- European Union — Copyright — Licensing — Rights — Technology transfer agreements — United States
7.1 Innovation policy is about balancing open and closed systems. The distinction lies in proprietary control. Open networks allow technology creators and users to operate freely, building on what others have done without permission. In such collaborative and norm-driven environments, innovation can flourish. Closed platforms, in turn, operate under owners’ rules. The property rights underlying those systems can be indispensable to R&D, especially when new technologies require significant capital investment. By reducing free-riding, proprietary interests encourage owners to coordinate network usage and optimally to invest in their improvement. Thus, open and closed platforms can both foster innovation, but each can also depress it. For instance, when a closed system becomes dominant, its owner may charge monopoly prices and hinder third-party innovation by dictating the path of technological development. Meanwhile, open networks are vulnerable to negative externalities, leading to coordination failures and suboptimal investment. In practice, the choice is not about polar opposites of zero property rights and absolute control. It is about the optimal blend of open and closed attributes—that is, about the bundle of rights that property ownership should entail, including when property rights should end and free access should begin. Those are charged questions that draw strong sociopolitical views.
7.2 The open–closed debate goes to the heart of the antitrust–patent intersection. Joseph Schumpeter famously argued that monopoly is the best driver of innovation, while Kenneth Arrow claimed the opposite. Those conflicting ideologies, and the theoretical and empirical literature that followed, inform competition enforcement in the new economy. Indeed, the entirety of antitrust and patent law arguably boils down to a debate when closed attributes fostered by IPRs should give way to open-access principles. Some of the most thorny issues here arise because patent law approaches problems from a perspective favouring closed systems, while antitrust law’s traditional predisposition is toward open structures.
(p. 230) 7.3 Some preliminary observations are in order. First, even the most ‘open’ system is partially closed because a ‘system’ presupposes parameters that demarcate it from others. Second, the socially optimal confluence of open and closed attributes varies between firms and industries. The social optimum, however, may differ from the open-closed blend that maximizes private value. As closed platforms facilitate monetization, firms may sometimes adopt a more tightly controlled platform than would be ideal for consumers. Third, the antitrust challenge is to identify when the private incentive to create a closed platform departs from the social-welfare criterion, and then to figure out whether intervention is in the public interest. Error analysis looms large here, as it well it should. Using antitrust to override property rights is aggressive, and threatens to use competition law as a regulatory function for which it is not well suited. Type I errors may wrongly punish procompetitive innovation, which is why antitrust enforcers sometimes stay their hand. Yet, Brussels and Washington take different views. EU law is quicker to intervene, requiring dominant undertakings to open their networks when doing so is indispensable to effective and viable competition. US law imposes mandatory-sharing duties only in vanishingly few situations.
7.4 To move away from abstract discussion of closed and open systems, consider real-life examples. Apple’s business model epitomizes a successful closed system. The firm keeps the hardware, software, and design underlying its iPhone, iPod, iPad, and other revolutionary products under its exclusive control. Although its iOS app store boasts over a million applications and the company has more than nine-million registered developers, Apple oversees all apps in its store. Users of Apple mobile devices wishing to use unauthorized apps not available in the iOS app store can only do so if they unlock (or ‘jailbreak’) their devices. The result is a seamless user experience across Apple’s interoperable product line, and major but linear innovation. Still, the firm’s control-driven development of a stand-alone ecosystem—pejoratively characterized as a ‘crystal prison’ by some—draws criticism from open-source proponents and excluded competitors.
7.5 Apple’s approach contrasts with that of Google. Indeed, the Android–Apple platform war is arguably the defining battle of the new economy. It represents competing visions for the future of computing: Google v. Apple; open v. closed. Specifically, Google makes much of its Linux-based Android mobile operating system freely available under the Android Open Source Project. Apple and Microsoft keep their iOS mobile and Windows Mobile operating systems closed. That does not mean, though, that Google’s Android-based smartphones and tablets are exclusively open source. Such devices contain proprietary Google applications in search, Gmail, Google Maps, YouTube, and so on. Despite committing to open information, Google rejects open source across its entire business, which is unsurprising since it needs to monetize its innovation. Sound business judgment often requires proprietary control over a product and the network in which it operates. (p. 231) Google thus keeps the source code underlying the core of its business—Internet search and advertising (i.e. AdWords and AdSense)—under lock and key.
7.6 Going beyond the mobile example, it bears noting that IBM-based PCs began as a largely open platform based on common components. Today, although Microsoft is not a renowned open-source proponent, its desktop operating systems are more open than those of Apple. For instance, although Microsoft and Apple keep their Windows and MAC OS X products closed at the operating-system layer, Windows is open at the content and software layers, while MAC OS is not. As these examples illustrate, openness is a question of degree and some proprietary control is ubiquitous. The public-policy question is how best to tailor devices and networks to foster innovation and competition. The antitrust question is what rules, if any, should limit a firm’s right to control a device and its associated network. Above all, when—if ever—should competition law oblige a dominant firm to supply a third party or otherwise grant them access to its infrastructure?
7.7 The answer depends on empirics, but lacking hard data policymakers often resort to priors. From one viewpoint, closed networks suppress freedom, creativity, innovation, and competition.1 Others argue that closed platforms spur development, coordinate network activity, foster interoperability, enable quality control, and beget linear innovation. Offsetting considerations abound. Property rights encourage investment and drive firms to improve their products in order to outdo their rivals and to reap monopoly profits. Such incentives are critical when R&D is capital-intensive and risky, and its fruit is vulnerable to third-party appropriation. Yet, absent competition, property rights allow a firm not only to harm static efficiency via supracompetitive pricing, but potentially to suppress follow-on innovation by controlling a mandatory input to cumulative R&D or the delivery mechanisms by which firms can realize new technologies.
7.8 Patents obviously infuse a protected good, method, or system with closed attributes. Antitrust law fosters competition, sometimes in the form of an unconcentrated market structure conducive of marginal-cost pricing and choice. The antitrust–patent intersection is thus a clash between closed and open principles. Indeed, the antitrust debate on closed systems is a microcosm of the intersection of patent and competition law more generally. This chapter explores when the law requires a patentee to license its technology or otherwise accommodate competition. It explains when it makes sense to use competition rules to open up a network. Of course, open-access duties require limiting principles to protect the basic incentives underlying the free-market economy.
7.9 One can summarize the substantive law as follows: US antitrust law rarely imposes a duty to deal. Indeed, the Supreme Court in Trinko threw cold water on any forced sharing beyond the narrow situation where a property owner terminated a (p. 232) mutually beneficial course of prior dealing.2 Mandatory patent licensing is the subject of a circuit split. The Federal Circuit disclaims any requirement other than in the extreme case where patent assertion would be a sham.3 The most hospitable circuit for a refusal-to-license plaintiff is the Ninth Circuit, which rejects pretextual reliance on IPRs.4 As that holding precedes Trinko and in any event attracted much criticism, its continuing vitality is open to question.
7.10 EU competition law similarly rejects any general duty to deal, even on the part of a dominant undertaking.5 Yet, while an antitrust duty to license a patent is vanishingly rare in US law—where ‘no reported case ... has imposed antitrust liability for a unilateral refusal to sell or license a patent’6 —the same is not true of the European Union’s competition regime. EU law has required dominant firms to license their IPRs to protect viable competition, and found Article 102 TFEU violations when they have not done so.7 In its controversial Microsoft opinion, the General Court held that an undertaking abuses its dominant position if it refuses to supply an input ‘indispensable to the exercise of the activity in question’, which refusal ‘prevented the appearance of a new product’ in a manner that ‘was not justified’ and that ‘reserved to the [dominant undertakings] themselves a secondary market’.8
7.11 When should antitrust require an owner to open its proprietary system? The first possibility is always. That would entail an antitrust abolition of property rights, and would be folly. The second—never—would be far preferable, reflecting a conviction that closed systems drive investment, invention, and competition, which References(p. 233) open-access duties would undermine. Nevertheless, treating closed systems as sacrosanct arguably goes too far, and in any event conflicts with EU and US antitrust law. Thus, the answer is sometimes.
