A Proposed Addition
to the Joint Status Report
The period of public comment regarding the proposed settlement between Microsoft, the Department of Justice and 9 of 18 states has recently ended. 30,000 comments were received by the Justice Department, and of those, 47 were flagged as being sufficiently "major" as to merit additional consideration.
For whatever reason, none of those 47 included comments from professional developers that favored the settlement. I have programmed professionally for the past eight years, writing software for a variety of platforms, both Microsoft and non-Microsoft. I keep abreast of the latest technologies and companies, as I make a living from what happens in those domains. I also happen to favor the settlement. I submit my "comments on the comments," therefore, in hopes of balancing the record, as well as helping to explain why consumers in general, and the majority of professional developers in particular (most of whom, logically, program for the dominant desktop platform), have disproportionately favored the Windows operating system. In the course of that explanation, I will show that those reasons have little to do with anti-competitive behavior on Microsoft's part and much to do with the nature of the market for computer software.
1. The number of comments opposing the settlement is not indicative of popular sentiment.
Of the approximately 30,000 comments sent to the Justice Department, half were against the settlement and a quarter in favor. From these comments, 47 were flagged as being sufficiently "major" as to merit consideration, only five of which came out strongly in favor of the settlement.
It is normal, however, for those opposed to a proposed action to spend the most time and effort fighting it. As Vilfredo Pareto (1848-1923), a famous economist, has noted:
"If a certain measure A is the case of a loss of one franc to each of a thousand persons, and of a thousand franc gain to one individual, the latter will expend a great deal of energy, whereas the former will resist weakly; and it is likely that, in the end, the person who is attempting to secure the thousand francs via A will be successful."
In other words, those who have most to lose from the settlement tend to protest most strongly, whereas those who gain from the settlement, albeit less profoundly but in greater numbers, protest little if any at all. Granted, those arrayed against the settlement are numerous and include large and important corporations within the computing industry. However, they are dwarfed by the thousands of third party development shops who specialize in software for the Windows platform and the billions of consumers around the world each of whom benefit greatly from the "network effects" of a large, dominant provider of desktop operating systems (explained in the following section). It is Microsoft's competitors, and the developers who program for them, who have the most to gain from a less agile Microsoft, and thus have the greatest incentive to ensure that the restrictions placed on Microsoft are as severe as possible.
2. Microsoft's dominance is market-derived, and
as such, there are economic reasons for it.
Microsoft was not created by government fiat, as was the US Post Office, through government-ordered market regulations, as was the case with AT&T. Microsoft has also not benefited from any artificial market conditions created by government or other entities. Their dominance was derived from the simple fact that consumers disproportionately favored computers which ran Operating Systems made by Microsoft. This was true even before Microsoft contemplated the integration of Internet Explorer into Windows.
Markets work because they assume that free participants in the market make better choices than centralized regulators. These choices are better because individuals have access to knowledge about particular markets which affect them directly in greater depth and detail than any disconnected regulator can hope to gather. In other words, free actors are specialists in their own respective markets, the markets which directly affect them as individuals. By letting them make their own decisions, knowledge is maximized, leading to greater efficiency.
For this reason, there must be real, substantive reasons why consumers gave Microsoft a dominant market share. That reason is the need for standards.
The market for computing products is not like other
markets. Oil from
This isn't the case with computing products. Software is not directly replaceable with a similar product made by a competitor. This isn't due to any ill will on the part of software engineers. Rather, there are a large number of ways to offer the same functionality in a particular software product.
For example, compare the manner in which user interfaces (or graphical "windows" on the screen) differ between word processing software packages. Corel WordPerfect and Microsoft Word both offer the same basic features, but organize them differently. Both packages enable users to create "tables" of data on a page, but present that capability using different interface conventions and options.
User interfaces are designed to be used by human users seated in front of a computer. Programs which reuse functionality contained in a particular software product (such as a Word document embedding an Excel spreadsheet) don't (usually) employ user interfaces. They interact with a particular software program through its Application Programming Interface, or "API." An API governs the semantics through which a particular operation is exposed to software that wants to use it. For instance, a mail program might have an operation named "sendEmail" which, as the name suggests, sends email. That operation might require certain types of information as "arguments." In the case of the "sendEmail" operation, it might require the text of the message to be sent and a list of addresses to which that message will be sent. In return, the operation might pass back to the calling software some value that the program can use to determine whether or not the message was successfully sent.
