This taken alone doesn't solve the tower riddle, no doubt, nor provide the sort of closure some Serial fans feel they didn't get. We only spoke for about 15 minutes, after all. But it does beg the question: How different would details of season one have looked if this particular expert were called instead of the two engineering professors from episode five?