The Wretched of the Networks

The years between Kodak and Microsoft were hardly quiet times in either IO or antitrust. Network economics and lock-in exploded into the real worlds of business and management—and in particular, into the businesses of the information sector.

Shapiro and Varian revealed that in network industries, the answer is lock-in. They told corporate managers how to recognize the network nature of their businesses, how to maximize their prospects for emerg­ing as the owners of the de facto standard, and how to maximize their profits. Of course, as good Post-Chicago economists, they also cautioned their readers not to run afoul of the antitrust laws.²⁴ That cautionary note reiterates the distinction between natural and artificial commercial advantages. There’s nothing inappropriate about a company exploiting natural barriers to charge consumers higher prices, nor is there anything wrong with exploiting natural lock-in to do the same. In fact, in some industries the ability to overcharge after securing a naturally advanta­geous position is the only way to motivate investment in product devel­opment. If the antitrust laws spared consumers from having to overspend after one firm secured a comfortable position through an appropriate combination of product quality, timing, and luck, there might not be any products for us to consume.

Lock-in in a network industry can allow an erstwhile competitive company to become a monopolist protected by a “network barrier to entry.” Tipping to a standard suggests that whatever standard emerges will be a monopoly.

A network based upon an industry open standard is likely to be a monopoly only in some senses. It’s certainly likely to restrict consumer choice. Producers rarely push technologies that compete with industry open standards. Instead, they produce competitive products that conform to the standard; competition within the standard replaces com­petition for the standard. The standard is thus a “monopoly” network, though open standards are rarely called monopolies, because they raise a different set of antitrust concerns than do true monopolies.

Proprietary standards are different. When a single, private entity con­trols a network standard, keeps at least part of it secret, and “closes” it to the rest of the world, the network barrier to entry will secure the owner’s true monopoly. In order to displace the monopolist, new com­petitors will have to produce a superior network. That challenge can be daunting—and it provides the network monopolist with ample oppor­tunity to exploit consumers.

How much exploitation is possible? In a network industry, switching imposes even more costs than it would in a nonnetwork setting. If I ever switch out of a network, I’ll lose the external value of membership—the value that I gained from all of the other members.

The incumbent monopolist has a distinct cost advantage over all would-be competitors. New entrants are unlikely to displace the incum­bent unless either switching costs are low or their new products are so far superior that they can absorb the switching costs and still be prof­itable. A monopolist faced with the first situation is in a weak position. Monopolies of simple products with low switching costs don’t really have locked-in customers to exploit. Such “weak” network properties— network growth without lock-in—may be interesting as a matter of industry dynamics, and they could still drive business strategy, but they’re unlikely to make either a firm or its investors obscenely rich. Most dot­coms fell prey to this trap; they rushed to monopolize Internet spaces without lock-in, only to discover that profits were elusive.

Monopolists in the second camp face a different challenge. Their prod­ucts exhibit the “strong” network properties of network growth coupled with lock-in. Owners of such standards can extract significant rents as long as no one produces a vastly superior product. Such superiority, however, rarely arises through a direct challenge. It usually occurs when an innovator develops a fundamentally new technology capable of dis­placing the old standard.²⁵ The incumbent wishing to protect its network monopoly must constantly monitor technological developments and either co-opt them or cut them short before the threat materializes. Owners of monopoly networks are thus wary of innovation. Their first choice is to own all innovation relevant to their networks. Their second is to control it, and to direct all third-party innovation toward their net­works. Their third is to squelch it. Network owners are happier when technology fails to progress than when it progresses in the “wrong” direction and leads to a competing network—which is, in turn, the worst thing that can happen to a network monopolist.

The first strategy is extraordinarily difficult to implement. Just ask Ma Bell, who owned everything connected to the telephone network well into the 1950s. An entrepreneurial innovator invented a small plastic cup that you could screw onto your phone’s mouthpiece. With a Hush-a- Phone in place, you could talk into the cup to gain a modicum of privacy even in a crowded office. Ma Bell went ballistic. She threatened to cut off the telephone service of any business that allowed its workers to tamper with their telephones in this outrageous manner. She claimed that Hush-a-Phone users threatened the integrity of the telephone network and thus imperiled national security. The Federal Communications Commission (FCC) agreed. Fortunately, the federal judges who heard the appeal had a bit more sense.²⁶ They reversed the FCC and allowed anyone to connect physical (i.e., not electrical) devices to a telephone handset. Hush-a-Phone was the first chink in Ma Bell’s armor. It began the erosion of her total power over the phone network. A few decades and several landmark lawsuits later, the government dissolved most of the rest of her “natural” monopoly. But in its heyday, AT&T owned the telephone network and everything connected to it—and the government was willing to enforce that monopoly. It’s hard to imagine how anyone could gain that sort of power without a government guarantee.

