Two principles are central to the law of patent infringement damages. The first is that, pursuant to the long tradition of common law burdens of proof, plaintiffs bear the burden of proving damages. The second is that, pursuant to the statute, damages awards must not be less than a reasonable royalty.

In the context of patent damages (and, indeed, in many other similar contexts as well) these two principles end up being in deep tension with each other. This is because, in almost all cases, we can be certain that the correct amount of a reasonable royalty is greater than zero. At the same time, in almost all cases, the evidence about the precise amount of what a reasonable royalty would have been involves unreliable guesswork. The main standard for determining the precise amount of a reasonable royalty is to ask what royalty a hypothetical licensing negotiation before infringement would have yielded. Since the negotiation is hypothetical, it inherently requires some element of speculation to know its precise result.

This tension was exposed most starkly in the recent case of Apple v. Motorola. In Apple, the district court (Richard Posner of the Seventh Circuit sitting by designation) found all of the damages evidence presented by the parties to be speculative and unreliable, and therefore excluded them. Since there was no admissible evidence concerning the precise amount of damages, he reasoned, it followed that no damages could be awarded because the plaintiff failed its burden of proof. The result was that the defendant may have infringed a patent, but it needed not pay any damages for that infringement.

The Federal Circuit overturned this result. It held that, because the statute guarantees a minimum award of a reasonable royalty, the failure of a patentee to prove the precise amount of what a reasonable royalty would have been does not justify awarding no royalty at all. “If a patentee’s evidence fails to support its specific royalty estimate, the fact finder is still required to determine what royalty is supported by the record.” In this regard, it specifically noted that “it seems unlikely that a willing licensor and willing licensee would agree to a zero royalty payment in a hypothetical negotiation” and “we know of no case where we found that the record supported an infringement award of a zero royalty.”

The implication of the Federal Circuit’s Apple decision seems to be that, even when the entire evidence concerning the content of the hypothetical negotiation is highly speculative and self-serving—for example, if the patentee testifies he would have only accepted a royalty of 99% and the defendant testifies he would have only accepted a royalty of 1%—the jury is required to make its best guess of what the true outcome would have been. It cannot choose an award of zero simply because the evidence in support of a precise number is speculative.

One can understand the impulse behind this. As mentioned above, determining the precise amount of reasonable royalty damages is inherently a speculative exercise, since it involves asking a counterfactual hypothetical. At the same time, we can be highly confident that the true answer to the question is almost never “zero.” Holding that plaintiffs should receive nothing merely because they cannot prove exactly how much they would have received is a recipe for systematic under-compensation of patentees and gives an unjust benefit to infringers.

At the same time, it is important to recognize that the impulse toward awarding successful plaintiffs at least something—reflected most openly in Apple but also embedded in older cases—has a pernicious effect. Much of the recent reform of patent damages law involves demanding that patentees meet their burden of proof to supply reliable evidence regarding the precise quantity of damages. Most famously, the Federal Circuit’s opinion in Lucent v. Gateway, 580 F.3d 1301 (2009) set aside a $357 million jury verdict because the court found that Lucent’s evidence concerning comparable licenses was too speculative to support that award. The logic of Apple—that juries must award something, no matter how speculative the plaintiff’s evidence is—greatly undermines this effort.

For example, what if Lucent had not bothered to provide any evidence at all besides its own testimony that it would have demanded $357 million? And what if, pursuant to the logic of Apple v. Motorola, the jury had found that, although it was highly uncertain about the precise quantity of a reasonable royalty, it was confident that the reasonable royalty was more than zero, and $327 million was its best guess? A court that overturns such an award must either substitute an award of zero (which would be inconsistent with Apple), or it must take its own guess, one no less speculative than the jury’s. A court that sustains such an award puts itself on a path towards ever-lower standards of speculative evidence.