From Marconi to Microsoft: The Rise And Fall of the “25 Percent Rule” for determining damages from patent infringement

What is a patent worth?  In 1943, Nikola Tesla, who patented many fundamental inventions only to watch others become wealthy using them, died impoverished in New York City.  Tesla had outlived his rival in radio technology, Guglielmo Marconi, who had received riches, a Nobel Prize, and a noble title for his own inventive efforts.

The last laugh was Tesla’s.  Just a few months after he died, in Marconi Wireless Telegraph Co. v. United States, 320 U.S. 1 (1943), the Supreme Court invalidated Marconi’s broad radio patent, chiefly on the basis of Tesla’s earlier – and far less remunerated – patent on “transmit[ting] intelligible messages to great distances.”  Before the decades of litigation over the invention of radio came to an end with this landmark decision, however, they had given birth to a curious rule, later to become known as the “25 percent rule,” concerning how much patentees—like Marconi or Tesla—ought to be compensated.  This “25 percent rule” would later come to be a key tool of patentees for setting a value for infringement of their patents in federal court.  Until this year, that is when, on January 4, the U.S. Court of Appeals for the Federal Circuit struck down the “25 percent rule” in its decision in Uniloc USA, Inc. v. Microsoft Corporation.

By disapproving this ancient, yet arbitrary, rule, the Federal Circuit left patent damages doctrine in a state of uncertainty.  When one sets out to prove an amount of damages for infringing a patent, much like when one sets out to make an invention worth patenting, one of the most difficult questions can be where to begin.   By eliminating a popular rule of thumb for such a starting point, the Court returns patentees and future courts to that original, vexing question.  The answers given in future cases will influence whether many patentees will receive remuneration more resembling that of Marconi or that of Tesla.


Six years ago, Uniloc sued Microsoft alleging that a feature in Microsoft’s Word and Office software infringed Uniloc’s Patent No. 5,490,216 (“‘216 patent”).  Uniloc’s patent covers a mechanism for combating “casual copying” of software by consumers attempting to install one copy of software in several different machines.  The patent claim recites an apparatus in which a software installer enters a string of information (e.g., a serial number) which is uploaded to a remote system for comparison to a unique licensing string.  If the strings match, the system returns data to the local computer that allows the software to run.  According to Uniloc, Word and Office were sold with an anti-copying feature known as “product key activation” that practiced this invention without permission.

Before trial, Microsoft moved in limine to exclude the testimony of Uniloc’s damages expert Dr. Gemini on the amount of harm Uniloc had suffered from the alleged infringement.  Microsoft’s motion argued that Dr. Gemini’s damages analysis was arbitrary because it was based on (1) an unsupported baseline value of $10 per software copy and (2) an unsupported application of the “25 percent rule” to estimate the royalty value of the anti-piracy feature.  The trial court refused to exclude Dr. Gemini’s testimony, and at trial, Dr. Gemini proceeded to tell the jury that, in his expert opinion, Uniloc should be awarded infringement damages of $564,946,803.

Dr. Gemini based his calculations on a hypothetical negotiation between Uniloc and Microsoft before the alleged infringement had began, under a set of factors (widely used in patent cases) set forth in Georgia-Pacific  Corp. v. U.S. Plywood Corp., 318 F. Supp. 1116 (S.D.N.Y. 1970).  But Dr. Gemini’s analysis of this hypothetical negotiation and the Georgia-Pacific factors also relied on the “25 percent rule.”  This rule assumes—at least in the absence of specific information to the contrary—that about 25 percent of the value of a product containing a patented invention comes from the patented invention and should go to the patentee, while the remaining 75 percent of that value is attributable to the rest of the product and should go to the infringer.  Dr. Gemini began his valuation by estimating that the “product key” feature in each copy of Microsoft’s software was worth $10, based on an internal Microsoft document that arguably said product keys could have a value of “between $10 and $10,000 depending on usage.”   Uniloc USA, Inc. v. Microsoft Corp., 632 F. Supp. 2d 147, 150 n.2 (D.R.I. 2009).  He “conservatively” took the lowest of these numbers ($10), and then applied the “25 percent rule” to estimate that 25 percent of the value of the copy ($2.50) should go to Uniloc.  Dr. Gemini then multiplied that figure by the number of Microsoft Windows and Office licenses sold during the pertinent period (nearly 226 million) to arrive at a total damages number of over half a billion dollars.

