By David S. Evans[*]

[download pdf]

[Editor's Note: This week, we are pleased to present Part I of Professor Evan's Essay on antitrust issues in the global internet economy.  Part II of this Essay will appear next week.]

Introduction

Web-based businesses are increasingly the subject of antitrust concerns. Plaintiffs in the United States have sued eBay for tying its online payments service to its transaction service.[1]  Multiple jurisdictions in the European Community have claimed that Apple has violated the competition laws by limiting the ability of its music player to play music from competing music stores and limiting the ability of competing music players to play music purchased from its music stores.[2]  During 2007, although the U.S. Federal Trade Commission decided not to block Google's acquisition of DoubleClick after a lengthy investigation, it expressed its intent to "closely watch these markets" involved in online advertising.[3]

Of course, competition policymakers have not just discovered the web.  In 1998, shortly after the start of the commercial internet three years earlier, the U.S. Department of Justice and various states filed an antitrust case against Microsoft for engaging in various practices related to web browsers.[4]  The European Commission started an investigation of Microsoft's practices related to media players that stream music over the internet in 2001.[5]  However, the Microsoft cases mainly involved the use of the company's market power in personal computers to influence competition in web-based markets that threatened it.  The matters involving Apple, Google, and eBay concern market power in web-based products and services themselves.

The internet economy is likely to raise antitrust concerns—and possible demands for regulation—for years to come. Global gargantuan firms have emerged, which will likely attract scrutiny by competition authorities and by policymakers concerned with competition issues.  The companies mentioned above, for example, have shares in putative antitrust markets that rival those held by Microsoft.[6]  Apple has more than a 70 percent share of paid music downloads in the European Union,[7] Google has more than an 80 percent share of search queries in Europe,[8] and eBay has more than a 90 percent share of auction site page views in France, Germany, Italy, Spain, and the UK.[9]

By David S. Evans[*]

[download pdf]

II.  The Economics and Technology of Web-Based Businesses

A.  The Economics of Multi-sided Platforms

Many of the key businesses that have arisen on the web are what economists call "multi-sided platforms."[30]  A multi-sided platform provides goods or services to two or more distinct groups of customers who need each other in some way and who rely on the platform to intermediate transactions between them.[31]  Multi-sided platforms usually lower transactions costs and thereby facilitate value-creating exchanges.  They tend to arise when there is some value available from getting multiple sides together but transactions costs or other obstacles stand in the way.  eBay, for example, drastically lowered the cost of exchange between buyers and sellers of second-hand goods.

Multi-sided platforms usually perform each of three interrelated core functions to some degree.[32]  First, they serve as matchmakers to facilitate exchange by making it easier for members of each group to find each other.  That can be for love (matchmaker.com) or money (eBay).  Second, they build communities (or audiences) because this makes it more likely that members of a group will find a suitable match.  Facebook provides value in part because people are more likely to find people they want to meet and because advertisers can reach a large audience. The value of the platform grows as the audience grows.  Third, they provide shared resources and reduce the cost of providing services to multiple groups of customers. This is an especially important characteristic of software platforms discussed below.

One key feature of multi-sided platforms is the presence of the "indirect network effects" mentioned earlier.[33]  That means that the value that a customer on one side realizes from the platform increases with the number of customers on the other side.  Consumers looking to buy something value a search engine more if it provides advertisements that are more relevant to their search, while companies value advertising on a search engine higher if they are more likely to reach potential consumers.

Another key feature is that multi-sided platforms must cater to multiple, distinct customer groups simultaneously.  To establish a two-sided platform, for example, the founders must solve a chicken-and-egg problem: customers on Side A will not participate without customers on Side B, but customers on Side B will not participate without customers on Side A.  YouTube had to pursue people who want to post videos, people who want to watch videos, and advertisers who want to reach these viewers.  These features make the profit-maximizing calculus for a multi-sided platform more intricate than for a traditional business.  A firm operating one of these platforms must consider the demands of all sides, the interrelationships between these demands, the costs directly attributable to each side, and the costs of running the platform.