7.12 The challenge, then, is to identify when forced sharing is appropriate. A promising enquiry is whether a network owner closed a previously open system after becoming dominant—in which case antitrust scrutiny is proper—or whether the proprietary infrastructure arose under competition. In the latter case, the closed structure is presumptively efficient and the potential for error in imposing a duty to deal is greater. A useful analytic method would thus look backward to the ex ante world to focus on the competitive dynamics in which the closed network elements arose, and consider whether the market structure or strategic milieu has changed to warrant a shift from open to closed elements.
A. Closed systems usually pose no antitrust problem
7.13 The proposition that ownership rights promote economic activity is widely accepted. Indeed, that premise underlies every market economy. To say that a system is in some respect ‘closed’ is of little interest, for that attribute is ubiquitous and often essential. No major technology company has built and maintained a sustainable business model out of a purely open structure. Google rightly protects its vaunted search algorithm, as do its search competitors. Apple keeps tight control over the hardware, software, and applications comprising its entire product line. Microsoft keeps Windows closed at the operating-system level. Facebook guards the code underlying its social-media programme under wraps. Google, Yahoo, Microsoft, and other sellers of internet-search advertising refuse to disclose their ad-placement code. Pharmaceutical companies rely on patents, regulatory marketing exclusivity, and product-life-cycle management to monetize their R&D. Coca-Cola jealously guards its secret formula. This is unsurprising. Abolishing property rights over easily replicated, but expensive to invent, products leads to third-party appropriation, inadequate returns, and suppressed innovation. After all, exclusive rights can facilitate capital investment in R&D by creating a prospect. It is the raison d’être of the patent system. To require firms to open their products and technologies upon reaching a critical success threshold would compromise core economic incentives.
7.14 Proprietary rights can also facilitate transactions and spur network development. An important example in the new economy is two-sided platforms, which facilitate interactions among distinct sets of actors. Typically, they do so by overcoming barriers to efficient exchange. Manufacturers of video-game consoles, for example, bring developers and gamers together. Social-networking sites combine people with mutual interests and goals, while allowing advertisers to reach consumers whose interests align with their products. Dating websites allow mutually interested people to meet one another. Network ownership instils incentives to foster (p. 234) the platform’s value by coordinating usage, interoperability, and quality control. By aligning private and social utility, property rights encourage owners to make cost-justified investments and regulate usage.
7.15 Property rights eliminate externalities, incentivizing owners to maximize the network’s worth. Console producers ensure that titles for their platforms work seamlessly and meet minimally acceptable quality standards. Social networks prohibit certain noxious behaviour. Dating websites impose rules on their users that improve the overall quality of the service. In such cases, an absence of property rights could lead network users to act in their self-interest, but potentially against the larger interest of those using the platform. That shortcoming would be especially problematic for today’s most successful technology firms, which increasingly offer not simply a stand-alone product, but a combined experience in the form of hardware (e.g. smartphone components), software (e.g. operating system), and interoperable applications. A closed system allows its owner to build an environment, while an open system may be chaotic, raising coordination and capital-investment problems.
7.16 This is all to say that property rights in networks are often desirable and sometimes vital, which makes any per se or even general rule against proprietary control of a platform inappropriate. Indeed, since property rights are the cornerstone of capitalism, an antitrust rule banning exclusion would undo the parameters of the modern economy. Such a rule would use antitrust to override patent and proprietary rights in all cases. The open–closed debate, like others in innovation policy, entails nuance. A responsible antitrust rule must discern when mandatory sharing can undo market failures to promote the public interest, not only in the case at hand, but in future settings. The right question is which restraints within a network detract from welfare and do so in a manner that raise antitrust concerns.
7.17 We thus dispense with the threshold question whether closed platforms inherently harm competition, and consider the larger one: when does a duty to deal improve welfare? Further, when does such a duty properly arise under the antitrust laws? Guidance exists in the law and economics that guide analysis of restraints in bricks and mortar industries. Owner-imposed limits on use of its product generally raise no anticompetitive concerns, especially under US law.
B. Closed systems are presumptively procompetitive
7.18 That closed systems can promote competition does not mean that they always do so. Of course, inter-platform rivalry can promote efficiency and protect freedom of choice, even where each network owner keeps a tight rein on use of its infrastructure. Suppose that a platform in a network market reaches a critical tipping point, however, and becomes dominant as consumers flock to the winning (p. 235) system.9 ‘Path dependence’ can perpetuate a network owner’s monopoly, even over superior alternatives that lack sufficient market share to avail of network effects. There, closed networks can produce imperfect outcomes. If a single owner controls a dominant network, it will not only charge supracompetitive prices, it may dictate terms to users and exclude those seeking to advance technology in new directions.
7.19 By contrast, open systems allow innovators to contribute in a collaborative exercise free of unilateral fiat. Zero pricing can foster static efficiency and drive uptake. No one entity can dictate the path of future innovation. The result is a permission-free environment that may promote competition and choice. For some observers, open systems contrast favourably with proprietary networks, especially those that are not subject to the disciplinary pressures of competition.
7.20 As open platforms sometimes promote welfare better than closed ones do, a tailored duty to deal could theoretically improve outcomes. Further, profit-maximizing firms often voluntarily disclose otherwise-proprietary information and relinquish control over technology and platform elements. This practice suggests that open networks can themselves provide a source of competitive advantage. That is especially so in network markets where achieving scale economies is critical to industry adoption of a standard and hence to a firm’s survival. For instance, Google granted open-source access to its Android operating system in late 2007, when its share of the mobile OS market was paltry vis-à-vis those of Microsoft, Apple, and Blackberry. Today, of course, that market position has shifted in Google’s favour, though the firm’s lack of control over the platform limits its monetization. As another example, in 2014, Tesla promised not to sue any third party that wishes in good faith to use its patented technology. More generally, technology companies routinely enter into royalty-free portfolio cross-licensing deals. Further, standard-setting organizations promulgate codes available to all industry participants, and often prefer to incorporate nonproprietary technologies or those subject to royalty-free licence commitments in lieu of other patented methods or products.
7.21 Given these offsetting considerations, what are the implications for antitrust policy? First, competition law cannot solve every problem in the new economy. Antitrust policy is not a cure-all. Still, there is a case for limiting a dominant network owner’s right to foreclose competition from rival emerging networks or in downstream markets. Legal and economic principles governing restraints in bricks and mortar industries can shed light on these questions. The overarching factor, however, is a limiting principle in light of potential Type I errors.
(p. 236) 7.22 The law of vertical restraints provides the most guidance, as such restraints have parallels to closed-network issues. In traditional industries, manufacturers routinely appoint certain distributors and retailers over others. Even under EU law, which takes a less hospitable view of vertical restraints than US law, distributors and retailers have no right to carry a particular manufacturer’s goods.10 Producers sometimes impose maximum resale price restraints on retailers carrying their products. Neither EU nor US law categorically prohibits such restraints.11 In the European Union, suppliers can set up exclusive territorial distribution networks, and prevent active sales by any distributor into another distributor’s exclusive territory, so long as they do not prohibit passive sales (which would be a by-object restriction).12 Indeed, if such a supplier and distributor account for a market share no higher than 30 per cent, such an exclusive-territorial restraint falls within the Commission’s safe harbour.13
7.23 In those cases, US and EU law will examine the vertical restraint to see if it harms competition. The reason is a recognition, drawn from economics, that upstream producers have an incentive to maximize the quality, and minimize the cost, of downstream distribution of their goods. Only where retailers conspire to raise the cost of distribution, thus harming consumers and the upstream manufacturer, does antitrust economics take issue. Intrabrand restraints imposed by a supplier presumptively bolster and improve the distribution system, thus fostering greater interbrand competition. Hence, other than for hardcore offences like vertical minimum price restraints (EU law only), territorial bans on passive sales in EU law, or an upstream or downstream horizontal conspiracy, high levels of interbrand competition defeat claimed antitrust violations from a vertical restraint.14 Only when a manufacturer faces little demand- and supply-side limits on its market power will vertical restraints generally draw scrutiny, and even then there is no general reason to presuppose the restriction to be anticompetitive.
