The opinion, authored by Judge Taranto and joined by Judges Linn and Chen, is Prism Techs. LLC v. Sprint Spectrum L.P., came out this morning and is available here. The patents in suit "claim and describe methods and systems for managing access to protected information provided over certain networks that . . . must be 'untrusted' networks" (p.2). The jury returned a verdict for Prism in the amount of $30 million. The key substantive issue on appeal was whether the district court improperly allowed Prism's expert to modify the court's claim construction of the term "Internet Protocol Network. (The court holds that it did not, because the expert actually didn't modify the meaning of the term.) I'll focus my attention on the damages issues, which mostly center on Prism's expert's use of a Settlement Agreement Prism entered into with AT&T in a case involving inter alia the same two patents, which settled "[o]n the last day of trial, just before closing arguments" (p.5).
The use of settlement agreements as comparables is at least somewhat controversial, so this could prove to be a very significant opinion on this issue. Here is the Federal Circuit's take on when district courts may use settlements as relevant evidence of the value of the technology, and when (by contrast) their probative value is outweighed by the risk of unfair prejudice:
This court has recognized that, depending on the circumstances, a license agreement entered into in settling an earlier patent suit sometimes is admissible in a later patent suit involving the value of the patented technology, and sometimes is not. . . . As to settlements generally, the Supreme Court has explained the normal settlement calculus for litigants: “Most defendants are unlikely to settle unless the cost of the predicted judgment, discounted by its probability, plus the transaction costs of further litigation, are greater than the cost of the settlement package.” Evans v. Jeff D., 475 U.S. 717, 734 (1986); Staton v. Boeing Co., 327 F.3d 938, 964 (9th Cir. 2003) (quoting Evans, 475 U.S. at 734). That formulation—enumerating “the cost of the predicted judgment,” “its probability,” and “costs of further litigation”—helps identify why and when a district court, conducting the inquiry required by Rule 403, can find earlier patent-suit settlements admissible in valuing a patented technology.
On one side of the Rule 403 balance is the strong connection a settlement can have to the merits of an issue common to the earlier and later suits. Specifically, a settlement involving the patented technology can be probative of the technology’s value if that value was at issue in the earlier case. The reason is simple: such a settlement can reflect the assessment by interested and adversarial parties of the range of plausible litigation outcomes on that very issue of valuation. And given the necessary premise that discovery and adversarial processes tend to move a legal inquiry toward improved answers, the parties’ agreement seems especially probative if reached after the litigation was far enough along that the issue was already well explored and well tested. See AstraZeneca, 782 F.3d at 1336–37. On the other side of the balance, for various reasons a settlement may be pushed toward being either too low, as in Hanson, or too high, as in LaserDynamics, relative to the value of the patented technology at issue in a later suit. As to the former, for example, even if the technology is identical in the earlier and later suits, the earlier suit’s settlement figure may be too low to the extent that it was lowered by the patent owner’s discounting of value by a probability of losing on validity or infringement. As the unchallenged jury instructions in this case indicate, the hypothetical-negotiation rubric for the assessment of royalty damages assumes that the asserted patents are valid and infringed. See J.A. 23473–75; Lucent Techs., Inc. v. Gateway, Inc., 580 F.3d 1301, 1325 (Fed. Cir. 2009). Accordingly, whereas a settlement reached after a determination of liability (though subject to appeal) is particularly reliable as evidence of value, AstraZeneca, 782 F.3d at 1337, a settlement tends to undervalue the technology where it reflects a discount for the probability of losing. A patent owner may also accept too little, relative to the patent’s value, when it accepts an amount out of a desire to avoid further expenditure of (presumptively unrecoverable) litigation costs.
At the same time, various factors may work in the opposite direction, tending to make a settlement of an earlier suit too high as evidence on the valuation question presented in a later suit. An earlier settlement may cover technology either not the same as or comparable to the patented technology at issue in the later suit, or may cover the patented technology plus other technologies. The earlier suit may have included a risk of enhanced damages, a factor in the settling parties’ assessment of risk that would push settlement value above the value of the technology. And, of course, the litigation costs still to come at the time of settlement may loom large in parties’ decisions to settle. . . .