Further complicating this calculus is the fact that the profit-maximizing prices for multi-sided platforms can result in users on one side getting a price that is less than the incremental cost incurred by a customer on that side, and even less than zero.[34]  The side that is "needed more" or that is "harder to get" may receive a price break; conversely, the side that gets the most value out of access to members of the other side likely bears more of the cost.[35]  As an empirical matter, many multi-sided platforms make their money from one side and make access to the platform available to another side for a price that does not cover the cost of provision.[36]  Facebook, for example, is free to users and makes money by selling advertising.[37]

There are several major classes of industries in which most if not all of the businesses are based on multi-sided platforms.  These include advertising-supported media including newspapers, magazines, radio, television; payments including credit and debit cards; exchanges including auction houses, commodity exchanges and financial exchanges; and dating and matchmaking such as singles bars and matchmaking services.  Another major class consists of industries that have software platforms as their underlying technology.[38]  These include computer operating systems, mobile telephones, personal digital assistants, and video game consoles.[39]  They also include many web-based businesses.

By William H. Page[*] & Seldon J. Childers[**]

[download pdf]

[Editor's Note: This week, we are pleased to present Professor Page's and Mr. Childers's Article on the Microsoft-Samba Protocol License.  Part I appears today.  Part II of this Article will appear later this week.]

Introduction

On March 24, 2004, the European Commission (EC) held that Microsoft had abused its dominant position under Article 82 of the European Community Treaty[1] by, among other actions, refusing Sun Microsystems' request for information that Sun needed to interoperate with Windows workgroup server products[2]  The EC ordered Microsoft to disclose "complete and accurate specifications for the protocols used by Windows work group servers in order to provide file, print, and group and user administration [i.e., directory] services to Windows work group networks."[3]  On September 17, 2007, the European Court of First Instance (CFI) affirmed the EC's liability ruling and its remedial order.[4]  About a month later, Microsoft's CEO, Steve Ballmer, reached an agreement with the head of the EC's competition authority, Neelie Kroes, on the terms under which Microsoft would license the protocols.[5]  In December 2007, with the active encouragement of the EC, Microsoft reached a licensing agreement for the covered protocols with Samba, an open-source development project that produces server software that emulates the behavior of Microsoft's server operating systems.[6]  The parties have begun to implement the agreement.[7]

The Microsoft-Samba agreement is by far the most important tangible outcome of the European Microsoft case.  The EC's other remedial order in the case, which required Microsoft to create a version of Windows without Windows Media Player, was an embarrassing failure.[8]  Immediately after the Ballmer-Kroes agreement, some anticipated a similar fate for the remedial order addressing Microsoft's refusal to supply rivals in the workgroup server market.[9]  The Samba agreement, however, is significant because it requires Microsoft to provide, to its most important rival in the server market, detailed documentation of its communications protocols, under terms that allow use of the information in open-source development and distribution.  There is good reason to believe that Samba will be able to use the information to compete more effectively with Microsoft because Samba's development methods depend specifically on analysis of communications protocols.  In a closely related development, Microsoft has now published all of the covered protocols on its website.[10]  While these actions will certainly enhance interoperability, they may also facilitate cloning and thus devalue Microsoft's intellectual property. Thus, it remains unclear whether the license will enhance or inhibit dynamic, innovative competition in the long run.

In this short Article, we assess what the Microsoft-Samba license might mean, both for the market and for antitrust policy.  In doing so, we rely on published sources and on interviews with some of the key players in the negotiations.  On the Microsoft side, we spoke to David Heiner, Microsoft's lead in-house antitrust counsel, and to Craig Shank, its lead negotiator for the Samba license.  On Samba's side, we spoke to Eben Moglen, a professor at Columbia Law School, whose Software Freedom Law Center provided legal representation for Samba.  In Part I, we briefly describe the function of servers and communications protocols in computer networks. We then discuss the special significance of the Samba project in the server market.  Part II summarizes the reasoning of the EC and CFI in the workgroup server side of the European Microsoft case.  Part III describes the negotiations that produced the agreement and spells out the terms of the resulting license.  In Part IV, we consider the possible implications of the license and the disclosure process for Samba, Microsoft, and competition policy.