7.24 Those principles suggest that antitrust law should conduct an effects-based enquiry of network restraints imposed by platform sponsors. When a network owner limits the terms on which a user can operate on the platform, refuses to (p. 237) disclose the code underlying the network, performs quality controls, imposes restrictions to guard the system’s security such as against malware, selectively chooses partners to provide applications and follow-on services at the expense of other candidates, charges a non-zero price for network operation, imposes privacy rules, or declines to share the fruits of its innovation with third parties, no claim of a per se violation or by-object restriction should lie. Rather, the requisite analysis should identify the price and incentive effects of the challenged restraint in the relevant market, weighing any competition-limiting elements against procompetitive justifications in furthering network quality, user coordination, and interoperability. Absent a hardcore restriction, meaningful inter-platform competition should exclude an antitrust claim founded on challenging an intranetwork restraint.
7.25 Thus, the law on vertical restraints guides analysis, but the instructive value of the analogy has limits. Supplier-imposed terms on the distribution of its product typically arise in static environments where relatively few actors at separate points on the vertical supply chain play distinct, stable roles in well-delineated market segments. Platforms in high-network-effect environments, by contrast, are dynamic, populated by a large and rapidly changing universe of users. Dividing lines are often elusive, as tech firms combine numerous functions within a single device and its attendant network. Nascent threats to dominant platforms are often present, though whether they will materialize and, if so, to what extent is difficult to predict. Static pictures may be unreliable, meaning that competitive analysis of a challenged restraint over a proprietary network should focus on dynamic effects over time. Yet, predicting network evolution in the context of existing and potential competitor platforms and innovations is difficult, not least because the operative counterfactual is immeasurable.
7.26 Several reasons justify conservative antitrust enforcement in policing restraints that keep platform markets closed. First, the costs of erroneous intervention may be uniquely large in this setting, especially in high-risk and large-capital-investment platforms where undercompensation may undermine R&D. Type I errors dilute the incentives responsible for some of the world’s most valuable historic and contemporary innovations. By contrast, Type II errors—failing to open a closed system component when society should do so—may be ephemeral, especially if closed networks remain vulnerable to displacement by third-party innovation. Notably, even monopolists have reason to improve their technologies and product offerings, not only to boost consumer demand and hence profit, but to safeguard their dominant networks against disruptive technologies.
7.27 Second, error may not just be worse in the new economy, it is more likely to happen. Discerning optimal licensing terms requires information that may be impossible to acquire. The long-run social value of accessing a network depends on unpredictable events, many of which are outside the owner’s control. The (p. 238) unknown is not susceptible of expected-value calculation. A recurring question is why dominant-network owners do not always offer a licence at the monopoly—i.e. profit-maximizing—price. Sometimes, platform owners may refuse to discuss terms at all. The reason may be that no one can ascertain the ‘monopoly price’ due to uncertainty. The network owner has superior information than any enforcer, court, or prospective licensee. If the owner cannot even identify the monopoly price, how can others reliably calculate the socially optimal price? In that event, competition agencies should be reluctant to estimate social-welfare-enhancing access terms. The problem is magnified if access is sought under ‘open’—i.e. zero-price—terms, rather than under a liability rule at a third-party-determined reasonable price.
7.28 Third, rapid innovation makes it difficult for antitrust intervention to remedy market imperfections in a timely manner. New-economy markets can outpace public enforcement, making challenged restraints defunct by the time of decision and behavioural or structural remedies academic.
7.29 These considerations caution against uncritical parallels to long-established laws in static industries. Of course, reasonable minds can differ and error analysis can point in opposing directions. In the right circumstances, the law may open a platform without undermining network-management and investment incentives. The law on vertical restraints can help to identify such cases. For instance, acquiring a company that provides a critical input to one’s competitors suggests strategic exclusion rather than bona fide investment worthy of deference. Further, errors costs may be less pronounced if a vertically integrated firm chooses to stop providing an essential input that it had previously sold to its downstream rivals. In either case, a dominant firm that closes a previously open platform has a weaker claim to efficiency justification.
7.30 Of course, the analogy to vertical restraints has limits, which are themselves instructive. Consider horizontal situations where a rival wishes to use a monopolist’s technology to create a viable competing product. The case for overriding a network owner’s property right is weakest in that setting, since the party seeking open access wishes to copy the network sponsor’s product. It is exceedingly rare for antitrust to order a dominant firm to share its proprietary network or technology to facilitate the emergence of an equivalent competing network or patented product. Finally, a demand to open up a network may entail conglomerate issues. Such issues arise when a firm wants to use a proprietary good in a new application, thus to operate in a market where the owner has no presence. For instance, a proprietary technology may be susceptible of a use that the owner has not previously realized.
7.32 A firm may refuse to share its proprietary infrastructure.15 There is no general duty to deal, and one can lawfully design and maintain a closed network.16 Although the right to close a system has limits, US law skews heavily against compulsory access. It echoes a conviction that liability rules can undercompensate property owners and harm efficiency, especially in low-transaction-cost settings where private contract is feasible. A firm that fails to gain access may not wield the legal process to achieve what it could not do through the market. The law is aware of error in fashioning rules governing mandatory sharing. Property rights drive technology- and facilities-based competition. Diluting those rights by overriding owners’ decisions may encourage free-riding and stymie investment. To be sure, proprietary control centralized within a monopolist’s control can be problematic, yielding excessive prices and potentially dampening dynamic efficiency. Yet, distinguishing cases where compulsory access promotes efficiency from those where it does not is notoriously difficult. The law errs on the side of Type II errors, preferring not to intervene even where perhaps—on balance—it should so. The goal is to avoid mistakenly compromising property rights. In short, the US approach heeds the first rule of medicine: First, do no harm.
7.33 Further, courts are ill-suited to be price regulators and obligatory sharing violates deep-rooted aspects of the US sociopolitical tradition. The near-century-old Colgate doctrine, for instance, allows people to do business only with those of their choosing.17 An aversion to forced sharing also reflects the antitrust regime’s References(p. 240) limited mandate. The law forbids anticompetitive conduct, but does not require competition. The Supreme Court has thus recognized that mandatory sharing often contravenes antitrust policy.18 Emphasizing the ‘uncertain virtue of forced sharing’, it has refused to add ‘to the few existing exceptions from the proposition that there is no duty to aid competitors’.19 The result is that lawsuits premised on a unilateral refusal to deal typically flounder at the pleading stage.20 Nevertheless, a lawsuit challenging efforts to create or maintain a closed system can state a claim in the right circumstances.21
7.34 Refusal-to-deal issues recur in three contexts. First, successful goods may create secondary markets for complementary services and parts. Aftermarkets flourish, servicing products from cars to PCs, while many goods require accompanying services to satisfy consumers. Smartphones have little value, for instance, absent a telecommunications provider. The dominant seller of a primary good can sometimes limit competition in secondary markets by refusing to sell key inputs or by designing its good to be compatible only with its own parts. Competitors in secondary markets have sued for monopolization when a dominant seller tries to reserve the aftermarket for itself.
