Friday, January 25, 2019

The FTC has defeated Qualcomm with respect to industry as well as expert testimony: checkmate

Before today's trial session (Day 9), it was already clear (based on what had happened on the first eight days and Qualcomm's witness list for today) that there was no way Qualcomm could even come close to equalizing the score with respect to industry testimony. But Qualcomm's only meaningful partial victory had been how Cravath's Gary Bornstein impeached one of the FTC's three expert witnesses, Michael Lasinski, and different experts will always have different methodologies with strengths and weaknesses even the most reasonable people can normally disagree on. My thinking before today's bench trial session was that the FTC was going to win based on the combination of three factors: Qualcomm's royalty rates being totally out of line (so no one can possibly rationalize them); Judge Koh's pretrial rulings, which already resolved or expressed an opinion on some central issues; and a huge advantage with respect to industry testimony--which, if given significant weight, could outweigh other considerations and, along with the Donaldson Report based on 31 years of active licensing expertise, could provide a solid basis for Professor Carl Shapiro's theory of anticompetitive harm in the form of supra-FRAND royalties being passed on, in part, to consumers.

I thought Qualcomm would save the best for last, calling Professor Aviv Nevo (Pennsylvania) on the final day of its case-in-chief. And I expected him to be a worthy adversary for Professor Shapiro. He might have been on an equal footing with him under different circumstances, but to my surprise, the FTC's lead counsel, Jennifer Milici, has already impeached him beyond recognition before Professor Shapiro will rebut on Monday.

I seriously don't think the FTC even needs a rebuttal anymore in order to prevail on at least a couple of key claims. If Judge Koh canceled (as she obviously couldn't and wouldn't) the rebuttal and the closing arguments in order to take the case under advisement, the FTC would be in great shape already. By beating Qualcomm at the game where its vast resources could have given it an advantage (expert testimony), after thrashing it with respect to industry testimony, the FTC's litigators have reached the point at which I guess we're going to see ever more Apple- and Huawei-related conspiracy theories between now and Judge Koh's opinion, all of which is to no avail as long as Qualcomm can't get a third commissioner to support some fake settlement based on arbitration.

Arbitration is a topic that the FTC addressed today in response to testimony Qualcomm's outside counsel had elicited from inhouse counsel. The FTC pointed to where the real issue is: Qualcomm wants to collect royalties set by arbitrators based on portfolio size, without previously establishing that valid patents are actually infringed (and not exhausted; patent exhaustion is a Godzilla-size specter down in San Diego). This month we've all seen what happens when Qualcomm's (like anybody else's, to be fair) patents in this field are challenged: the USPTO instituted 21 inter partes reviews of Qualcomm patents at Apple's request, and the European Patent Office preliminarily agrees with Apple and Intel on two patents Qualcomm is asserting over there.

I can claim without fear of substantiated contradiction that I'm the longest-standing and most loyal "Trumpian" among tech law and policy bloggers, and I'm disappointed that some conservatives with whom I agree on many other issues think it's conservative when companies try to manipulate government decisions to their competitive advantage. Rush Limbaugh explained a long time ago that conservatism is about fair competition and not about corporatism, not about lobbying, not about pandering. Applying this to FTC v. Qualcomm, I can't see what would be conservative about Qualcomm securing a settlement based on an unfairly parameterized arbitration mechanism allowing it to get paid for 130K patents when ligitation statistics for the industry at large suggests only a minority of them, possibly just a small minority, are actually valid and infringed.

The battle of the experts was the final battle here, and Qualcomm has lost it, too (even prior to rebuttal). When the FTC's Mrs. Milici cross-examined Professor Nevo, it was the equivalent of a Waterloo.

Compared to him, Dr. Chipty (another Qualcomm economist seeking to rebut a part of Professor Shapiro's analysis) had performed well up to a certain point, though it obviously impeached her big-time that she couldn't even say whether Samsung had possibly bought more than 200 million thin models during a certain period, and had to concede she wouldn't have considered such a fact. And Professor Andrews was just a dreamer talking about patents he deemed superstrong, but if a judge--such as patent-savvy Judge Koh--had handed down a claim construction after a defendant making the case for narrow interpretation, and if they had been challenged like the 23 Qualcomm patents that came under pressure just over the last 9 or 10 days, the result would be sobering. However, Professor Nevo lost too much credibility today.

Mrs. Milici asked him what (unless something better comes up on Day 10) was the single best question of the entire trial:

She was the star today. Big Law should consider making her and other members of her team an offer.