Dr. Gemini told the jury that he had then “checked” this damages figure to make sure it was reasonable, by comparing it to the “entire market value” of the infringing products.  Since a copy of Microsoft Windows or Office sold for $85, he testified, his total damages figure represented only a tiny fraction of an “entire market value” of nearly $20 billion for the accused software copies, and therefore was quite reasonable.

The jury returned a verdict of infringement, found the patent valid and willfully infringed, and awarded $388 million in damages.  Microsoft moved for judgment as a matter of law, arguing again that the “25 percent rule” was illegally arbitrary and that Dr. Gemini had misapplied the “entire market value” rule.  The district court declined to overrule the jury’s verdict.


Microsoft appealed to the Federal Circuit, again arguing that the “25 percent rule” was inappropriately arbitrary and that Dr. Gemini had misused the entire market value rule.  In an opinion that relied heavily on scholarly criticism and empirical studies as well as prior court decisions, a panel of the Federal Circuit ruled in Microsoft’s favor, overturned the damages award, and rejected the “25 percent rule” as a matter of law.

The panel first ruled that the 25 percent rule “fails to pass muster under Daubert and taints the juries damages calculation.”  Opinion at 47 (citing Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993)).  The panel noted that “a reasonable royalty is the predominate measure of damages in patent infringement cases” and may be determined by estimating the royalty the parties would agree to in a “hypothetical negotiation.”  Id. at 36.  The panel then examined the basis for the 25 percent rule’s assumption that, in such negotiations, parties would assign 25 percent of product value to the patentee as a rule of thumb.  The panel looked at studies that had endorsed the “25 percent rule” and identified methodological flaws in each study, including problems with sample size and geographical reach and a failure to take into account such variables as the importance of the alleged technology, the ability to design around a patentee’s invention, whether the patentee is able to exact monopoly pricing on the licensee, and any pre-existing relationship or past dealings between the parties.  The Federal Circuit concluded that the “25 percent rule” was based on scant evidence, and that its inclusion in numerous patent decisions over the years had been based on an assumption of its utility rather than hard evidence.

A chart of the various courts’ rulings on the pertinent issues is presented below.

Simplified Table Of The Disposition Of Issues During The Uniloc Litigation

Issues Pre-Trial Or Trial Post-Trial Appeal
Was The Patent Infringed? Yes Yes Yes
Was There Willful Infringement? Yes No No
Is The Patent Invalid? No No No
Was Use Of The 25% Rule For Damages Improper? No (This was decided by the judge on a motion in limine) No Yes
Was Use Of The Entire Market Value Rule  Improper? No (This was decided by the judge on a motion in limine) No Yes

A Rule Of Thumb, Or A Thumb On The Scales?

In fact, the origins of the “25 percent rule” as a starting point to estimate a reasonable royalty can be traced all the way back to the 1942 decision of the Federal Court of Claims that the Supreme Court had overturned in 1943.  Marconi Wireless Tel. Co. v. United States, 99 Ct. Cl. 1, 20 (1942).  Before being overruled by the Supreme Court’s finding that Marconi’s broad radio patent was invalid, the Court of Claims had found the patent valid and infringed, and had apportioned 25 percent of total profits to the patent owners as compensation for infringement.  Interestingly, 25 percent happened to be one of a sliding scale of several royalty levels, ranging from 20 percent to 33 percent, that had been expressly set forth in an earlier license granted on Marconi’s patent.  But rather than cite the 25 percent figure as one in an actual past license, the Court of Claims chose to present it as a hypothetical, which happened to involve the entire market value rule:

