Parishioners, Purchasers, and Parliamentarians

Stallman spread the gospel among the hackers. Raymond evangelized the tech-savvy parts of the business community. But the ultimate battle between open-source and trade-secret-protected software will take place in the minds and wallets of consumers, and it’s not at all clear who’s been talking to them.

Ironically, it’s pretty tough for anyone outside hackerdom to under­stand open source; most attempts to navigate the literature are likely to veer quickly into issues that aren’t relevant to anyone whose interests lie beyond computer programming or software marketing. Yet, at least three other groups should be interested in the movement and its products: software users, potential investors, and government decision makers.

Users’ interests should be clear, though they’re likely to differ for dif­ferent types of users. Broadly speaking, computers today serve two pur­poses. Individuals use their desktop computers for a variety of interactive tasks such as word processing, calculating, e-mail, and Web browsing; organizations also use servers to “serve up” information or services. Because Windows dominates today’s desktop so decisively, relatively few individual consumers consider alternative platforms that would secure their membership in a competing network.

Investors’ interests may be somewhat harder to discern, but the rele­vant definition of an investor goes far beyond equity shareholders. Because platform software is a network industry, every user becomes an investor in every network that she chooses to join. Everyone running a copy of Windows on her computer has a vested interest in the ongoing maintenance, development, and support of the Windows platform and of the applications that run upon it. That means that she’s also inter­ested in seeing Microsoft survive and thrive. Everyone running a copy of Linux on her computer has an analogous interest in the maintenance, development, and support of Linux. But because Linux was born of a “movement,” or a “community,” rather than of a company, it’s harder to know how to translate that interest into practice. As a result, anyone interested in “investing” in open-source software—whether that invest­ment comes in the form of a cash outlay or simply a commitment to invest time and effort in the use of the product—needs to understand the open-source movement well enough to be confident in its survival.

Government decision makers possess both sets of interests—those common to users and those common to investors.

Such considerations tend to rise to the fore only at critical junctures in an industry’s development. The relatively recent emergence of open­source systems into powerful and important products suggests that the software industry may stand today at such a juncture. Legislators around the world have begun to inquire about the implications of this new model of software development not only to their own computing environments, but to the future of their domestic software and technology industries. Linux, Apache, and a small number of other open-source success stories thus represent far more than simply new entries into the competitive software arena, to be considered for adoption and evaluated as are all other software systems. They redefine what it means to be a “software company,” and thus provide policy makers with a rare opportunity to contemplate the direction in which they would like to move industrial policy and economic development.

And then there’s always the leviathan in the room. Because govern­ments tend to be not only software users, but also the single largest users in their jurisdictions, their decisions can help provide a way around Microsoft’s formidable applications barrier to entry. If a single very large user adopted a new platform, developers would follow—precisely because the new applications market serving the entrant would reopen niches that had already converged to a standard in the dominant plat­form’s more mature applications market.

Governments may be the only users large enough to trigger the rebirth of full-blown competition among software networks. The existence of competing networks, though, would mean more than simply a reintro­duction of consumer choice. It would also complicate the relationships among the various players in the computing and software industries— the original equipment manufacturers (OEMs) who build the hardware, the competing platform developers, the independent software vendors (ISVs) who write application programs, and consumers. Various com­panies may attempt to use copyright law, patent law, and contract law to manipulate the market to their advantage. Government decision makers can also influence which of these attempts succeed and which ones fail.

Government decision makers thus face a tough challenge. They have to consider all of the issues that drive consumer software purchases, including costs, service, reliability, and product quality.²⁷ But they also face a unique set of concerns relating to issues like national security and economic development. From a national security perspective, the open­ness or secrecy of source code, particularly for critical infrastructure software like a platform, can affect a cracker’s ability to wreak havoc with national infrastructure. Furthermore, some governments might feel uncomfortable trusting their critical infrastructure to a private company’s secrets—particularly when the company is foreign.

Beyond security, different countries have different concerns.

China is hardly alone. Legislators across Europe, Asia, and Latin America reportedly have introduced open-source legislation—though most have been little other than posturing by minority parties or back­benchers trying to grab headlines. These bills have run the full gamut, from truly intrusive command-and-control to more benign statements of preference. In some cases, for example, legislators have tried to ban proprietary software from government computers—a horrible idea that would block government’s access to powerful applications for which no viable open-source alternative exists. Others have advocated requiring open-source platforms on all government computers, with no preference among applications—showing a bit more pragmatism and flexibility, but hardly enough. The most reasonable of them—and the only idea that could work without threatening to cripple government functioning— simply express a preference for open-source where available, and require procurement agents to justify selecting proprietary software over an open-source competitor.

Procurement legislation of even this measured sort is rather unusual. By and large, commercial considerations should guide government’s com­mercial decisions. The elevation of social concerns risks causing more harm than good by hampering government efficiency. At the same time though, most people recognize that governments should consider indus­trial and economic development. Governments at all levels representing economies at all stages of development need to ensure that they’ve created environments that foster sustainable education, training, inno­vation, jobs, and commerce.