What was just as impactful: she showed that Professor Nevo had based his analysis on what he called "contractual" royalty rates and Mrs. Milici more descriptively labeled "headline royalty rates": if there was a royalty clause in a contract, he took that one, but in most cases it wasn't the actual royalty after considering rebates, kickbacks, and all sorts of other deal terms. That is, by the way, also an issue with how Qualcomm presented its 2018 amendment to an old agreement with Samsung: as the FTC's Dan Matheson (who also did very good cross-examinations during this trial) pointed out, the royalty rate under that contract gets adjusted downwards in various ways, through direct payments to Samsung, and was just part of a package deal that also involved Qualcomm receiving manufacturing services ("foundry") from Samsung. In such a package deal, you can theoretically also have a nominal royalty rate of 50%, not just 5%, but bring it down elsewhere. It means nothing for the purposes of a FRAND-compliance analysis.

Even before the cross-examination I had already noticed that Professor Nevo didn't consider all contract terms, as I highlighted on Twitter.

On redirect, Professor Nevo attempted to rationalize his selectivity--which is still selectivity, even if there were criteria as he claimed--by saying that Professor Shapiro had also listed those "contractual" ("headline" as per the FTC) royalties in his report. But that's an apples-to-bananas comparison given that the two experts undertook disparate things here:

Professor Shapiro had made it very clear that he didn't endeavor to quantity the damages resulting from Qualcomm's allegedly anticompetitive conduct. He merely sought to show that if there was whatever quantity of damage, there was also anticompetitive harm in the form of supra-FRAND royalties being passed on to consumers, and because of indirect effects on competition in the chipset market. So for Professor Shapiro's purposes, there simply wouldn't have been any point in digging deeper.

But Professor Nevo tried with respect to patent royalties what Professor Snyder had previously attempted with respect to design wins (getting one's chipset into devices): to call the presence of anticompetitive conduct into question by explaining it away, saying that other factors simply explain the real-world outcomes. (There are some gradual differences, and this time around Qualcomm itself asked its expert the question that came down to conceding that if there had been anticompetitive behavior, some effects such as falling prices would have been even stronger under fair conditions.)

The expert who doesn't even pretend to quantify harm--Professor Shapiro--doesn't need to look at all deal terms bringing down an effective royalty. But the expert attempting to explain away anticompetitive conduct on the basis that the outcome (the royalties companies ended up paying) of real-world processes (such as negotiations) supports Qualcomm's royalty rate must be precise. He cannot possibly afford shortcomings that materially affect the numbers. Nor can he afford such selectivity that several of the largest licensees during a relevant period are just ignored.

It's a mystery why Qualcomm couldn't at least win the battle of the experts. In the end, its best expert indeed was its own employee Lorenzo Casaccia, who definitely knows everything about 3GPP. He's an employee, which adversely affects his credibility as an expert (even if an external experts makes millions, he still has a reputation to defend because he wants to be hired by others later), and Mr. Casaccia's topic was narrow, but he knew it inside out.

It could be that some people wouldn't even have been willing to testify for Qualcomm (if Qualcomm approached them at all, which I cannot know) but looked at the facts and said "thanks, but no thanks" even to a multi-million dollar Mission Impossible.

Even before Mrs. Milici's brilliant cross-examination, Professor Nevo had already lost his credibility with me. Apart from little things that didn't convince me, there were two fundamental flaws during his direct examination that showed me his testimony in this case (again, he may be great in other industries or other cases) can't be taken seriously, to put it mildly:

  1. He made reference to an analogy that already came up during Professor Shapiro's initial testimony: Microsoft's operating system license deals that raised antitrust issues. Professor Nevo said the difference was that Qualcomm wasn't giving away chipsets, but selling them at a profit (even under the exclusive deal with Apple, for instance), while Microsoft was giving away operating systems. #FACEPALM

    At this trial, there were at least two people in the audience who know what utter nonsense he spouted. One is a former DOJ official who was involved with the related investigation. The other person is yours truly because I did business with a company named Berkeley Softworks, later renamed GeoWorks, at the time headquartered on 2150 Shattuck Avenue in Berkeley, CA. In 1990 I received stock options in that company--the first time I owned "a piece of America." And that company, which I already knew several years before because of a Commodore home computer software product range, was trying to compete with Microsoft Windows. GeoWorks' product was named GeoWorks Ensemble, and contained the proprietary PC/GEOS operating system.