The status of a patent in the art with which it is associated is of importance in determining the base which is to be used in an accounting. . . . “The patentee . . . must . . . give evidence tending to separate or apportion the defendant’s profits and the patentee’s damages between the patented feature and the unpatented features . . . or he must show, by equally reliable and satisfactory evidence, that the profits and damages are to be calculated on the whole machine, for the reason that the entire value of the whole machine, as a marketable article, is properly and legally attributable to the patented feature.” . . . [T]he application of this rule relates more directly to an accounting based upon profits rather than the type of accounting which is based on reasonable royalty . . . . This becomes apparent if we consider a theoretical instance, in which the profits, due to the patented portion of a machine, have on apportionment been found to be 25 percent of the total profits on the machine, the remaining 75 percent being due to extraneous elements or elements patented by others.  In such a case plaintiff would be entitled only to the profits on the features or elements covered by his patent.(Id. at 46-47 (emphasis in original).)

The Supreme Court’s splintered and winding opinion overturning the Court of Claims’ decision never addressed the validity of assigning damages based on a “25 percent” rule.  Many years later, however, the Court of Claims’ 1942 analysis was cited with approval by a later Court of Claims decision, Tektronix, Inc. v. United States, 552 F. 2d 343, 358 (Ct. Cl. 1977).  Twenty years after that decision, judge Joseph Longobardi of the District of Delaware was to observe that a “Rule-of-Thumb” was now in use in patent cases—although, contrary to the Court of Claims’ original conception, it had by then become a rule used to estimate the “royalty rate,” not the percentage of profits, due the patentee.  Procter & Gamble Co. v. Paragon Trade Brands, Inc., 989 F. Supp. 547, 612 (D. Del. 1997).  “Although the Court will consider the Rule-Of-Thumb analysis in determining the royalty rate,” Judge Longobardi concluded doubtfully, “this approach will not receive substantial weight,” for “the Court has found no case adopting this test as a matter of law.”  Id. As the years went on, the courts’ initially tentative adoption of this “Rule-of-Thumb,” as in Procter & Gamble, gradually became a stronger and stronger embrace—until, by 2010, the trial court in Uniloc v. Microsoft could quite reasonably assert that “[t]he ‘25% Rule’ has been accepted as a proper based line from which to start” the damages calculation.  In Limine, 632 F. Supp. 2d at 151.

As the “25 percent rule” had garnered wider adoption by patentees and courts over the years, it had garnered wider criticism in academia.  Empirical studies pointed to the fact that licensing deals vary greatly between industry and gave little support to a “one size fits all” approach.  Opinionat38.  Further studies indicated that royalty rates had wide statistical distributions, making application of an “average” rate ill-suited to the long tails of such distributions.  The Federal Circuit’s citation to such studies in Uniloc v. Microsoft to undercut the “25 percent rule” highlights the impact that such empirical research may sometimes have on the development of patent law.

Other, less mathematical critiques of the “25 percent rule” had focused on the fact that licensing negotiations are highly heterogeneous and their outcomes highly factbound.  Id. at 29.  Licensing agreements commonly reflect such unpredictable facts as the unique relationship between each patentee and licensor, their past dealings, the ease or impossibility of designing a workaround to the patent, the stage of any litigation proceedings, and so on.  Although such facts do not disprove the reasonableness of a starting “25 percent” estimate, the idiosyncratic nature of such negotiations also seems to counsel against a universal, one-size-fits all “Rule-of-Thumb” for royalties.

The Rule Is Dead; Long Live The New Rule

The Federal Circuit’s abolishment of the “25 percent rule” may leave patentees at something of a loss.  Patentees still need a starting point for estimating a reasonable royalty.  Alternative theories for such starting points might prove more arbitrary than the rule they replace.