7.35 For instance, independent service providers sued Kodak in the 1990s for refusing to supply parts needed to compete in a post-sales market for camera-equipment repair.22 In its action against Microsoft, the US government challenged the software giant, in part, for designing its operating system to exclude rival internet browsers.23 More recently, a class-action lawsuit alleged that Apple’s exclusive contract with AT&T Mobility in launching the iPhone locked consumers into voice and data plans with that telecommunications provider.24 In another case, RealNetworks accused Apple of using software updates on its iPod to exclude (p. 241) RealNetworks music.25 US and EU antitrust agencies have investigated Google for alleged manipulation of its search algorithm to favour its services and products elsewhere. Each theory supposes that a monopolist may not hinder competition in secondary markets, and thus cannot maintain purely closed systems. They rest on the proposition that antitrust limits an owner’s right to wield and encumber its own property. That view inheres in an allegation that a firm excluded competition in a secondary market by designing a non-interoperable product in the primary market, or by selling the product on terms that bind consumers to it in the secondary market. As we shall see below, the law sharply circumscribes efforts by excluded rivals to demand interoperability or access to a dominant firm’s network.
7.36 Second, many joint ventures are partially or completely closed to non-participants, as when certain rivals in an industry choose a mutual standard and pool their IPRs, cross-licensing themselves for free and others at an agreed fee. Rivals not part of the club may demand access under the antitrust laws, claiming an inability to compete effectively otherwise.26 Third, network industries generate compulsory-access issues where naturally monopolistic components involving high fixed-to-marginal-cost ratios intersect with those conducive of competition. For instance, long-distance telecommunications and electricity distribution grids are natural monopolies, but wholesale and retail competition is often possible. Yet, competition can only emerge if entrants have access to distribution, which a vertically integrated utility may be loath to share. Rate-of-return or price-cap regulation was the traditional policy response in America, while Europe historically relied on nationalization. As economies on both sides of the Atlantic liberalized markets formerly thought to be natural monopolies, they have introduced various compulsory-access rules. Regulatory gaps invite antitrust questions, however, when the owner of an essential facility refuses access. Historically, US jurisprudence recognized an antitrust duty to deal in this setting, but has retreated from that position in the modern era.27
7.37 To reiterate: the basic rule today, subject to few exceptions, is that even monopolists may refuse to deal. The devil is in the details, of course, and the question is when the few exceptions to the rule apply. The last section distinguished keeping a closed system shut and closing one that was previously open. Antitrust should have little to say about the former situation, at least absent a change in industry dynamics that potentially alter the efficiency of a single-firm-controlled infrastructure. That appears to be the law. Heightened antitrust scrutiny arises where a firm closes a previously open platform, especially when that network grew into a (p. 242) dominant state by virtue of its open traits. Questions also arise in evergreening situations. For instance, when a drug comes off patent, its owner may endeavour to keep the resource under lock and key. The mechanisms used to accomplish that outcome vary, such as by contract, strategic manipulation of a regulatory structure, or further patent acquisitions, but the overarching question remains the same.
7.38 Note that analysis differs between closed systems that flow from unilateral and concerted activity. Antitrust scrutiny is sharper where horizontal rivals foreclose rivals by designing a network open only to the participants.28 For instance, joint ventures may create lucrative structures to which only the contracting parties are privy. Such concerted arrangements attract greater antitrust scrutiny. Those furthering a procompetitive rationale, of course, generally pass muster. Illustratively, firms holding essential patents for practising an industry standard do not necessarily run afoul of antitrust rules in forming a pool that licences non-members at an agreed rate.29 Still, competitive dangers are more acute when a product’s or network’s closed characteristics are not a function of unilateral design choice under competition, but of rival agreement.
7.39 We begin with the refusal-to-deal situation most likely to attract antitrust scrutiny: terminating a prior course of cooperation to deprive a rival of access to a critical network. That liability scenario finds expression in Aspen Skiing.30 This principle of law, however, by no means condemns every decision by a platform owner to close previously open access.
a. Early jurisprudence recognized a broad duty to deal
7.40 There is Supreme Court precedent for finding unlawful monopolization when a firm closes a system or designs a closed infrastructure. Perhaps the most famous case is the 1912 Terminal Railroad decision, in which competing railroads created a terminal that presented the only feasible option for crossing the Mississippi River into St. Louis.31 The terminal association, controlled by a subset of rival railroads, violated Section 1 when it refused to grant access to non-owners. Being a conspiracy case, however, it does not inform the question when antitrust imposes unilateral obligations on a network owner.
References(p. 243) 7.41 In Lorain Journal in 1951, a newspaper with a near-monopoly in local advertising violated Section 2 when it refused to contract with firms that also advertised with an emerging radio station.32 That was not a case, however, where competitors tried to access a dominant firm’s infrastructure or network. The conduct focused not on using a proprietary platform, but on influencing rivals’ access to competing platforms. Specifically, it concerned a monopolist’s effort to deprive its fledging rival of sufficient scale to become a material constraint.33 Lorain Journal, too, provides limited guidance to the issue at hand.
7.42 Two aged Supreme Court cases, however, suggest that a platform owner may have to accommodate competition by granting access to its rivals. In 1927 in Southern Photo, a monopolist stopped selling essential inputs at a discounted rate to a firm after it had acquired that plaintiff-firm’s competitors.34 In paying retail prices for those inputs, the plaintiff could not compete in the market for photographic materials. The Supreme Court affirmed a monopolization finding based on a jury determination founded on the ‘intention and desire to perpetuate a monopoly’.35 It did so, however, without much analysis. That shortcoming dilutes its force, if not precedential value, in the modern age.
7.43 Perhaps most on point is the 1973 case of Otter Tail, which concerned a utility that generated, transmitted, and distributed electricity over its network in three US states.36 Voters in four towns decided to create municipal systems for electricity distribution in lieu of Otter Tail’s distribution service. To operate a retail service, however, the municipalities needed a conduit to get electricity from their suppliers. To asphyxiate its emerging rivals, Otter Tail refused to ‘wheel’—that is, transmit—electricity over its lines from utilities and refused to sell power at wholesale. As Otter Tail’s transmission lines were the only ones available, and because the municipalities could not feasibly construct their own duplicative transmission grids, the result was a bottleneck under Otter Tail’s exclusive control. By shutting the door, and hence closing its system, Otter Tail could suppress competition at the retail level. The Supreme Court affirmed a Section 2 violation, holding that ‘Otter Tail’s refusals to sell at wholesale and to wheel were solely to prevent municipal power systems from eroding its monopolistic position’ and observing that ‘[i]nterconnection with other utilities is frequently the only solution’ to the problems faced by ‘isolated electric power systems’.37
References(p. 244) 7.44 One might argue that Otter Tail created an essential-facility doctrine, requiring owners of indispensable networks to accommodate rivals. Importantly, though, the Court was attuned to the dangers of ‘compulsory interconnection’, noting its potential to undermine the utility’s integrated system. In the case at hand, however, it thought those dangers to be remote. In construing the precedential force of the opinion, one should recall that Otter Tail was at its heart a natural-monopoly case. The problematic issues raised by the utility’s refusal to interconnect are regulatory in nature, since the economics of large fixed investment in infrastructure coupled with subsequent low-marginal-cost operation create unique challenges for industrial organization. Antitrust law today eschews broad duties to deal, in part, because doing so incentivizes companies to invest in competing networks and technologies. In the context of natural monopolies subject to subadditive production functions, however, spurring facilities-based competition may not only be impractical, it may harm efficiency. Otter Tail would not obviously extend to a situation in which a closed network is anything other than indispensable to competition.
b. The modern era rejects mandatory sharing in all but outlier cases
7.45 The last four cases suggest an incursive antitrust duty to deal. If given full expression, they might curtail network owners’ right to design closed systems or to shut out those competitors that they had previously accommodated. Yet, while the Court has not overruled Terminal Railroad, Southern Photo, Lorain Journal, and Otter Tail, they are hard to reconcile with modern jurisprudence. Whatever force a duty to deal commanded in antitrust’s early years, it has since whittled away—though perhaps not to vanishing point. The right to refuse to deal with third parties, and horizontal competitors in particular, is almost sacrosanct. In modern case law, the major—and perhaps sole—exception is where the owner of a dominant platform closes a formerly open element to foreclose a rival at short-term cost to itself.
7.46 The distinction between closing an open system and maintaining a closed one finds clear expression in US law. In 2004 in Trinko, the Supreme Court explained that its 1985 Aspen Skiing decision ‘is at or near the outer boundary of § 2 liability’.38 Aspen Skiing’s hallmark feature was that the dominant firm unilaterally terminated a profitable course of dealing with its rival. There, the owner of three of four mountains in the famous skiing resort decided no longer to cooperate with the owner of the fourth mountain, irritating consumers who valued the previously available four-mountain pass and trampling demand for its competitor’s product. In affirming an antitrust violation, the Supreme Court emphasized the owner’s failure to show that ‘its conduct was justified by any normal business purpose’, allowing the jury to conclude that it ceded short-term profits ‘because it was more References(p. 245) interested in reducing competition in the Aspen market over the long run by harming its smaller competitor’.39
7.47 Trinko subsequently cabined Aspen Skiing. It involved allegations by a competitive local exchange carrier (LEC) in New York that the incumbent LEC, Verizon, had denied interconnection services to its telecommunications network to suppress entry. Finding no duty to deal and no antitrust violation, the Court observed that Verizon had not ‘voluntarily engaged in a course of dealing with its rivals, or would ever have done so absent statutory compulsion’.40 That holding echoes the rationale explored in the last section: closing an open system raises greater antitrust concern than keeping a closed system shut. Trinko remains the law. Indeed, five years later in linkLine, the Court embraced the holding, stressing that an ‘upstream monopolist with no duty to deal is free to charge whatever wholesale price it would like’.41
7.48 Post-Trinko, a prior voluntary course of dealing appears to be a requisite of a viable refusal-to-deal claim under Section 2.42 Indeed, the Second Circuit has described the termination of a prior course of dealing as ‘the sole exception to the broad right of a firm to refuse to deal with its competitors’.43 It is hard to reconcile Otter Tail with Trinko, and it is doubtful the Supreme Court would reach the same decision in the former case today.
7.49 This law comports with the last section’s analysis. Open systems are not categorically superior to proprietary ones, any more than single-firm-controlled platforms necessarily dominate free systems. Open-access incursions can undermine incentives to invest in networks and coordinate their usage, exclusive property rights in a dominant system may allow the owner to consolidate control, suppress third-party innovation, and deny consumers freedom of choice. Still, antitrust is a blunt instrument rather than a precise scalpel, so enforcers should wield it with caution. The economy relies on property rights to allocate resources efficiently, so opening up a proprietary infrastructure ought to be the exception. Open systems can offer compelling efficiency benefits, but the law lacks reliable means to determine when closed access harms competition.
7.50 An exception can arise where a platform ascends from within a competitive milieu, based on open access and cooperation. Its popularity vis-à-vis other networks References(p. 246) suggests the efficiency of its attributes, and so the law can more confidently deem the danger of forced sharing to be less severe when the network operator later shuts the door. By contrast, closed platforms becoming dominant may require investment and coordination uniquely suited to proprietary control through a single entity. Antitrust efforts to open up such systems should be received coolly. Trinko and its progeny reflect these considerations.
7.51 Although a voluntary course of prior dealing may be a requisite of liability for an actionable refusal to deal, cooperating with one’s rival does not necessarily create a duty to keep doing so. Aspen demarcates the band of conduct in which a duty to keep a network open may arise. As Judge Posner has since explained, construing Aspen for a broader proposition would create perverse incentives.44 This facet of the law makes sense. Circumstances change, and what may have been a mutually beneficial course of dealing yesterday may cease to be so today. Hence, under Trinko and Aspen, a refusal to share must contradict the owner’s short-term self-interest other than for its value in eliminating a competitor.
7.52 Finally, de facto immunity for owners who consistently refuse to open up closed systems is the most controversial feature of Trinko and its progeny. Any bright-line rule, of course, invites error in outlier cases. To be sure, there are situations where requiring a network owner to facilitate interoperability or to grant access to a platform element would promote efficiency, even absent a prior course of dealing. The problem lies in reliably distinguishing those cases from others where open-access obligations would harm welfare.
7.53 This section explores three applications of the open-closed problem in antitrust law. The first considers the issue of predatory innovation, where a firm designs a product to be closed in the first instance—that is, not to work or interoperate with rivals’ goods. The second summarizes the law applicable to joint ventures that exclude non-participants. The third looks to the new economy for recent cases in which plaintiffs have challenged high-tech firms for designing or implementing closed systems.
7.54 The false-positive risk is greatest when the conduct challenged as anticompetitive is innovation. The danger of subduing R&D is real, and so the law receives such claims with scepticism. The typical issue is whether antitrust requires an innovator to open its product specification to accommodate rivals’ components and products. Inventors often rely on property rights, including patents, to protect References(p. 247) their new products. The difficulty in articulating an optimal rule has produced a four-way circuit split.
7.55 The Ninth Circuit holds that any improvement immunizes an innovation, as long as the inventor does not force its new technology on consumers.45 In Tyco, plaintiff health-care providers challenged a patented pulse oximetry system allegedly designed to be incompatible with generic sensors. Finding the existence of a patent illuminative, but not dispositive, on the question of an improvement, the Ninth Circuit nevertheless found that the impugned sensors added new capabilities at lower cost. As Tyco had not forced consumers to purchase its new monitors, the fact of improvement meant no Section 2 violation. It was of no moment that ‘Tyco could have made its monitors compatible with the old sensors’ because ‘a monopolist has no duty to help its competitors survive or expand when introducing an improved product design’.46
7.56 In a famous 1979 opinion in Kodak, the Second Circuit held that monopolists can permissibly maintain closed systems and further held that success in the marketplace renders a new product immune from antitrust attack as long as consumers had a choice.47 Berkey Photo, a competitor of Kodak in photofinishing services though not in selling cameras, sued after Kodak introduced a ‘Pocket Instamatic’ camera and accompanying colour film, Kodacolor II.
7.57 First, in Berkey’s view, the law required Kodak to predisclose information on that new system to enable its rivals to compete. The Second Circuit rejected that view and the idea that a monopolist had ‘“to live in a goldfish bowl,” disclosing every innovation to the world at large’.48 The court thus recognized a dominant firm’s right to maintain a closed innovation process, and held that there is no mandatory open-access rule under the antitrust laws. The Second Circuit noted that ‘a firm may normally keep its innovations secret from its rivals as long as it wishes, forcing them to catch up on the strength of their own efforts after the new product is introduced’.49 Even for a monopolist, ‘any success that it may achieve through “the process of invention and innovation” is clearly tolerated by the antitrust laws’.50
7.58 Second, Berkey challenged Kodak’s simultaneous release of the new camera and film, and advertising the new film’s relative advantages. The court refused to assess References(p. 248) whether the new film was objectively ‘superior’ to discern a possible antitrust violation. It held that, ‘[i]f a monopolist’s products gain acceptance in the market, ... it is of no importance that a judge or jury may later regard them as inferior, so long as that success was not based on any form of coercion’.51
7.59 In its landmark Microsoft decision, the DC Circuit articulated an influential—albeit not a particularly workable—test for judging the legality of a product innovation.52 One aspect of the government’s case was that Microsoft had integrated browsing software into its operating system, commingling code and excluding Internet Explorer from the Add/Remove Programs feature of Windows 98. The purported anticompetitive goal was to deter original equipment manufacturers and consumers from substituting towards alternative browsers.
7.60 The DC Circuit observed that ‘courts are properly very skeptical about claims that competition has been harmed by a dominant firm’s product design changes’.53 It applied a burden-shifting standard, such that a plaintiff must first show that an impugned design had an anticompetitive effect.54 If the monopolist asserted a procompetitive justification, then the plaintiff must either rebut it or show that the net effect of the innovation harms competition. Applying that standard, the DC Circuit found the requisite anticompetitive effect. Indeed, Microsoft could not even articulate any justification for commingling code and excluding its browser from the add/remove feature.55 Thus, it imposed liability for those aspects of technological integration. It bears emphasizing, of course, that Microsoft’s failure to muster a justification for its scrutinized innovation is the key factor explaining the outcome. The DC Circuit was at pains to disclaim any implication that the antitrust laws inhibit closed systems, even if they are designed by monopolists.