    We were selling the product through retailers, but also to OEMs. From them I learned the term OEM. And I remember the problem we had with Microsoft's strategy. Microsoft gave OEMs two alternative choices: they could pay per actual installation of MS-DOS/Windows (so if they shipped 200K personal computers, but only 100K came with MS-DOS/Windows, they paid 100K times price A); or they could sign a contract under which they paid a price B, much lower than A (so they didn't have a rational alternative), but committed to paying it on each and every PC they sold, whether or not MS-DOS/Windows was included. The choice between A and B was a rhetorical question: of course everyone opted for B.

    The effect was that an OEM had the Microsoft "tax" on a PC even if another operating system was installed instead of Microsoft's product. So if an OEM with such a deal wanted to ship a product like Digital Research's DR-DOS or GeoWorks' PC GEOS, they had to pay Microsoft anyway and had the alternative system's royalty on top, so they had to sell those non-Microsoft-run PCs at a higher price and/or lower margin.

    By saying Microsoft was giving away software, he completely failed to understand the issue. The issue was that Microsoft's competitors would have had to give away their competing product to get a sizable installed base. At the time, venture capital wasn't available in such quantities that anyone could have survived it. But Professor Nevo also got this wrong for another reason: those OEM deals were pure license deals. The OEMs had to preinstall and/or duplicate the software on floppy discs, and had to print the manual. It was a 100% threshold-margin business for Microsoft--which is impossible for chips, which do involve cost of goods and manufacturing.

    The parallel between the Microsoft issue of about three decades ago and Qualcomm's patent tax on competitor's products is simply that device makers have to pay Qualcomm anyway, whether or not they use one of its chips, and it gets even worse when someone like Apple, during the term of those meanwhile-expired exclusive agreements, faced a de facto contractual penalty for shipping a single unit with a competitor's baseband chip. Compared to such a penalty, the Microsoft "per hardware unit regardless of actual software" operating system license was less of a problem because it only had a gradual effect on competitiveness, not a "clawback provision" over huge amounts.

  2. In connection with licenses to rival chipsets, he said the same nonsense about chipset makers being more competitive without paying royalties--a fallacy debunked even before the trial. Then, in this context, he also argued it was more efficient not not grant licenses to chipset makers since some patents would still have to be licensed at the device level. But the question of licensing rival chipset makers is only about cellular SEPs, and Judge Koh has already taken a clear position on the smallest salable patent-practicing unit (in GPNE Corp. v. Apple and, in practical terms, in her summary judgment in the FTC case). The worst part, however, is that Professor Nevo warned against having too many license agreements in place, when there's only a few baseband chipset makers (like Intel and MediaTek), but huge numbers of device makers. For instance, Qualcomm's licensing president Mr. Rogers later testified about a meeting with about 30 Chinese device makers. So the number of license agreements doesn't really increase much if you license a few chipset makers, but the net efficiency gain results from the fact that you then have fewer patents to sort out with those huge numbers of device makers.

Even though Cravath's Mr. Bornstein is extremely good at the intersection of law and economic analysis, and he and Professor Nevo were a highly efficient tandem (probably the highest text-to-time ratio of the whole trial), this day was dreadful for Qualcomm. In the courtroom, it's now practically Game Over. The FTC won't necessarily get all the injunctive relief it's seeking, but a resounding victory is the most likely outcome. Qualcomm's PR and lobbying efforts, with the Chinese bogeyman named Huawei and other theories, won't end until all appeals have been exhausted, however. When I read such crap as in the Washington Post (which President Trump dislikes even more than the New York Times, with which I think it's actually a hate-love relationship) about Huawei testifying for the FTC, I wonder whether some of those Qualcomm-aligned columnists would also defend the flat-Earth theory only because Huawei says the Earth is round: let's face it, Huawei is just one piece of a huge puzzle here and merely says what American companies and companies from an ally nation like South Korea have said as well. Of course, Huawei should not get an unfair advantage, or (a subject on which I don't have an opinion) pose a security risk, but letting Qualcomm get away with its conduct does nothing to address those actual or potential problems.

As for the FTC's huge advantage with respect to industry testimony, Qualcomm only gained limited ground before resting its case. The three companies supporting it through testimony are also often seen alongside Qualcomm as members of anti-FRAND lobbying efforts: Nokia, Ericsson (this one only with respect to rival chipset licensing, while it supports the FTC in other ways), and InterDigital. Nokia and Ericsson stopped making handsets a while ago, and InterDigital never made any product (unless litigation counts as a product--they've been quite prolific in that regard). Two European failed businesses (with respect to handsets) and a U.S. patent assertion entity (PAE)--those are not the kinds of allies that would Make America Great Again.

Share with other professionals via LinkedIn: