The plaintiff had up to 10/28 to rebut the report. Instead, Malibu fired its own motion for summary judgement. Among numerous exhibits to this motion, a couple can be counted as a rebuttal attempt. A very interesting transcript of Tom Parker’s deposition deserves a separate analysis, and I’ll get back to it in the coming days. Today’s post is about another document: a very strange “Report on the probability that two Americans live next to each other and both use Bittorent and the Bittorent client Transmission,” prepared by Dr. Dan Sarel who is an associate professor of marketing in the School of Business Administration at the University of Miami.
This report’s aim was to rebut Tom Parker’s suggestion that accessing wi-fi by a neighbor was possible and plausible. Dr. Sarel’s argument was largely based on the fact that the Germans determined that the alleged infringer had used a relatively rare Bittorent client, Transmission — the very same client the defendant admitted to own. The expert argued that such a coincidence is unlikely:
Subject to the assumptions and limitations set forth below, I conclude the probability that two Americans live next to each and both use BitTorrent and the BitTorrent client “Transmission” is 1 out of 37,679.
Dr. Sarel was paid a $2,500 fee for what, in my opinion, is a half-assed argument at best. Read the report and scroll down: I’ll try to explain why I think it is misleading and doesn’t rebut the defendant expert’s report at all.
Buzzwords used incorrectly
Dr. Sarel begins his report with
The method for calculating the probability employed in this report is based on a well-accepted methodology that has been used and admitted into evidence by courts for many years. The method often called the “Multiple Dependent Event Probability Determination Analysis.”
I dare to say the “often called” is a slight overstatement because searching Google for this phrase yields exactly zero results (excluding this post).
Of more concern is conflating “dependent events” and the multiplication formula, which is applicable only to calculating overall probability of an independent event combination. The probability of a dependent event combination should be calculated using the Bayes’ Theorem.
Given that the author is a tenured professor who undoubtedly knows statistics and the probability theory better than yours truly, I assume it was a typo.
Unnecessary paragraph that proves nothing
Paragraph A that estimates the probability of two “heavy pirates” leaving next to each other is, while technically correct (assuming that the input numbers are accurate), proves nothing and borders on making a legal conclusion — by implying that both the defendant and his mysterious wi-fi piggybacker are both “heavy pirates.”
A wrong assumption
The entire report is based on the assumption that the distribution of tech-savvy digital movie owners is uniform across the US, and the the same 14% is a correct estimation everywhere — be it rural Missouri or Silicon Valley (this assumption is not listed in the “assumptions and limitations” section). I don’t know what they teach at the U of Miami School of Business’ marketing program, but a person with a functioning brain immediately sees how wrong this assumption is. For example, the probability of a US household having a yearly income of $380,000 or more is 1%, so, I will calculate the probability of two wealthy people leaving next to each other: it is 0.01 * 0.01 = 0.0001, or 1 in 10,000. Then I will picture you driving along the Malibu coast and [rightfully] calling me an idiot.
A wrong problem
Dr. Sarel is solving a wrong problem here. The question is not what a probability of two Transmission users leaving next to each other is. The defendant admitted that he is a Transmission user, so the probability of him using Transmission is, unsurprisingly, 100%. The expert should have been tasked to determine what the likelihood of another Transmission user living next door is. Even with the [incorrect] numbers Dr. Sarel uses on p. 7, it will be 0.5%, not 0.5%2 . If we roll two dices, the a priori probability of both 6s is 1/6 * 1/6 = 1/36, yet if we already have 6 after the first roll, the probability of the second 6 is 1/6.
More than one neighbor
The problem being solved is also wrong because it only assumes a single neighbor. The correct way to formulate the task is: what is the likelihood that at least one neighbor out of N uses Transmission. The independent events here are “Neighbor X does not use Transmission”:
P(1 neighbor: no Transmission) = 1 – P(1 neighbor uses Transmission)
P(N neighbors: no Transmission) = (1 – P(1 neighbor uses Transmission))N
P(at least 1 neighbor of N uses Transmission)= 1 - (1 – P(1 neighbor: no Transmission))N
The defendant alleges that he lives in a unit with close proximity of other households (Defendant’s motion for summary Judgment, p. 11). Moreover,
While he was in his kitchen and living room, and upstairs (and not next to a window or open door), Defendant viewed the network preferences of his computer and noticed that at least 12 wifi signals other than his own were visible. These signals had not been generated from within Defendant’s unit and came from neighbors.
Given that at least 12 units are located inside the 50 meter radius from the defendant’ unit, seeing that many wireless connections is not surprising. 50 meters is a minimal range of a typical wi-fi router. Judge Wright in the infamous Ingenuity 13 v John Doe case (CACD 12-cv-08333) used the same analysis and concluded that in a suburban environment, wi-fi connection could be used by dozens of neighbors.
Below are my calculation results for different numbers of neighbors (rows), and different percentages of households that have movies/TV shows in a digital form (columns; starting with the US average of 14%). Red font is for the numbers higher than probability to be killed playing a single round of my ancestors’ favorite sport (16.7%).
Probabilities of at last one neighbor using the Transmission client
Misusing statistics in courts is nothing new. 15 years ago a flawed “expert report” resulted in a wrong incarceration:
Sally Clark (August 1964 – 15 March 2007) was a British solicitor who, in November 1999, became the victim of a miscarriage of justice when she was found guilty of the murder of two of her sons. Although the conviction was overturned and she was freed from prison in 2003, the experience caused her to develop serious psychiatric problems and she died in her home in March 2007 from alcohol poisoning.
Clark’s first son died suddenly within a few weeks of his birth in September 1996, and in December 1998 her second died in a similar manner. A month later, she was arrested and subsequently tried for the murder of both children. The prosecution case relied on significantly flawed statistical evidence presented by paediatrician Professor Sir Roy Meadow, who testified that the chance of two children from an affluent family suffering sudden infant death syndrome was 1 in 73 million. He had arrived at this figure erroneously by squaring 1 in 8500, as being the likelihood of a cot death in similar circumstances. The Royal Statistical Society later issued a statement arguing that there was “no statistical basis” for Meadow’s claim, and expressing its concern at the “misuse of statistics in the courts”.
Of course, our case is not that dramatic; yet don’t forget that juries in civil cases use a very low standard (preponderance of evidence), and can be easily swayed over the 50% threshold by a misleading report, especially given that historically expert witnesses are often being taken at their word without any healthy scrutiny.
So, Mr. Lipscomb, when can I collect my $2,500 paycheck?
Thanks to Raul for finding this gem.
At paragraph 7 of his declaration, Dr. Sarel touts having given testimony in several lawsuits including Innovation Ventures, LLC v. N2G Gistributing, Inc., et al. (MIED 08-CV-10983), in which four years ago Judge Borman granted the defendants’ motion in limine to exclude expert witness Sarel:
The Court finds that Dr. Sarel’s survey suffers from multiple flaws which resulted in an unreliable scientific foundation for its conclusions. […] The Court further finds that the reliability of this survey is so questionable that it is not appropriate material to assist the jury in deciding the issues in this case. Defendants’ motion is therefore granted.
I planned a couple of updates/followups, but the majority of them are moot now: this case settled on 11/6/2015. I bet Lipscomb celebrates — not the cash he was able to wrestle from the defendant (if any), but rather the fact that he dodged a jury trial.
Still may write about the defendant’s expert deposition: it has a couple of interesting tidbits.