7.61 The problem with Microsoft is how to apply the burden-shifting standard when an innovator mounts a credible justification. It is not obvious how a court would quantify the harm of exclusion vis-à-vis the magnitude of a genuine improvement, weighing the two to inform a reliable calculation. Even if a jury could do so, innovators would struggle to predict how others might—in hindsight—perceive the consumer benefits and exclusionary effects of a given product–design change. The DC Circuit’s rule, fashioned in a case where it was uniquely easy to apply, is questionable from a policy perspective.
References(p. 249) (d) Modifying a product with the intent to exclude competition might be illegal
7.62 Finally, the Federal Circuit has affirmed antitrust liability based on an incompatible product design motivated by exclusionary intent.56 The circuit is a surprising outlier, given its proclivity for bolstering patentees’ rights. In C.R. Bard, a divided panel of the Federal Circuit affirmed a verdict of attempted monopolization based on a biopsy gun engineered not to work with the plaintiff’s replacement needles. The dissent argued that the improvement patent covering the redesigned biopsy gun excluded antitrust liability.57 The majority rejected that view, holding that the innovator’s anticompetitive intent supported the jury’s verdict. The court’s focus on subjective intent is questionable, which is likely why other courts have not adopted its holding. Indeed, it is not clear that C.R. Bard remains good law even within the Federal Circuit due to the Supreme Court’s subsequent decision in Trinko.
7.63 Competition can sometimes depend on access to an infrastructure controlled by a dominant firm. A recurring question is whether a monopolist can keep its essential facility closed or whether antitrust requires it to grant access to facilitate downstream competition.
7.64 Otter Tail is the classic authority for an essential-facilities doctrine.58 The Supreme Court has never explicitly recognized that doctrine, however, and its Trinko and linkLine decisions poured cold water on compulsory access under the Sherman Act.59 Nor has the doctrine fared well in the lower courts.60 At a minimum, a plaintiff challenging a refusal to grant access must show that the platform is essential to downstream competition.61 The few cases where courts have required access to a network involved a natural monopoly in the upstream facility.62 Further, an owner’s willingness to grant access may foreclose antitrust liability, if the price it charges—though unpalatable to competitors—is not equivalent to a refusal to deal.63
References(p. 250) c. Refusal to license patent rights
7.65 The mandatory licensing of patented technology is a discrete issue underlying the larger question of antitrust rules on closed systems. In short, the Federal Circuit regards a patentee’s right not to license as unqualified, save for sham litigation or other acts that independently violate the antitrust laws.64 The outlier is the Ninth Circuit, which has affirmed a Section 2 violation based on a blanket refusal to sell both patented and unpatented parts where reliance on IPRs was pretextual.65 That opinion has been the object of much criticism, however, and it may not have survived Trinko. Tellingly, the federal antitrust agencies have concluded that ‘liability for mere unconditional, unilateral refusals to license will not play a meaningful part in the interface between patent rights and antitrust protections’.66
7.66 A brief note follows on the status of patents vis-à-vis other forms of property in analysing open and closed platform issues. Although some decisions place special reliance on patents in justifying a refusal to deal,67 there seems little reason to treat patents in a categorically different fashion than other property interests, since the right to exclude defines them all. What matters is not the form of property right, but the economic context surrounding the resource potentially subject to mandatory access. If IPRs and tangible property are subject to disparate analysis, it is because the dynamic incentives surrounding certain patent rights can differ. In some cases—most obviously when a competitor seeks to copy a patented product—compulsory access in the patent sphere may be especially problematic, given the ease of copying and danger of inadequate patentee compensation. The problem of appropriation and negative externalities is not unique to the patent setting, but may be especially pronounced in it. Today’s law thus makes sense. A unilateral refusal to license a patent, absent a profitable course of dealing abruptly terminated at short-term cost to the patentee, does not violate the antitrust laws.
7.67 Antitrust’s reluctance to impose open-access obligations stems from the difficulty of regulating unilateral conduct and the dangers of over-enforcement. Different considerations apply when competitors band together to create a network, facility, or technology to which access is a requisite of competition. Acknowledging that joint ventures and patent pools are procompetitive in most cases, competition law nevertheless takes a more discerning eye to essential facilities created by agreement.
7.68 The Supreme Court’s opinions in Terminal Railroad and Associated Press establish that rivals cannot create an indispensable network and then refuse to grant access References(p. 251) to their competitors.68 The platform thus established must be a requisite of realistic competition potentially to trigger an open-access obligation. In Terminal Railroad, it was, ‘as a practical matter, impossible for any railroad company to pass through, or even enter St. Louis ... without using the facilities entirely controlled by the Terminal Company’.69 In Associated Press, a cooperative association made up of 1,200 newspapers had a by-law that prohibited any member from selling news to nonmembers and gave each participant a veto over admitting a new firm into the association. That restraint’s effect was ‘seriously to limit the opportunity of any new paper to enter’ relevant markets.70 Combined, these decisions limit horizontal competitors’ ability to create a network closed to their rivals. That such a limit exists, however, does not imply a general rule or presumption against joint ventures. To the contrary, firms can ordinarily join forces to build a network for their exclusive use.71 A refusal to grant access becomes actionable only when the facility can impede competition from non-members and thus harm efficiency.
7.69 In its modern jurisprudence, the Court has tied a closed, horizontal joint venture’s presumptive illegality to market power or ‘exclusive access to an element essential to effective competition’.72 In Northwest Wholesale, a cooperative buying agency composed of rival office supply retailers expelled a member, thus raising that firm’s costs vis-à-vis those of the enduring members. The Supreme Court held that the expulsion was not a per se illegal group boycott or concerted refusal to deal. Recognizing that such ventures often promote efficiency, the Court observed that certain rules—i.e. closed attributes—are necessary to permit the cooperative to function. Absent market power or denying access to a platform element that is a requisite of effective competition, the rule of reason governs. Lower-court decisions following Northwest Wholesale have held that ventures may lawfully exclude rivals in some circumstances.73
7.70 The economics of antitrust incursions into proprietary systems apply equally to the European competition regime. Property rules encourage investment and force competitors to invent their own technologies when the law does not permit them to free ride on others’ networks and infrastructures. Yet, a dominant undertaking’s References(p. 252) unfettered control becomes problematic when competition is not viable without access. EU competition law thus grapples with the same error-cost analysis that informs US policy. Indeed, EU policy statements note that compulsory access, even at a reasonable rate, may compromise R&D incentives and ultimately harm social welfare.74 EU law thus ostensibly arrives at the same general rule, such that there is no overriding duty to deal with one’s rivals, even for a dominant firm.75 The parallel quickly dissolves, however, because EU law imposes more aggressive interoperability and mandatory-access duties than the US antitrust regime does. Indeed, the issue of state aid to one side, refusal-to-deal jurisprudence is likely the largest substantive point of divergence between US and EU antitrust law. This section summarizes EU law on refusals to deal, which were the object of more detailed discussion in Section III.
7.71 EU law distinguishes keeping a system closed to achieve functional, efficiency, or other procompetitive benefits and doing so solely to foreclose competition. It thus views the productive and exclusionary use of assets differently. The European Commission takes a dim view of exclusion, and will look past purported reliance on property rights if the real purpose is foreclosure. Illustratively, in its 2009 enquiry into competition and innovation in the pharmaceutical sector, the agency noted various efforts by originators to evergreen their lucrative drug products. It reached a stark enforcement conclusion, which contrasts with the US approach, to the effect that ‘defensive patenting strategies that mainly focus on excluding competitors without pursuing innovative efforts and/or the refusal to grant a licence on unused patents will remain under scrutiny in particular in situations where innovation was effectively blocked’.76 Nevertheless, it remains controversial that the act of patenting—absent misrepresentations—could itself be abusive. Further, the enforcement principles espoused by the Commission as to defensive patenting strategies have yet to find backing in case law.