The standardization issue comes into play because there are multiple ways to expose an operation which sends email. One mail client might call its operation "NSSendEmail", and take as arguments the message of the text, a list of addresses, AND a value which indicates whether the program should encrypt the message before sending it. A similar operation on another mail client might expose the same functionality using an entirely different set of semantics. To add even more confusion, some mail software may not expose ANY API to send email. The software in question might only be designed to be used via the user interface, not from other pieces of software. This was the case with Netscape Navigator (the web browser), which for years did not expose most of its functionality via APIs. Internet Explorer, in contrast, did, leading to many third party software products which reused Internet Explorer features in their own applications.
Other incompatibilities exist, of course, such as CPU incompatibilities which cause software developed for an Intel CPU (Windows software, among others) not to work on a Motorola CPU (Apple software). However, the potential for API incompatibilities exist in every type of software, whether it is software tailored to a particular hardware configuration or software designed to run on some middle layer abstraction such as .NET or Java (which gets rid of CPU incompatibilities, but doesn't solve the problem of API incompatibility).
Given the inherent incompatibility of software product, the market historically has gravitated towards one dominant provider for a particular category of software product. This was the case with IBM, who dominated the market for computing hardware, operating systems, and software for most of the industry's history. This has been the case with Microsoft, who inherited the mantle from IBM when IBM introduced the first Personal Computer (PC), leveraging it's dominance of business hardware to popularize the new platform. At the time, they chose Microsoft to provide an operating system for the new PC, leaving Microsoft the leverage to license their OS to non-IBM hardware makers and eventually the power to define standards for desktop PCs.
Microsoft dominance has benefited consumers in a number of ways.
First, standardizing on one platform creates a larger market for software and even hardware product by unifying the market behind one development convention. The market for Windows compatible products dwarfs all other desktop-oriented markets. This expanded size leads to economics of scale, which essentially means the per unit cost of a particular product goes down the more you produce of it. This principle is the same as "buying in bulk," and explains why housing contractors pay less for building materials than regular consumers might at their local hardware store.
Similarly, this expanded market creates more room for competing software providers, who compete with each other on the common Windows platform. This lowers costs for software on that platform, a fact which should be obvious in any comparison of the cost of software for Apple computers versus the cost for Windows systems. Similarly, witness the lower cost historically of Windows PCs versus the relatively higher cost of machines that run the Mac OS, a reflection of the competition which exists for Windows-compatible hardware.
The size of the Windows software market also creates more software options for consumers due to the number of vendors attracted to the market. This creates greater consumer value, since consumers can find products which more precisely fit their needs. Furthermore, the expanded market guarantees future compatibility with new hardware and software, since the large market guarantees that vendors will expend the effort to make their products compatible. As an example, most hardware currently arriving on the market works with the Windows operating system (often first), whereas this is not the case for less popular operating systems such as the Mac OS.
Second, standardizing on Windows saves developers time and money, cost savings which are passed on to consumers. Windows, at least when compared to competing operating systems such as Unix, historically has treated the "operating system" as a large development library as opposed to a traditional "core" operating system. This means that Microsoft tends to integrate more features into the "operating system" than competitors. These features include development libraries such as component management frameworks (COM/COM+), transaction management features (MTS, now part of COM+), telephony features (TAPI), and even features such as Internet Explorer, which includes the useful HTML rendering engine.
These extra features create a common base that software developers can assume every consumer has available, for free, as a part of Windows. This obviates the need to create these complicated features themselves, allowing them to concentrate on what makes their software unique while including new Operating System features at no extra cost. Not surprisingly, Windows programs tend to be feature-rich, and new features find their way into new or updated software as reusable code which supports those features appear in the latest Windows release.
Furthermore, these standard features, as an integral part of Windows, boost developer productivity by allowing them to learn the intricacies of only one API. This is useful when multiple APIs exist which perform the same function. Any developer can learn a new API, but developers work fastest when using an API the idiosyncrasies and details of which are already well known to them.
Third, Windows market share enables Microsoft to create "de facto" standards by including technology (and new APIs) into every release of Windows, thus greatly boosting the use of a given technology and simplifying the adoption of a new technology. I use "de facto" to differentiate such standards from "international standards" created by standardization bodies such as the W3C (maker of the HTML and XML standards), IEEE and ECMA. These groups create standards which are open to everyone, allowing products to be developed that use the same API across operating systems.
International standards are an important part of software development. Without HTML, HTTP, or TCP/IP, the internet as we know it would not exist. However, international standards often take a long time to develop, and thus cannot hope to be broad enough to encompass every situation, particularly in the fast moving technology market. De facto standards fill the gap between critical, fundamental protocols of the sort standardized by international organizations and real world demands.
De facto standards take less time to develop because they are made by one company or group and do not require the lengthy peer review which goes into an international standard. They fill the void between international standards, which provide specifications for core protocols and technologies required for interoperability, and real world consumer demand, a realm profit-oriented corporations are best positioned to understand due to their direct interaction with customers. For instance, a standard for HTML has existed for many years, and is fully supported by every major web browser in existence (including Internet Explorer). This standard, however, does not define a specification for the manner in which HTML rendering "components" can be reused in applications.
Consider the API history of Internet Explorer and Netscape Navigator. Internet Explorer, until Netscape 6.0's release in the middle of 2000, was the only browser to be "componentized." By "componentized," I mean that it was the only browser a programmer could reuse in another application. For instance, a programmer could take Internet Explorer's HTML rendering component (the portion which interprets HTML to paint the user interface seen in a web browser), "embed" it into another application, and tune the HTML environment to fit her needs. That programmer could change the "right-click" menus to display new options offered by her application, programmatically navigate through the HTML "document" to find data or alter pieces of it, or change the environment of executable scripts which run inside the web page to use information and functions supplied by her application. I myself have used this feature for corporate clients, but larger companies, such as Intuit with their TaxPro software, have built their entire user interface on the Internet Explorer HTML rendering component.
De facto API standards require a wide distribution so as to make it possible for most developers to expect the presence of the API on a given platform. The gain is increased developer productivity, faster integration of new technology and more feature-rich applications. Microsoft's market share guarantees that distribution, thus yielding benefits to consumers.
Fourth, standardizing on the Windows operating system and Windows-oriented software saves companies money by making it easier to find personnel trained to use, program and maintain Windows-based systems. Most new hires have used Windows and Windows-oriented software in the past. Few enterprises expect that they have to train new users to navigate the Windows desktop. Similarly, IT departments have a giant pool from which to draw IT personnel, given the preponderance of Windows in businesses and homes around the world. In comparison, Apple-oriented companies (what few exist) have a harder time finding developers and administrators familiar with the Apple Operating System, leading to higher costs.
These, in my opinion, are the reasons that Microsoft has managed to acquire so large a proportion of the desktop operating system market. Microsoft's dominance cannot simply be blamed on anti-competitive practices. Rather, that outcome is a reflection of the realities of the software market. Real cost savings accrue to consumers from Microsoft's dominance, cost savings which would not exist were multiple OSes, each with relatively similar market share, to exist. The market made a rational choice in favoring a primary vendor of desktop operating systems, and to ignore the reasons underlying that choice can only lead to a solution which harms consumers.
3. Some comments on antitrust law
Antitrust was originally intended to ensure that no one company should acquire so large a portion of the market as to constrain choices available to consumers. It was theorized that such "monopolists" could force consumers to accept higher prices, leading to a general degradation of consumer welfare. Antitrust, in other words, was created with the aim of maximizing the consumer benefits of trade. This leads to a number of conclusions.
First, antitrust is not about individual competitors. In fact, placing too much weight upon the opinions of competitors may well lead to a situation which is LESS competitive than existed prior to a proposed "remedy" to antitrust violations. Microsoft is the primary competitor to many of the companies whose comments are included in the 47 submitted by the Department of Justice as part of the "Joint Status Report." They compete with Sun Microsystems as a server operating system, Nokia as a platform for high-end cell phones, Oracle in database management software, Palm Inc. in handheld operating systems, Sony in games consoles, AOL/Time-Warner in internet access, Instant Messaging and a number of other technologies, Novell in server operating systems, RedHat in desktop and server operating systems, and RealNetworks in the provision of streaming media formats, media servers and media players. Each of these companies would greatly prefer a Microsoft less capable of posing a competitive threat to their core business.
These companies' interests are not necessarily in line with the interests of the buying public. A remedy which relies too heavily on their preferences would be one where Microsoft is less able to compete, which would not be good for competition or consumers.
Second, antitrust is about consumer welfare first and foremost. Consumers benefit from a market where any one corporation lacks the power to reduce the choices available to them. On the other hand, in a market lacking inherent standardization between products as described in section 2, consumers also benefit from a large, dominant company with the power to define standards for the industry. The former benefit does not trump the latter benefit, and any remedy must endeavor to preserve the benefits currently derived from Microsoft's standard-setting operating system.
Third, antitrust must be interpreted within the context of
current American economic policy as a whole.
Modern policies strongly favor free markets.
In 2002, the vast majority of American economic policy is predicated upon the proven success of free markets. That modern thrust should be considered when attempting to apply 110-year old notions of antitrust.
4. Why the settlement makes sense
The settlement adheres to the principle of "do the least damage." It does not view Microsoft's dominance of desktop operating systems as problematic in itself, as well it shouldn't. Rather, it views Microsoft's market power as problematic, and does its utmost to reduce Microsoft's ability to use that power to reduce consumer choice.