Empowered Juries

Ironically, the demise of the “25 percent rule” might give jurors increased, rather than decreased, freedom in determining patent damages.  Without the rule to provide a numerical baseline, jurors may rely even more than before on the non-numeric Georgia-Pacific factors—themselves a heterogeneous, often-criticized collection of miscellaneous considerations, which not only lack the simplicity of the “25 percent rule,” but often point in many directions at once, as recently noted by Professor Mark Lemley and Daralyn Durie.  Durie & Lemley, “A Structured Approach to Calculating Reasonable Royalties,” 14 Lewis & Clark Law Review, 627 (2010). Because the Georgia-Pacific factors are so flexible, it may be even more difficult to police juries’ determinations based on them than it was to police juries’ application of the old “25 percent rule.”  This result would appear to be in some tension with other recent Federal Circuit jurisprudence, which has tended more to reduce, or at least question, lay jurors’ flexibility in deciding patent law arcana.  The panel in Uniloc v. Microsoft itself voiced just such a concern.  Opinion at 33.

Increased Empiricism

Therejection of the “25 percent rule” continues a recent trend in patent law away from general one-size-fits-all rules in favor of more flexible approaches.  For example, in Princo Corp. v. ITC, 616 F.3d 1318 (Fed. Cir. 2010), the Federal Circuit recently refused to extend the patent misuse doctrine to situations where a patentee who offers to license a patent also induces a third party not to license its separate, competitive technology.  In In re Seagate Technology, LLC, 497 F.3d 1360 (Fed. Cir. 2007) (en banc), the Federal Circuit required more fact-intensive allegations to establish willful infringement.  And in KSR Corp. v. Teleflex, Inc., 550 U.S. 398 (2007), the Supreme Court overturned a rigid doctrine of nonobviousness previously applied by the Federal Circuit, and replaced it with a more flexible approach to patent invalidity.

In Uniloc v. Microsoft, the Federal Circuit seemed to indicate that estimation of patent damages should be based on empirical data, and that the analysis undergirding any rule be closely tied to the factual specifics of the case.  Under this reasoning, a general rule can rarely be appropriately applied across situations involving divergent technologies or parties.  But the requirement that damages calculations must be tightly bound to the factual specifics of a case may prove challenging to implement, for when a damages expert analyzes, as he must, a “negotiation” that is inherently hypothetical, how can the expert’s testimony on that hypothetical satisfy the standards necessary to qualify the testimony as reliable under Daubert?   Every negotiation over licensing, even real ones, contains a large number of unknown variables, and even the use of prior licensing deals might be suspect since they would probably involve different companies, address different components, and occur at different times.   And damages experts do not even have the luxury of opining on real negotiations, but are usually required to analyze hypothetical ones, where the entire litigation was hypothetically avoided by an agreement that was never actually reached.   If damages calculations should start by assuming a hypothetical negotiation between the parties, but data used in determining the outcome of a hypothetical negotiation must be closely tied to the facts of a case, parties may struggle to find any empirical data that actually matches the likely shape of a negotiation between the actual litigants.  It is such situations that likely caused parties to turn to “illogical” valuation methods like the “25 percent rule” in the first place.    These conflicting signals concerning how to prove damages could lead, at least in the short term, to use of new, even more obscurantist ways of valuing damages.

How experts will deal with this mandated hypothetical inquiry, while satisfying the reliability requirement under Daubert, with fewer rules of thumb to guide them remains to be seen.  As patentees seek ways to prove damages, new techniques that confuse the issues may also confuse the courts. The time may have come to revisit the fundamental tenets of determining patent damages by imagining hypothetical negotiations under the Georgia-Pacific factors.  A new approach may give parties more guidance and provide more rational damages calculations.

Continuing Questions

Despite having dozens of seminal patents, Nikola Tesla died alone and deep in debt.  In contrast, Guglielmo Marconi received greater reward for his inventions even though Tesla’s were eventually found to have preceded them.  Yet both were great inventors.  Their different fates highlight the difficulty inherent in determining what rewards that patentees should receive.  By coincidence, the clash between their creations itself helped lead to creation of a rule that long served as a guidepost for valuing those rewards.  Now that this flawed rule is gone, the time has come to invent a better rule to replace it.

What is a patent worth?  The Federal Circuit has taken away a rule that was once used to look for the answer.  But the question is still there.

By Dorian Berger and Kenneth Weatherwax, Irell & Manella LLP, Los Angeles

All images and illustrations used in our posts are licensed and have been legally acquired through official sources such as Adobe Stock