7.72 The law recognizes exclusion, and hence allows closed systems, because of the incentive and coordination benefits discussed above. It tolerates the negative implications of deadweight loss, fewer choices, and in some settings inhibited innovation accordingly. When an undertaking closes a system only to exclude competition or to limit technological advance, the purported benefits ring hollow and condemnation properly follows. The complicating factor, of course, is error. How to discern beneficial exclusion from the anticompetitive variant? That question is exceptionally difficult, and explains much of EU–US divergence on open (p. 253) and closed systems. The European Commission and courts are more sceptical of efficiency claims behind purported exclusion, particularly as we now discuss when an owner closes a system to firms in the upstream or downstream markets.
7.73 The analytic starting point is that EU law imposes a special responsibility on dominant undertakings not to distort competition.77 That duty limits the circumstances in which dominant firms may close a network. For instance, in the context of IPRs, acquiring patent rights to exclude rivals can be an abuse of a dominant position.78 More generally, a monopolist may run afoul of Article 102 if it discontinues a prior course of dealing. Thus, in Commercial Solvents, the CJEU held that the dominant supplier of an ingredient for the manufacture of a drug abused its position when it stopped supplying its customer because its subsidiary had entered the market to make the drug itself.79 An abuse existed because the refusal to supply risked eliminating all competition on the part of the rival that sought access.80 Further, a dominant firm may not discipline a distributor by refusing to supply the distributor until it stops carrying a competitor’s goods. Thus, in United Brands, a dominant banana producer violated Article 102 when it discontinued supplies to a distributor that supported the advertisement of a competitor.81 Thus, dominant undertakings may not ‘stop supplying a long-standing customer who abides by regular commercial practice, if the orders placed by that customer are in no way out of the ordinary’.82
7.74 Closing a platform that becomes infused with monopoly power may be problematic when the network ascended to prominence by virtue of its open nature. As explained above, that dynamic may reduce the probability of Type I errors in enforcing an open-access rule. Thus, an undertaking may improperly close a system by exercising property rights it had not previously asserted, if it does so without objective justification.83
7.75 Nevertheless, it would be a mistake to construe the CJEU’s aged judgments in Commercial Solvents and United Brands in economic terms. Concerns for fairness and open access to markets—independent of efficiency goals—drove those decisions, as they have much of the EU judiciary’s Article 102 jurisprudence. Given the ordoliberal tradition that has informed EU competition law, an undertaking that impedes access to inputs, customers, or markets undermines goals that do not materially guide contemporary US policy: hence, EU limits on a dominant firm’s right to discontinue References(p. 254) supply.84 Modern guidance from the Commission, by contrast, evidences a stronger focus on economics and consumer welfare, which may mark a directional shift in enforcement.85 Further, some case law supports an approach that respects undertakings’ right to deny rivals access to systems that have always been closed.86
7.76 On the broader issue of refusals to deal—which involve the assertion of property rights to keep a system element closed—Article 102 distinguishes a firm’s refusal to deal in a primary market from its refusal to do so in a secondary market. A monopolist can lawfully reject its competitor’s request to copy its proprietary goods. Thus, in Volvo v. Veng, the CJEU held that refusing to license a registered industrial design to a rival wishing to market that design reflected the specific subject matter of the IPR, and thus did not constitute an abuse.87
[I]n the first place, the refusal relates to a product or service indispensable to the exercise of a particular activity on a neighbouring market; in the second place, the refusal is of such a kind as to exclude any effective competition on that neighbouring market; [and] in the third place, the refusal prevents the appearance of a new product for which there is potential consumer demand.89
7.78 That standard would permit an expansive duty to deal. The qualities of ‘effective’ competition and indispensability inform the scope of the duty, and narrow the concomitant right to build and maintain a closed network. In practice, however, the Commission usually requires access only where the principles justifying non-intervention—undercompensating a property owner, moulding an effective sharing obligation, and promoting free-riding—seem attenuated.
7.79 Thus, in Magill, television stations abused their dominant positions in refusing to license a third party to create a weekly programme listings guide that neither defendant offered.90 Compulsory dealing that facilitates the emergence of a product that an IPR holder does not sell may not materially compromise core incentives to create or invest, absent some deflection of sales in the primary market. IMS Health References(p. 255) expanded the duty to deal when refusals prevent the emergence of new goods in a secondary market.91 The CJEU was careful to hold, however, that the dominant undertaking could properly refuse to license when the intended use would be ‘essentially to duplicat[e] the goods or services already offered on the secondary market by the owner of the copyright’.92 In this way, a patentee could prevent copying of its proprietary technology in a second market, but may not foreclose innovation in that downstream market by denying access to its technology.
7.80 The exception is Microsoft, which is the most controversial antitrust decision on the mandatory-access question.93 Its divisiveness flows in part from its expansive duty. First, the court relaxed two mandatory-access requirements. Post-Microsoft, a refusal need not eliminate all competition in a neighbouring market to be abusive; it need merely foreclose ‘effective’ competition. Further, the interconnected network at issue, and the dynamic innovation at play, raised more complex economic questions than those accompanying prior open-access cases like Magill and IMS Health.References(p. 256)
5 See, e.g., Case C-7/97, Oscar Bronner GmbH v. Mediaprint GmbH, 1998 E.C.R. 1-7791 (declining to find that a dominant undertaking violated Art. 102 in refusing to deal when the refused input was dispensable to competition); see also European Comm’n, Guidance on its enforcement priorities in applying Article  of the EC Treaty to abusive exclusionary conduct by undertakings, 2009 O.J. (C 45) 7, paras. 75–90 (‘When setting its enforcement priorities, the Commission starts from the position that, generally speaking, any undertaking, whether dominant or not, should have the right to choose its trading partners and to dispose freely of its property.’).
7 See Case T-201/04, Microsoft v. Comm’n, 2007 E.C.R. II-3601, paras. 312–36, 369–436, 479–620, 643–65, 688–712 (affirming Commission decision that Microsoft had abused its dominant position in refusing to supply proprietary information to facilitate interoperability and discussing antecedent case law of the EU courts establishing the circumstances in which a dominant undertaking’s refusal to deal is abusive).
9 Such winner-takes-all competition is common. IBM–Microsoft, Facebook, Google search, Blu-Ray, and other examples of systems-based competition in network markets show that one model can rapidly supplant competing alternatives, thus bestowing the winner with a powerful market position.
10 On US law, see United States v. Colgate & Co., 250 U.S. 300 (1919); see also Westman Comm’n Co. v. Hobart Int’l, 796 F.2d 1216, 1229 (10th Cir. 1986) (‘[M]anufacturers should be free to choose and terminate their distributors free of antitrust scrutiny so long as their motivation does not involve illegal price or tying arrangements’). On EU law, see, e.g., European Comm’n, Press Release Database, Antitrust: Commission adopts revised competition rules for distribution of goods and services, IP/10/445 (‘The basic principle remains that companies are free to decide how their products are distributed, provided their agreements do not contain price-fixing or other hardcore restrictions, and both manufacturer and distributor do not have more than a 30% market share.’).
11 State Oil Co. v. Khan, 522 U.S. 3 (1997) (maximum resale prices not per se illegal); European Comm’n, Notice—Guidelines on Vertical Restraints, 2010 O.J. (C 130) 1, para. 111 (‘Maximum and recommended resale prices ... are not hardcore restrictions.’).