That market power is expressed through Microsoft's ability to compel third parties to use Microsoft products preferentially or exclusively. To that end, Microsoft may not penalize OEMs (Original Equipment Manufacturers, such as licensees of Windows) for using non-Microsoft products by changing the licensing terms with that company (Section III.A of Proposed Settlement). Microsoft must provide a standard pricing schedule across all OEMs, leaving room for normal price differentials based on language and volume (Section III.B). Microsoft must give OEMs greater flexibility in desktop configurations, leaving OEMs free to include any third party products or middleware as part of their standard Windows installation, even middleware that is used in place of Microsoft-supplied defaults (Section III.C).
These items all deal with Microsoft's market power as opposed to Microsoft's market share, which is rightly given to Microsoft as the result of consumer choice. Likewise, it does not prevent Microsoft from including new features, however complex, as part of Windows. This is important, because Microsoft's ability to include new features is part and parcel of Microsoft's role as setter of standards in the desktop market, as described in section 2.
Next, the settlement attempts to make it easier for third-party products to replace integrated features in Windows (Section III.D and III.E). To this end, the settlement requires that Microsoft document fully all interactions between the base operating system and "Middleware products" (such as web browsers, media players, and email clients). This information is critical to third-party ability to write software which functions in the same fashion as Microsoft-supplied defaults.
In Section III.H.1, Microsoft must make it possible for desktop users to remove the icons, links, and menu options for Microsoft-supplied middleware. As some have pointed out, this does not mean that Microsoft must remove the underlying programs which normally would have been invoked by these icons. Icons on the desktop are like hyperlinks on a web page. Removing a link to CNN.com does not make the code underlying CNN.com disappear.
However, as I explained in section 2, Microsoft's ability to create API standards rests upon their ability to include a particular implementation in Windows which adheres to a particular API. As the default implementation, the feature's presence ensures that every copy of Windows (at a particular version) has at least the functionality which exists in the default. This means that the system default (the Microsoft implementation) CANNOT be removed if the Microsoft-provided feature is to act as a default. Through the mechanisms described previously, Microsoft will have to make it possible for third parties to create features which function in an identical fashion to the Microsoft default. However, if a third party implementation is removed, the Microsoft default must exist to maintain the base level of functionality.
The Microsoft supplied code is the guarantor of base level functionality. Therefore, requiring that such functionality be completely removed (as the opposing states suggest through their requirement that Microsoft sell a lower-cost, stripped down version of Windows) makes it harder to unify the development environment of a particular version of Windows. If complete removal is required, then the standardization benefits of the middleware API are erased, with all the cost increases that such an outcome entails.
Much has been made of the provisions in section III.J.1, which states:
No provision of this Final Judgment
shall: 1. Require Microsoft to document,
disclose or license to third parties: (a) portions of APIs or
Documentation or portions or layers of Communications Protocols the disclosure
of which would compromise the security of a particular installation or group of
installations of anti-piracy, anti-virus, software licensing, digital rights
management, encryption or authentication systems, including without limitation,
keys, authorization tokens or enforcement criteria.
The common complaint is that Microsoft is the sole arbiter of what is considered to "compromise the security" of the aforementioned subsystems. On the other hand, keep in mind that Microsoft, as maker of the standard desktop Operating System used by 90% of all desktop computers, has a responsibility to ensure that third parties aren't, in fact, individuals or groups aiming to compromise the security of desktop computer users. Arbitrarily releasing information that has security implications to third parties would harm consumers, and runs counter to the principle that antitrust should maximize consumer benefit.
Likewise, I am assuming that the oversight committee detailed in Section IV.A would have the right to confirm Microsoft's claims regarding the security implications of access to certain aspects of the Windows API. It is right that Microsoft should have the ability to protect sensitive information which might assist the "cracker" community in the creation of viruses or in breaking digital rights management on future music downloads. On the other hand, such claims are subject to oversight, and thus third parties have an avenue for complaints should they feel Microsoft is abusing the terms of the settlement.
The
terms of this settlement are intended to reduce Microsoft's market power vis à
vis non-Microsoft software providers and hardware vendors without reducing the benefits consumers derive from a (mostly) unified desktop operating system market. It endeavors to strike a balance between the benefits consumers derive from the Microsoft standard and the market power such an arrangement grants to Microsoft. For this reason, the proposed settlement is good for consumers and in keeping with the intent of antitrust as well as the general thrust of modern economic policy.
Sincerely,
Email: gawain24@yahoo.com
P.S. Though I live in Switzerland, I have only been here for two years. I am an American citizen, and have lived and worked in the U.S. as a programmer for most of my career.