158KB Sizes 0 Downloads 6 Views


MAKAN DELRAHIM Deputy Assistant Attorney General Antitrust Division U.S. Department of Justice

Presented at the

British Institute of International and Comparative Law London, England

May 10, 2004


Introduction It is a pleasure to be here today to discuss an issue at the forefront of antitrust and

intellectual property law: compulsory licensing as a remedy in antitrust cases. Compulsory licensing of intellectual property rights, for good or bad, has been a potential remedy to a whole host of antitrust liability theories, from consummated mergers to refusals-to-deal. Today, I will focus on compulsory licensing purely as a remedy in antitrust matters and not in relation to any particular cause of action. From the U.S. Supreme Court’s decision in Besser Manufacturing 1, to the district court’s decision fifty years later in United States v. Microsoft Corporation 2, courts have recognized that compulsory licensing can be a necessary remedy in some cases. Compulsory licensing, however, also has the real potential to harm innovation. My goal today is to illustrate the policy and practical issues posed by compulsory licenses, to advise caution when using this remedy, and to suggest ways to use compulsory licensing, when required, that place a minimal burden on innovation and future competition. As an antitrust remedy, compulsory licenses can be divided into two types: those related to mergers and those related to non-merger cases. 3 Compulsory licensing as a merger remedy is a well-established tool and has not been particularly controversial. Non1

United States v. Besser Mfg. Co., 343 U.S. 444, 447 (1952) (imposing compulsory licensing on a "fair" and, apparently, nondiscriminatory basis). 2

231 F. Supp.2d 144, 190 (D.D.C. 2002) (decision conditionally approving the consent decree). The consent decree at issue in that reported decision, and currently in effect, is available at 3

Statutory compulsory licenses also exist. See, e.g., 7 U.S.C. § 2404 (patents necessary for the nation's food supply); 42 U.S.C. § 2183 (patents necessary for national atomic energy needs); 35 U.S.C. § 203 (patents developed through the use of government research funding); 17 U.S.C. § 115 (copyrights in certain musical works). While these clearly qualify as compulsory licenses, they are not remedies -- a private or governmental demand for such licenses does not require proof that the owner has violated any law.

merger compulsory licensing imposed by an agency or the courts, though, should be a rare beast. As a point of structure, I will turn first to the history of the compulsory licensing remedy in the US and the EU. Next, I will discuss the policy and practical issues raised by compulsory licensing, particularly in the non-merger, non-consent context. Last, I will mention those limited situations where I believe such licensing is the most appropriate, and I will suggest some key items for making a compulsory license potentially less problematic. With that in mind, let’s turn to the history of the remedy in the United States. II.

A Brief History of Compulsory Licensing as a Remedy A.

United States

Any discussion of licensing in the United States must start with our Antitrust Guidelines for the Licensing of Intellectual Property, which state that the United States enforcement agencies attempt to "apply the same general antitrust principles to conduct involving intellectual property rights that they apply to conduct involving any other form of tangible or intangible property."4 That statement is an excellent introduction into my discussion of compulsory licensing in the context of mergers. If we start from the principle that an antitrust agency can certainly demand divestiture of a physical asset in a merger case, we would expect, from the principle articulated in the Guidelines, that the agency would also be able to demand divestiture of an intellectual property asset. Taking this one step further, since compulsory licensing is often less burdensome than outright divestiture, one should expect an agency to be able to use compulsory licensing as a 4

See 1995 Department of Justice & Federal Trade Commission Antitrust Guidelines for the Licensing of Intellectual Property § 2.1, available at


potential merger remedy in some situations where, without the license, the merger would result in a substantial lessening of competition. And this is exactly what does, in fact, occur. In the United States, appropriately narrow compulsory licensing is an established remedy in the context of mergers that otherwise would violate the law and can serve as a lessrestrictive and more efficient alternative to divestiture. 5 This remedy is relatively uncontroversial, perhaps due to the fact that most such cases appear with the consent of the parties and involve consent decrees and most involve grants to only a single licensee, which is a way merely to preserve the same number of competitors that existed pre-merger. 6 But it is crucial to make such a license as narrow as possible to avoid collusion among competitors in markets not affected by the proposed merger. This is an important point to remember. As the U.S. Supreme Court acknowledged in its recent Trinko decision, “compelling negotiation between competitors may facilitate the supreme evil of antitrust: collusion.” 7 Turning now to non-merger remedy cases, compulsory licensing has a long but contradictory history. At the turn of the last century, the dominant mode of thinking was to view patent law and antitrust law as opponents -- one granted monopolies and one took monopolies away. That mindset required courts to choose sides in a patent-versus-antitrust


See, e.g., USDOJ/NK AK Steel consent decree, press release available at; USDOJ/3D System Corp. consent decree, press release available at; USFTC/Dow Chemical Co. consent decree, analysis available at; USFTC/Boston Scientific consent decree, analysis available at 6

Id. (patent license to a single competitor; preserved two-competitor market).