We’ve already missed the appropriate arena for debating software policy. The proprietary model of software development is only possible because IP law allows developers to protect their source code as a trade secret while simultaneously circulating their copyrighted object code. In other words, we adopted an industrial policy that enabled some business models that otherwise wouldn’t have been viable. Unfortunately, we didn’t debate this policy before we adopted it; we neglected to weigh its benefits against its costs, evaluate its impact on innovation, or project its contributions to scientific knowledge. No U.S. Congress—a body charged with this analysis—ever considered triple protection explicitly. And American courts are required to presume that Congress actually reached a reasoned conclusion, despite knowing that such a presump­tion is entirely fictional. We thus made a critical policy decision without ever evaluating either its merits or its implications—and then foisted it upon the rest of the world, where it may cause even more harm than it does in the United States. While procurement is hardly a good place to make industrial policy, for many governments it may be the only place left.

Our “decision” to award triple protection had a particularly adverse effect on the market for platform software. Platform software is an unusual product for a number of reasons. First, it’s a network product prone to lock-in, high switching costs, and tipping to a standard. Second, it’s a critical component of the public’s information infrastructure. Third, a single private company owns its de facto standard. Fourth, that private owner operates the standard as a proprietary product. Fifth, the stan­dard and its private owner are dominant worldwide.

This combination is unique. For the most part, those relatively few products that both tip to a network standard and play critical infra­structure roles are subject to some form of government control—either through direct government ownership or through regulations restricting the behavior of a “natural monopolist.” These monopolists generally do pretty well for themselves and for their investors—though not always quite as well for their customers. But they do recognize that part of the price of being a monopolist is that they need to reserve that guest room for the market cops; we often designate entire regulatory agencies to police them.

The information sector’s growth turned platform software into a crit­ical infrastructure product. But for the first time, the provider is a private company unconstrained by either market forces or government regula­tion. Microsoft can set its prices wherever it wants, include whatever terms it wants in its contracts or licenses, and leverage its dominance in platform software into any middleware markets that it chooses. If anyone has a problem with Microsoft’s moves, they can call in the market cops—but we’ve already seen how well that works. Technology gave Microsoft the ability to integrate new products into its platform, IP law protected its distribution channels, contract law permitted its licenses, and its economic incentives drove it to squelch all innovation that it couldn’t direct through Windows.

Our inattention to industrial, software, and IP policy thus gave Microsoft an unregulated infrastructure monopoly, the ability to channel innovation in directions likely to serve its monopoly, and broad discre­tion about the directions in which its monopoly grows. Software should evolve as innovation above the platform diverges in competing direc­tions, and eventually converge to a de facto standard when competing vendors copy each other’s successful innovations. When convergence leads to stable, widely understood technology, developers can integrate the application safely into the platform to begin its slow journey down the translation chain. But we’ve given Microsoft the power to tamper with that evolution; the platform now evolves how, where, and when Microsoft wants it to evolve. And as Microsoft directs the platform’s evolution, its creeping infrastructure monopoly becomes increasingly central to our lives. As far as I can tell, no private company has ever had this level of power over a product that is so central, while receiving so little government oversight.

Could open-source procurement legislation circumvent the applica­tions barrier to entry? And even more to the point, should it try? Windows is a classic cathedral program. And though it might never have become dominant if we had a functioning bazaar, our information infra­structure wasn’t quite capable of generating emergent systems as complex as a platform when Microsoft launched and nurtured Windows. In a curious way, Windows may be a victim of its own success. Its top­down construction helped create an infrastructure that enabled open source to match its raw power, while also embodying a number of other attributes that better serve the public’s interests. The looming clash between these approaches to infrastructure could thus have a profound effect on more than just the information sector; it could help shape the future of economic and industrial development.

Governments worldwide will have to decide whether they want indus­trial policies that favor worshippers or shoppers—or policies that are indifferent between them. Existing policies play strongly in favor of cathedrals; if unchanged, we may all end up worshipping whether we want to or not. The debate over procurement legislation raises the pos­sibility of leveling the playing field. Good legislation could give us a choice. Open-source platforms would lead to an open standard for soft­ware development, and force future developers to compete strictly on the merits of their products, not on their ability to leverage strategically from one software product market into another. That sort of open standard for infrastructure would lead to a more neutral, market-driven evolution of the rest of the software industry. Innovators and entrepreneurs could focus on software features enabled by technology, rather than on those motivated by Microsoft’s business strategies. But bad legislation could destroy what we already have; it could force us into the bazaar even when we’d prefer to be lining the pews. Public policy may thus prove to be as important to the information sector’s maturation as it was to its birth. And as long as we remain unwilling to revisit our default decisions about software IP rights, procurement may be the only arena in which policy movement is possible.