Opinion
Civil Action No. 11-235-RGA
01-28-2014
Robocast, Inc., Plaintiff, v. Apple Inc., Defendant.
MEMORANDUM ORDER
Before the Court is Defendant Apple Inc.'s Motion to Exclude the Testimony of Robert J. Sherwood and Creighton G. Hoffman (D.I. 341) and related briefing. (D.I. 342, 377, 410). The Court heard oral argument regarding this motion on January 8, 2014. Mr. Sherwood is Robocast's expert on the subject of internet advertising and Mr. Hoffman is Robocast's damages expert. Apple contends that Mr. Sherwood's testimony should be excluded because it fails to meet the fit and reliability requirements for expert testimony. Additionally, because Mr. Sherwood's analysis forms the foundation for Mr. Hoffman's opinions, Apple contends that Mr. Hoffman's opinions have no basis.,
Apple also argues that Mr. Hoffman's opinion impermissibly applies a 50/50 profit split and violates the Entire Market Value rule. Because I conclude that Mr. Sherwood's report uses unreliable data, I do not reach these issues.
Apple also argues that Mr. Hoffman's Safari analysis must be excluded because the foundational Berger report has been withdrawn. Robocast does not contest this issue. Therefore I will grant Apple's request.
In his report, Mr. Sherwood concluded that Automated Browsing System ("ABS") advertising increases Click Through Rates ("CTR") an average of 33% over static advertising. (D.I. 343-1 at 129). Mr. Sherwood bases this conclusion on eleven internet sources (D.I. 343-1 at 129 n. 2), which he described as "credible and representative of an advertiser's expectations." (D.I. 343-1 at 136). At deposition, Mr. Sherwood could not identify sources for some data points and admitted that he mistakenly counted others multiple times. (D.I. 342 at 6). Apple contends that the remaining five sources are unreliable because they were either self-interested sales pitches or misread by Mr. Sherwood.
Robocast has not attempted to rebut these characterizations. I therefore assume that Mr. Sherwood can only base his opinion on the five sources that did not suffer from those defects.
The first source is a webpage maintained by Morf Media, a "gamification" company that creates game based internet advertising. The website stated that gamification results in a 68% rise in CTR, yet Mr. Sherwood did not know anything about the underlying data or methodology. (D.I. 343-1 at 248). The second source is a press release from an online advertising company which claims that its ads "drove 21% click-through rates," yet Mr. Sherwood mischaracterized it as increasing click through rates by 21%. (D.I. 343-1 at 249). The third source is a webpage by TLV Media that claims slider ads typically generate 42% more eCPM than standard ads, yet Mr. Sherwood mischaracterized it as increasing CTR by 42%, something he admitted would only occur at constant CPM. (D.I. 343-1 at 228). The fourth source is an article which claimed interaction rates of up to 25% for certain rich-media ad formats in some markets, yet Mr. Sherwood cited it for the proposition that ABS showed a 25% increase in CTR, even though CTR is a subset of interaction rates. (D.I. 343-1 at 236). The fifth source was a Microsoft study that found that certain ad formats resulted in a 30% increase in agreement with the statement that "this ad makes me want to click on it," yet Mr. Sherwood conceded that he could not use the study without a separate finding regarding CTR. (D.I. 343-1 at 234).
Robocast does not fully respond to the substantive arguments raised by Apple, but merely states that internet sources are not per se unreliable, that Apple has not cited a case for the proposition that results reported by "interested parties" are per se unreliable, and that the data is reliable because Mr. Sherwood "verified the reliability of his sources by researching the sources' background to satisfy himself that they were reliable and credible." (D.I. 377 at 17-18). Daubert requires reliable inputs reliably applied. See Power Integrations, Inc. v. Fairchild Semiconductor Int'l, Inc., 711 F.3d 1348, 1373 (Fed. Cir. 2013) ("[T]he Federal Rules of Evidence require that an expert's testimony be the product of reliable principles and methods applied to sufficient facts or data."); ZF Meritor, LLC v. Eaton Corp., 696 F.3d 254, 291 (3d Cir. 2012) (affirming exclusion where underlying data was not sufficiently reliable). The issue is not whether internet sources are always unreliable. The issue is also not whether results reported by interested parties are always unreliable. The question is whether they are reliable here, and whether they have been reliably applied.
It seems indisputable that internet sources, like other sources, have to be evaluated on their own merits.
The extent to which a party's interest in a result makes its self-reported results doubtful is also fact dependent.
Statistical information in press releases and other promotional materials is of doubtful reliability. "Figures often beguile me, particularly when I have the arranging of them myself; in which case the remark attributed to Disraeli would often apply with justice and force: 'There are three kinds of lies: lies, damned lies, and statistics.'" Mark Twain, Chapters from My Autobiography, North American Review, September 7, 1906, available at http://www.gutenberg.org/files/19987/19987.txt. I only have to consider that the five sources have a wide spread (21% to 68%), which one would not expect if they were accurately measuring the same thing. I also consider that the "statistics" have nothing approaching a sufficient explanation of what they are purporting to measure. There is nothing approaching an explanation of the circumstances surrounding the measurements. The measurements themselves are described imprecisely, and it is apparent that Mr. Sherwood does not, for the most part, know what the measurement actually measured. Garbage in, garbage out. I hold that Mr. Sherwood's underlying data was unreliable, and was unreliably applied, and I therefore exclude his testimony and any testimony for which it forms a foundation.
In fact, some sources reported increases of 300%, 500%, and 600%, but Mr. Sherwood "placed less weight on these upper end result data as not typical." (D.I. 343-1 at 135). This data is no more or less reliable than that relied upon by Mr. Sherwood, yet he chose not to use it, presumably because no jury would believe such implausible results.
See Wikipedia, http://en.wikipedia.org/wiki/Garbage_in,_garbage_out ("I have been asked, 'Pray, Mr. Babbage, if you put into the machine wrong figures, will the right answers come out?' ... I am not able rightly to apprehend the kind of confusion of ideas that could provoke such a question.") (quoting Charles Babbage, Passages from the Life of a Philosopher (1864)).
I need not reach the question of whether Mr. Sherwood's testimony fits the facts.
--------
Defendant's motion (D.I. 341) is hereby GRANTED.
__________
United States District Judge