Verizon Communications Inc. v. Law Offices of Curtis V. Trinko, LLP, 124 S. Ct. 872, 879



struggle. In the early 1900s, courts sided with the Patent Act. 8 Compulsory licensing was nearly unknown. But legal thinking gradually changed during the trust-busting period and the New Deal, and by 1952, the pendulum swung back to the point where courts consistently chose the "antitrust" side. A case in point is Besser Manufacturing, 9 which broke up a cartel involving concrete blocks. The Supreme Court in Besser described compulsory patent licensing as, I quote, "a well-recognized remedy where patent abuses are proved in antitrust actions and it is required for effective relief." 10 A year later, 1953, saw perhaps the most famous compulsory licensing case, General Electric, 11 where a district court broke up a light bulb cartel and required GE to issue "free" licenses to its competitors, the effect being essentially to wipe out GE's light bulb patents. The Supreme Court also endorsed compulsory licensing in the 1973 case of Glaxo Group, stating that compulsory licensing was an accepted remedy. 12 But the pendulum, as some observers call it, again began to swing in the other direction. Just a few years later, in the 1980 case of Dawson Chemical v. Rohm & Haas,13 the Supreme Court sharply criticized 8

See, e.g., E. Bement & Sons v. National Harrow Co., 186 U.S. 70, 92 (1902) ("The general rule is absolute freedom in the use or sale of rights under the patent laws of the United States. The very object of these laws is monopoly ... The fact that the conditions in the contracts keep up the monopoly or fix prices does not render them illegal."); United States v. General Elec. Co., 272 U.S. 476, 493-494 (1926) (holding that patent owner may condition a license to manufacture a product on the fixing of the first sale price of the product); see generally R. Hewitt Pate, Refusals to Deal and Intellectual Property Rights, 10 GEO. MASON L. REV. 429, 441 (2002). 9

United States v. Besser Mfg. Co., 343 U.S. at 447.




United States v. General Electric Co., 115 F. Supp. 835, 843-46 (D.N.J. 1953).


United States v. Glaxo Group Ltd., 410 U.S. 52, 64 (1973) ("Mandatory selling on specified terms and compulsory patent licensing at reasonable charges are recognized antitrust remedies."). 13

Dawson Chem. Co. v. Rohm & Haas Co., 448 U.S. 176, 215 (1980). The Court noted that "[a]lthough compulsory licensing provisions were considered for possible incorporation into the 1952 revision of the patent laws, they were dropped before the final bill was circulated." Id. at 215 n.21, citing House Committee on the Judiciary, Proposed Revision and Amendment of the Patent Laws: Preliminary


compulsory licenses, saying, "[c]ompulsory licensing is a rarity in our patent system ... Compulsory licensing of patents often has been proposed, but it has never been enacted on a broad scale." United States appellate decisions since 1980 have been almost silent on the subject of compulsory patent licensing. But I will get to that and clarify further in a little bit. The courts have dealt more frequently with copyright compulsory licensing; however, usually these cases only involve music sharing, not antitrust issues. 14 The exception, of course, is the compulsory license imposed in United States v. Microsoft Corporation. 15 That license is interesting because it involves not only a consent decree, due to the fact that Microsoft settled with the federal government, but also a fully litigated court-imposed decree, due to the fact that Microsoft did not settle with some of the states. Ultimately, the two decrees were similar in many respects. More importantly, although as yet there is no appellate decision about that remedy, the District Court strongly endorsed the compulsory licensing relief in that case. To quote the District Court, the remedy was "closely connected with the theory of liability in this case and [will] further efforts to ensure that there remain no practices likely to result in monopolization in the future ... As a result, the provisions plainly fall within the public interest."16 I will return to this later in my comments today.

Draft, 81st Cong., 2d Sess., 91 (Comm. Print 1950). The Court characterized antitrust law and patent law as equivalents in importance: "The policy of free competition runs deep in our law .... But the policy of stimulating invention that underlies the entire patent system runs no less deep." Id. at 221. 14

E.g., ABKCO Music, Inc. v. Stellar Records, Inc., 96 F.3d 60 (2d Cir. 1996); see also 17 U.S.C. § 115 (statutory compulsory license for some types of musical works). 15

Available at INK and 16

231 F. Supp.2d 144, 190 (D.D.C. 2002) (decision conditionally approving the consent decree).


So what we have in the United States is a bit of a confusing record. Before I try to make sense of this history, let me also mention the EU experience. B.

European Union

There are three high-profile compulsory licensing cases in the European Union, and as in the United States, the precedent is not a model of clarity. I will start with the Magill case involving television listings in Ireland and Northern Ireland. 17 In Magill, several local television stations had each published their own program guides. A publisher, the Magill company, wanted to publish a joint guide. The stations refused and asserted a copyright in their listings. Magill responded with a competition complaint, and the European Court of Justice eventually upheld a compulsory copyright license imposed by the European Commission in favor of Magill. The second major European case involving a compulsory license is IMS Health. In 2001 the European Commission sent a Statement of Objections to IMS, charging that IMS’s refusal to license its “brick structure” for dividing the German market into 1,860 geographical units -- used for purposes of collecting and marketing pharmaceutical sales information -- was an abuse of dominance, despite IMS's argument that the brick structure was protected under German copyright law. Shortly thereafter, the Commission adopted an "interim measure" decision ordering IMS to license the brick structure to its two competitors in Germany, pending the Commission’s final decision in the matter. 18 The Commission

17 Radio Telefis Eireann v. Commission, Cases C-241 & C-242/91P (1995), E.C.R. I-743, 4 C.M.L.R. 718 (1995), 1 C.E.C. 400 (E.C.J.). 18

NDC Health/IMS Health: Interim Measures, Case COMP D/338.044 (July 3, 2001), available at


withdrew its interim order last August. 19 The national court of Germany asked for clarification of the relevant competition law standard and, in the most recent pronouncement on this issue in Europe, on April 29, 2004, the European Court of Justice issued a reply. 20 The dECJ's decision stated that while a copyright owner's refusal to license "cannot in itself" constitute an abuse of a dominant position, such a refusal will be considered abusive if three additional factors are met: q

first, the refusal prevents the emergence of a new product for which consumer demand exists;


second, the refusal is not justified by any objective considerations; and


third, the refusal excludes competition in a "secondary market," which appears to mean a market different from the copyright owner's primary product line.21

It will be interesting to see how the German court applies this three-part test to the facts of IMS Health. Although the ECJ did not specifically say this, the decision strongly implies that dominant firms can lawfully refuse to license any competitor who would operate in the same "primary market" as the copyright owner. Such a rule would be a significant step away from the Commission's interim measure decision and would bring the state of European law much closer to where U.S. law is today. As you may be aware, many United States commentators disagreed with Magill and the Commission's interim measure decision in IMS Health. What may be less obvious is that these commentators' quarrel was not with compulsory licensing in the remedy phase of 19

See Decision, 2003/741/EC, Case COMP/D 38.044, at L 268/71, available at 20

IMS Health GmbH & Co. OHG, Case C-418/01, available at 21 Id. at ¶¶ 34, 38, 53.


each case. The quarrel was with the liability phase, which in essence involved the theory known in the United States as "unilateral refusal to deal." 22 Thus is a subtle, but important, distinction. In over a century of antitrust decisions in the United States, only one appellate case, known as the Kodak decision,23 has ever found liability even arguably on this theory, and the prevailing scholarly view in the United States is that Kodak was wrongly decided and that the unilateral refusal to deal theory is extremely difficult, if not impossible, to view as an antitrust violation. Compulsory licensing certainly exists in the United States, but it is quite separate from antitrust liability for the refusal to license intellectual property. This may amount to a significant difference between the United States and European Union experiences. United States commentators had another problem with Magill and IMS: each case involved a weak copyright in what might well have been considered uncopyrightable facts in the U.S. A major part of both decisions seems to have been the concern that the underlying intellectual property was questionable. 24 In the United States, a better way to address the copyright issue would have been to do it directly, through the copyright laws themselves, rather than through the blunt instrument of the antitrust laws. I recognize that the European Commission and courts were constrained by the absence of harmonized


In the context of patents, a unilateral refusal to deal occurs where a patent holder refuses to grant rights in the patent to a potential licensee, but does so without making this refusal conditional on conduct by the licensee and without basing the refusal on an agreement with other third parties. 23

Image Technical Services v. Eastman Kodak Co., 125 F.3d 1195, 1225 (9th Cir. 1997); cf. In Re Independent Service Organizations Litigation, 203 F.3d 1322, 1327 (Fed. Cir. 2000) ("CSU") (finding no liability for refusal to deal in a situation nearly identical to that in Kodak). 24

Cf. generally Feist Publications, Inc. v. Rural Telephone Service Co., 499 U.S. 340 (1991) (telephone listings not protected under the Copyright Act, even where a defendant engaged in wholesale copying to create a competing telephone directory).


intellectual property laws in the EU, but if I may, I would sound a note of caution. The Commission's approach could lead to unclear precedent under both bodies of law, antitrust and copyright alike. When uncertainty increases, innovation often decreases, which is exactly the opposite of what should be the long-term goal of competition law. But regardless, I believe that the Magill and IMS Health cases may provide little precedent for a future case that features undisputed software rights, for example, or strong patent rights. The third and last EU case I will mention is the Commission's recent decision against Microsoft Corporation. That case is very much a work in progress and we await the judgment of the European courts. Suffice it to say that the Antitrust Division has a number of concerns about the unbundling portion of the Commission's decision. But in terms of the interoperability remedy, which requires Microsoft to license technologies used by Microsoft server software to communicate with other Microsoft software on a network, there is considerable overlap with the United States' remedy. Like the US decree, the EC decision appears to focus on providing competing software developers with the opportunity to build products that communicate and interoperate with Windows-based PCs. Beyond that, for now, I will rely on the statements by Assistant Attorney General Hew Pate this March and April, which are available on the Antitrust Division's website. 25 What we are left with, in both the US and the EU, is a history without a great deal of consistency. How do we explain this, and particularly the fact that so many of these decisions seem to be all-or-nothing results? I submit that these binary outcomes from the past were the inevitable result of an incorrect, binary way of thinking about the antitrust and 25



patent laws -- either the intellectual property laws win, or the antitrust laws win, but never both. The Antitrust Division's view and, I submit, the correct and modern view, is that there is no conflict: antitrust law and intellectual property law serve the same goal of promoting innovation and competition, and we should bear this in mind in the remedy phase just as we do in the liability phase of a case. This brings me to a list of possible objections to compulsory licensing on both policy and practical grounds, and it also brings me to some thoughts about how the remedy can be used correctly in the future, in those limited situations where it is appropriate. III.

Policy Issues Raised by Compulsory Licensing There are important policy reasons to cause us to be cautious when considering a

compulsory licensing remedy. The most important of these is the concern that an improperly-designed compulsory license can stifle innovation. Some of the risks being taken by today's innovators are massive, with rewards systems that may be very fragile and that could potentially be destroyed by over-aggressive antitrust remedies. As an example I would offer the Iridium satellite telephone network. You may remember that this was a network of satellites shot into space, capable of bringing telephone service worldwide, like no other network, created at a massive cost. It was a stock-market darling in 1998 but unprecedented growth in plain old mobile phones drove the company into bankruptcy just a year later, in 1999. 26 If an antitrust enforcer had looked at Iridium's intellectual property in 1998, wouldn't Iridium have seemed so dominant that rivals needed a compulsory license to compete? If the stock market was so wrong, are we 26

See Iridium's stock price peaked at $70 per share in May 1998, followed by a bankruptcy filing in August 1999.


certain that antitrust enforcers or courts would have gotten it right? Or take the example of Xerox and Kodak, which were the target of compulsory license demands in the Independent Service Organization cases in 1997 and 2000. 27 Xerox was a money-losing company during this period and nearly went bankrupt in 2002. And when you need something photocopied, do you turn to your secretary and say, "Please go Kodak this for me?" I think we need to be careful not to take competitors’ or plaintiffs’ claims about so-called dominant firms at face value. Assuming that you think copier innovations are good things, or the development of satellite phone networks is a good thing, you have to permit systems that reward such innovations. The patent system does this well, when we allow it to work, so it is imperative that antitrust enforcers approach that system with some humility. We don't want to kill the goose that lays the golden egg.


Kodak, 125 F.3d at 1225; CSU, 203 F.3d at 1327.


Another, related problem is the "cost of false positives," which is a concept that combines the fear of stifling innovation with concerns over costs to both the government and the defendant. I mention this point due to its prominence in this year's Supreme Court decision in Trinko. Let me just quote Justice Scalia, writing for the unanimous Court, who in turn quoted largely from Professor Areeda: The cost of false positives counsels against an undue expansion of § 2 liability [and can] chill the very conduct the antitrust laws are designed to protect ... No court should impose a duty to deal that it cannot explain or adequately and reasonably supervise. The problem should be deemed irremedia[ble] by antitrust law when compulsory access requires the court to assume the day-to-day controls characteristic of a regulatory agency. 28 Trinko implies that courts can impose a duty to deal -- and by extension, have the power to order a compulsory license -- but should be very careful before doing so. This quote from the Trinko decision may not be as broad as it appears, since the case involved the local telephone service industry. That industry is already highly regulated in the United States. But the Supreme Court's warning leads me nicely into my last set of points, which relate to the practical difficulties of compulsory licensing. IV.

Practical Issues Raised by Compulsory Licensing A.

Benefits of the Consent Decree Approach

Drafting a compulsory license is a difficult process, even for intellectual property licensing lawyers. For antitrust enforcers, licensing is not what we would call our "core competence." In the United States, the great majority of compulsory licenses arise out of consent decrees. There are many reasons to prefer this approach, but I will mention only four: 28

Trinko,124 S.Ct. at 882, quoting Phillip E. Areeda, Essential Facilities: An Epithet in Need of Limiting Principles, 58 Antitrust L. J. 841, 853 (1989).




Input from interested parties: A consent decree results from a negotiation, which can involve the full spectrum of interested parties. This includes the government, the defendant, and competitors who may have been harmed. More heads at the table can make for a longer negotiation process, certainly, but can lead to more creative ideas.


Buy-in by interested parties: When parties negotiate a decree, the natural tendency is for them to feel some ownership and to give them an incentive to try to make it work. This can benefit enforcement.


Improved enforcement: Again, if a defendant agreed to a remedy that included a licensing program, it is much more difficult for the defendant to criticize that remedy at a later time.


Focus on future conduct: Parties negotiating a consent license focus less on past harm and more on future competition and future innovation. This clearer focus tends to reduce the chances of "false positives" or other harm to future competition, which was the Supreme Court’s concern in the Trinko decision. Considerations for Drafting the License

Whether the license is by consent or by litigated judgment, certain features will always be difficult. I will mention four of the most challenging, then I will suggest a checklist for drafting a license. First, we must define the terms of the license. The traditional definition of a compulsory patent license is: "The granting of a license by a government to use a patent without the patent-holder's permission," 29 but that simple definition is next to useless when it comes down to creating a document. Is the scope of the intellectual property, particularly if it is a patent, clear? Is the license only a negative prohibition, meaning simply that a patent owner is barred from suing for infringement? Or does the patent owner have to 29

Sara M. Ford, Compulsory Licensing Provisions under the TRIPS Agreement: Balancing Pills and Patients, 15 AM. U. INT'L L. REV. 941, 945 (2000) (discussing compulsory licensing related to AIDS medications).


provide know-how and technology support? Is this an implied license, such as a naked covenant not to sue, or is it a full written license, with normal license terms? Who gets to set field-of-use restrictions or geographic restrictions? Can the patent owner ever terminate for misconduct by the licensee? And this list does not even get into licensing boilerplate such as the term of the license, conditions for payment, accounting, or even indemnities to the licensor or licensee. Second, a compulsory license will inevitably require some oversight, making it difficult to administer. The oversight difficulty increases as you impose a royalty, and it increases even more if you impose a transfer of "know-how" and the trade secrets, or at least trade skill, needed to commercialize the patent. Third, what about future patents in the development pipeline? Some of the old cases, notably the 1953 General Electric case, imposed a compulsory license on future patents as well as already-issued patents. On the one hand, this makes sense because the value of a compulsory license would evaporate if a monopolist could come out the very next day with a new blocking patent. But on the other hand, the license could potentially harm innovation by taking away or at least reducing the incentives to invest in future research and development in that field. Fourth, is the license royalty-free, or if there are royalties, how do you set a reasonable royalty? There is precedent for doing this under patent infringement law, using what United States lawyers call the Georgia-Pacific factors. 30 But the Georgia-Pacific


Georgia-Pacific Corp. v. U.S. Plywood Corp., 318 F. Supp. 1116, 1120 (S.D.N.Y. 1970), modified on other grounds, 446 F.2d 295 (2d Cir. 1971). The 15 factors include previous royalty rates for licensing of the patent in suit, rates paid by the licensee for the use of other patents comparable to the patent in suit, and the established profitability of the product made under the patent.


factors use historical information for past infringement, whereas a compulsory license focuses on the future. As the market changes, would we periodically alter the rate, and if not, why not? A well-known commentator on IP-antitrust issues, Professor Carl Shapiro of the business school at the University of California at Berkeley, has expressed the opinion that when we impose compulsory licensing, we transform the courts into price regulators. 31 It is desirable to reduce the regulatory role, if possible. There is no one correct way to draft such a license, and each case will differ. But I propose that if you conclude that a compulsory license is necessary, ensuring that it has the following elements will address some of the most significant practical problems: q







Less is more: To minimize the policy objections and the practical concerns, an overriding goal should be to use the simplest, minimum necessary combination of transfer of rights and government oversight. Objective, verifiable criteria: Along the same lines, try as much as possible to make the benchmarks for compliance clear. If you do not, ancillary litigation is almost inevitable. Narrow scope: The coverage of a patent or copyright itself can be a matter of debate, so try to avoid that question by making the scope of the license narrow and clear as to field of use, products, or geography. Consider the question of future intellectual property, but also consider the negative consequences to innovation if future developments will automatically be licensed. Avoid "know-how" if possible: Transferring know-how requires ongoing collaboration, which increases the chances of ancillary litigation. Avoid this if you can have a successful license without it. Time-limited: There should be a sunset provision or at least a mechanism for reconsidering the license as market conditions change, particularly if future patents are involved. Clear royalty: The royalty term usually will be a "reasonable and nondiscriminatory" ("RAND") royalty. Be careful, however, to consider whether the defendant will try to make "reasonable" equate to "prohibitive." Clear status of licensees: In merger cases we often have specific, named licensees. In non-merger cases, we usually have nondiscriminatory licensing to any willing party. Consider which best fits your remedial goals, and make the


See U.S. Department of Justice & Federal Trade Commission Hearings on Intellectual Property and Antitrust, May 1, 2002 Hr’g Tr., The Strategic Use of Licensing: Is There Cause for Concern about Unilateral Refusals to Deal? at 147, at


choice as clear as possible. V.

Preferred Uses For Compulsory Licensing I hope I have explained why I think this remedy should be avoided where another,

simpler remedy is available. That said, I can identify three circumstances where compulsory licensing is particularly likely to be appropriate, although I do not mean to suggest that this is a comprehensive list. First, a compulsory license is a useful tool in a merger case. If a divestiture is required, a compulsory license can be necessary for the success of one of the divested entities. The narrowness of this license is its virtue -- the license typically would apply only to the one divested entity -- this remedies the “supreme evil of antitrust” to use the words of Trinko.32 A second circumstance is to use compulsory licensing as an alternative to divestiture. For example, the Antitrust Division applied this remedy in 1999 in the AK Steel merger case, where AK Steel sought to acquire the only other competitor that had patent rights to make a type of specialty automotive steel. 33 By a consent order, AK Steel licensed relevant patents to another competitor named Wheeling-Nisshin Inc., and thereby avoided the need to divest part of the combined entities' physical and patent assets or to block the merger completely. This type of remedy shows that compulsory licensing can serve not as a burden, but as a less-restrictive means of accomplishing a competitive result. And finally, compulsory licensing may be used in a non-merger case when other, less restrictive remedies would most likely fail to address anticompetitive conduct by a 32

124 S. Ct. at 879.




defendant. Before imposing the remedy in this type of case, we would look for an extraordinary level of market dominance and a demonstrated history of monopolization and resistance to reform. In other words, we would look for a situation where the chief objections to compulsory licenses evaporate, because monitoring the defendant's behavior has already been demonstrated to be a problem and the harm to other innovation, by other competitors, trumps the alleged harm to the defendant's innovation incentives. But even if compulsory licensing is justified in such a case, the antitrust authorities should draft the license as narrowly as possible. We should adopt the medical principle of "first, do no harm." If a defendant has placed a cancer on innovation, cut out the cancer, but be careful not to kill the patient. In conclusion, compulsory licensing presents many policy and practical issues. I believe, however, that the remedy is appropriate so long as antitrust authorities carefully consider the potential harm to innovation, and draft the license as narrowly as they reasonably can. Thanks for your time today and I look forward to the comments of my distinguished co-panelists on this important topic.