16 See, e.g., Trinko, 540 U.S. at 407–15; Novell, Inc. v. Microsoft Corp., 731 F.3d 1064, 1072–73 (10th Cir. 2013); LiveUniverse, Inc. v. MySpace, Inc., 304 F. App’x 554, 555 (9th Cir. 2008); MetroNet Servs. Corp. v. Qwest Corp., 383 F.3d 1124, 1132 (9th Cir. 2004); In re ISO Antitrust Litig., 203 F.3d 1322, 1327-28 (Fed. Cir. 2000), cert. denied, 531 U.S. 1143 (2001); Stop & Shop Supermarket Co. v. Blue Cross & Blue Shield of R.I., 373 F.3d 57, 61-62 (1st Cir. 2004). For recent examples in the new economy, see Feitelson v. Google, Inc., No. 5:13-cv-2007, 2015 WL 740906, at *1, passim (N.D. Cal. Feb. 20, 2015) (dismissing claim that Google forced mobile-phone manufacturers to make Google the default search engine in return for pre-installing Google applications); Apple iPod iTunes Antitrust Litig., 796 F. Supp. 2d 1137, 1143–46 (N.D. Cal. 2011) (claim that Apple violated Section 2 in releasing software updates in iTunes 4.7 with redesigned anti-piracy software to prevent RealNetworks songs playing on iPod failed); Facebook, Inc. v. Power Ventures, Inc., C 08-05780 JW, 2010 WL 3291750, at *13–14 (N.D. Cal. July 20, 2010) (dismissing Section 2 claim that Facebook monopolized market by using its customers’ account information to import further contacts into Facebook and by prohibiting users from obtaining their own data); Apple, Inc. v. Psystar Corp., 586 F. Supp. 2d 1190, passim (N.D. Cal. 2008) (dismissing antitrust claims that Apple maintained monopoly in its Mac OS operating system).
17 United States v. Colgate & Co., 250 U.S. 300, 307 (1919) (antitrust law ‘does not restrict the long recognized right of [a] trader or manufacturer engaged in an entirely private business, freely to exercise his own independent discretion as to parties with whom he will deal’).
18 Trinko, 540 U.S. at 407–08 (‘Compelling ... firms to share the source of their advantage is in some tension with the underlying purpose of antitrust law, since it may lessen the incentive for the monopolist, the rival, or both to invest in those economically beneficial facilities.’).
20 See, e.g., Novell, Inc. v. Microsoft Corp., 731 F.3d 1064, 1080 n. 5 (10th Cir. 2013); LiveUniverse, Inc. v. MySpace, Inc., 304 Fed. App’x 554, 557 (9th Cir. 2009); Covad Communications Co. v. Bell Atl. Corp., 398 F.3d 666, 672–73 (D.C. Cir. 2005); ASAP Paging Inc. v. CenturyTel of San Marcos Inc., 137 Fed. App’x 694, 697–99 (5th Cir. 2005).
21 See, e.g., iPod iTunes Antitrust Litig., 796 F. Supp. 2d at 1146–47 (a fact issue whether iTunes 7.0 update that excluded certain third-party applications constituted a genuine improvement prevented summary judgment in Apple’s favour); Datel Holdings Ltd v. Microsoft Corp., 712 F. Supp. 2d 974, 985–90 (N.D. Cal. 2010) (claim that Microsoft monopolized aftermarket for accessories and add-ons to its Xbox 360 product survived a motion to dismiss because consumers may have mistakenly believed that they would ‘be free to shop in the aftermarket despite their choice in the primary market’); In re Apple & AT&TM Antitrust Litig., 596 F. Supp. 2d 1288, 1294, passim (N.D. Cal. 2008) (allegation that Apple and AT&T agreed to restrict iPhone voice and data services to AT&T after two-year service contract had conspired stated a claim under Section 2).
27 Cf., e.g., Trinko, 540 U.S. at 407–15 with Aspen Skiing Co. v. Aspen Highlands Skiing Corp., 472 U.S. 585, 610–11 (1985) and Otter Tail Power Co. v. United States, 410 U.S. 366, 377–78, 381 (1973).
28 See Trinko, 650 U.S. at 410 n. 3 (distinguishing United States v. Terminal Rr. Ass’n of St. Louis, 224 U.S. 383 (1912) because it ‘involved concerted action, which presents greater anticompetitive concerns and is amenable to a remedy that does not require judicial estimation of free-market forces: simply requiring that the outsider be granted nondiscriminatory admission to the club’ (emphasis in original)).
29 Such arrangements, however, still draw antitrust scrutiny. See, e.g., Samsung, No. 4:10-3098, Dkt. 148 (N.D. Cal. Jan. 20, 2015) (alleging that Panasonic, Toshiba, and Samsung created a pool for patents relating to SD memory cards and enforced a licence requiring all non-participants to pay a fixed 6 per cent royalty with exclusionary effect).
33 Many courts still treat such conduct as actionable anticompetitive behavior. See, e.g., United States v. Dentsply Int’l, 399 F.3d 181, 189-90 (3d Cir. 2005) (a dominant firm’s refusal to supply replacement teeth to dealers that did not exclusively carry its products violated antitrust law).
42 See, e.g., LA. Wholesale Drug Co. v. Shire LLC, 754 F.3d 128, 134–35 (2d Cir. 2014); Christy Sports, LLC v. Deer Valley Resort Co., 555 F.3d 1188, 1197 (10th Cir. 2009); LiveUniverse, Inc. v. MySpace, Inc., 304 Fed. App’x 554, 556 (9th Cir. 2008) (citing MetroNet Servs. Corp. v. Qwest Corp., 383 F.3d 1124, 1132 (9th Cir. 2004)); Covad Communications Co. v. Bell Atl. Corp., 398 F.3d 666, 673 (D.C. Cir. 2005); Covad Communications Co. v. BellSouth Corp., 374 F.3d 1044, 1049–50 (11th Cir. 2004); Int’l Ass’n of Machinists & Aero. Workers, No. 1:14-cv-530, 2015 U.S. Dist. LEXIS 5912, at *90–91 (D. Mn. Jan. 20, 2015).
47 Berkey Photo, Inc. v. Eastman Kodak Co., 603 F.2d 263 (2d Cir. 1979). For a recent example where this standard supported an antitrust violation, see New York v. Actavis PLC, 787 F.3d 638, 652–53, passim (2d Cir. 2015).
60 See Alaska Airlines, Inc. v. United Airlines, Inc., 948 F.2d 536 (9th Cir. 1991); Twin Labs., Inc. v. Weider Health & Fitness, 900 F.2d 566 (2d Cir. 1990); Olympia Equip. Leasing v. Western Union Tel. Co., 797 F.2d 370 (7th Cir. 1986), cert. denied, 480 U.S. 934 (1987).
61 See, e.g., Alaska Airlines, 948 F.3d at 545 (dismissing antitrust claim because the airline computerized reservation systems to which plaintiffs demanded access ‘did not give the[ir owners] power to eliminate competition in the downstream air transportation market’).
73 See, e.g., SCFC ILC, Inc. v. Visa USA, Inc., 36 F.3d 958, 972 n.20 (10th Cir. 1994) (‘Forcing joint ventures to open membership to all competitors ... would decrease the incentives to form joint ventures’).
74 European Comm’n, Guidance on the Commission’s enforcement priorities in applying Article [102 TFEU] to abusive exclusionary conduct by dominant undertakings (‘Art. 102 Guidance’), 2009 O.J. (C 45) 7, para. 75 (‘[T]he imposition of an obligation to supply ... —even for a fair remuneration—may undermine undertakings’ incentives to invest and innovate and thereby, possibly harm consumers.’).
75 See, e.g., Case T-504/93, Tiercé Ladbroke SA v Comm’n, 1997 E.C.R. II-0923, paras. 126–33. See generally Art. 102 Guidance, para. 75 (‘[A]ny undertaking, whether dominant or not, should have the right to choose its trading partners and to dispose freely of its property.’).
86 For instance, in Bronner v. Mediaprint, the CJEU found a dominant newspaper firm’s refusal to grant its rival access to its paper-home-delivery system was not abusive because access to the system was not indispensable. Case C-7/97, Oscar Bronner GmbH v. Mediaprint GmbH, 1998 E.C.R. 1-7791, paras. 41–47.