What is needed for assessing the probativeness and prejudice components of the Rule 403 balance, then, is consideration of various aspects (of which we have mentioned some) of the particular litigation settlements offered for admission into evidence. . . . [A]s a logical matter, the mere filing of a complaint—shifting from potential to actual litigation—does not automatically turn the prejudice side of the Rule 403 balance into one that substantially outweighs the probativeness side. The particulars of the case that was settled and the settlement, as well as of the case in which the settlement is offered as evidence, matter to the Rule 403 balance (pp.11-14).
Applying these principles here, the court concludes that the lower court "had an adequate basis for admitting the AT&T Settlement Agreement," which "covered the patents at issue here, though not only the patents at issue here. In that common situation, evidence was needed that reasonably addressed what bearing the amounts in that Agreement had on the value of the particular patents at issue here . . . Prism supplied such evidence . . . (p.15). Further:
The circumstances of the AT&T Settlement Agreement affect the Rule 403 assessment in ways that support the district court’s admission of the Agreement. The Agreement was entered into, not just after all discovery was complete, but after the entire trial was finished, except for closing arguments and jury deliberations. Thus, the record was fully developed and thoroughly tested in the adversarial process, enhancing the reliability of the basis on which Prism and AT&T were assessing the likely outcome. The timing of the settlement also means that a very large share of litigation costs had already been sunk, reducing (though of course not eliminating) the role of litigation-cost avoidance in the settlement decision. Moreover, Sprint has not suggested that enhanced damages were at issue by the time of the settlement; and the proposed jury instructions and verdict forms suggest that they were not. On the other hand, validity and infringement were still open issues at the time of the settlement. But Sprint cannot rely on that fact: possible non-liability is a factor that tends to make settlements too low, not too high" (pp. 15-16).
The court also notes that AT&T itself offered some (other) settlements agreements in evidence (p.14). In addition, the court expresses skepticism regarding Sprint's argument that the Supreme Court's 1889 decision in Rude v. Westcott, which sets out certain conditions for proving an "established" royalty--including that the royalty was not the product of a settlement--establishes a categorical rule against the use of settlements as evidence of a reasonable royalty, though ultimately it rejects the arguments based on Sprint's failure to present the argument to the district court in a timely fashion (pp. 16-20.) (For more on the distinction between established and reasonable royalties, see also Roger D. Blair & Thomas F. Cotter, Rethinking Patent Damages, 10 Tex. Intell. Prop. L.J. 1, 7-9 (2001)). The court also seems skeptical of Sprint's argument that Federal Rule of Evidence 408, which bars the use of settlements under some though not all circumstances, applies here, but it ultimately rejects that argument too based on the failure to preserve (pp. 20-22). Finally, the court rejects Sprint's argument that the district court should have rejected expert testimony on the "estimated costs that Sprint avoided by infringing," concluding that the experts adequately substantiated their opinions in this regard (pp.23-26); and it rejects Prism's request that additional damages be awarded for the postinfringement time period, based on the judge's conclusion that the jury award itself was intended to compensate for all infringement, past, present, and future (pp. 26-29).
The court's analysis of the admissibility of settlement agreements tracks with Jonathan Masur's argument that licenses entered into in settlement of litigation sometimes may be more probative than other licenses, particularly if they are entered into late in the game when the parties perceive the probability of infringement and validity to be high. See Jonathan S. Masur, The Use and Misuse of Patent Licenses, 110 Nw. U. L. Rev. 115, 145-48 (2015). On the other hand, I wish the court had said something about the counterargument, that settlements (again, particularly those entered into late in the game) also may reflect "holdup" value, i.e., the defendant's desire to avoid eating its sunk costs or incurring higher switching costs if it has to design around. (Later on, however, in its discussion of the cost-saving point at pp. 24-25, the court does cite Riles v. Shell Exploration & Production Co., 298 F.2d 1302 (Fed. Cir. 2002), which as Norman Siebrasse and I point out in one of our papers does stand for the proposition that courts should avoid basing royalties on holdup value.) And while it may be that the desire to avoid litigation costs factors into the amount of a settlement, unless those expected costs are asymmetric (or the parties have different tolerances for risk) those costs should affect both parties equally and thus cancel out, I think.