In the information sector, Apple tried to apply the first strategy throughout the 1980s; it didn’t work out well. Apple refused to share its hardware specifications with other software computer manufacturers, and insisted that only Apple could develop equipment compatible with the Apple network. But too much pent-up innovation existed for Apple to shut it down. IBM and Microsoft tried variants of the second strat­egy, though with notably different results; they chose to control and to direct innovation, rather than to own it. IBM published the specifica­tions for its PC architecture, and allowed all interested manufacturers to build compatible components.

The battles over network standards in strong network industries can be brutal. But even the weak network story can be brutal, particularly when it’s the subject of massive confusion—as we learned the hard way during the bubble, when confusion reigned supreme. The investor chat rooms that helped to power the bubble buzzed incessantly with misap­plications of several reputable theories—notably the “Gorilla Game,”²⁷ the “Telecosm,”²⁸ and “disruptive technologies”²⁹—each of which allegedly provided a scholarly patina to the disscussants’ simplistic investment “analyses.” The most egregiously misapplied theory, however, was none other than network economics. Shapiro and Varian had written Information Rules for managers, not for casual equity investors; the book extracted practical lessons from economics and presented them as strate­gic advice to companies in technology industries characterized by varying degrees of network effects. Lock-in was the key to many of these lessons. Companies gain flexibility, negotiating strength, and pricing power when dealing with locked-in consumers—often gaining enough to warrant offering those consumers sweetheart up-front deals to lock themselves in.

Investors who recast this sound strategic management advice as strate­gic investment advice did an astoundingly poor job. They compounded their error by misinterpreting the daily accounts of the Microsoft trial. They “learned” that network growth combined with Microsoft’s large market share to make Microsoft’s profits inevitable. Investors fell in love with the idea that networks exhibit organic growth, and concluded that inevitable profitability was widespread. All that it took was the early identification of a network industry—say, a newly opened Internet space. Because that first mover into that space would inevitably emerge as the monopolist, profits and returns would flow as a matter of course. And therein lay the critical error.

The observation of network-driven growth should mark the starting point of industry analysis, not an entire theory. Two early network econ­omists, Stan Besen and Joe Farrell, for example, had explained that a final characteristic of network markets is that history matters. Outcomes in other markets can often be explained by contemporaneous consumer preferences and producer technologies, but network market equilibria often cannot be under­stood without knowing the pattern of technology adoption in earlier periods. Because buyers want compatibility with the installed base, better products that arrive later may be unable to displace poorer, but earlier standards.³⁰

Chat room investors tended to ignore the caveat “a final characteristic,” and viewed tipping as the only important characteristic in network markets. That misconception led them to equate all network growth with monopoly rents and first movers with inevitable monopolists.

Even scholars who undoubtedly knew better tended to oversimplify, at least when addressing the public. Business School Professors Michael Cusumano (of MIT) and David Yoffie (of Harvard), for example, blurred a number of related yet distinct terms:

Web commerce also exploded from nothing in 1993 to $22 Billion in 1998, with predictions of hundreds of billions of dollars early in the next century. This rapid expansion of the network is a classic example of what economists describe as “positive feedback loops,” “increasing returns,” and “network externalities.” Behind the jargon, the dynamics are easy to follow. As more people and organ­izations connect to the Internet, more people and organizations create more tools and applications that make the Internet even more useful. And the more users, as well as tools and applications, there are, the more valuable connecting to the Internet becomes. As a result, more people start connecting, more tools and appli­cations appear, and even more people sign on, ad infinitum. The technology com­munity likes to describe this phenomenon as Metcalfe’s Law, which states that the usefulness of a network, like the Internet, grows exponentially as the number of users grows.³¹

Among nonacademics, the confusion was even more obvious. A well- written Motley Fool posting, for example, explained that

market share is important because the software industry exhibits increasing returns, a phenomenon explained by W. Brian Arthur of the Santa Fe Institute. His theory states that once a company gets a market share lead, it gets farther ahead while competitors fall farther behind. This happens because technology buyers are conservative and they demand technologies that are standard and work effectively with the rest of their infrastructure. As more copies of a leader’s software are sold, it increases the likelihood of its becoming the standard, causing even more copies of software to be sold and reinforcing the growth cycle.³²

Chat room residents viewed network growth as a sufficient condition for dominance and monopoly profits. No one much bothered to mention lock-in.³³ The investment community came to believe that the Internet— the ultimate network—had to be a network industry, and that network effects were rampant. Investors then embarked on a wholly understand­able quest for the next Microsoft.³⁴

Investors seeking the next Microsoft found positive feedback every­where. On the Internet, positive feedback meant that the more people posting information to a site (or the more products offered for sale on a site), the more people who would look to that site for the information. The more people known to be looking to a site, the more eager infor­mation providers or vendors would be to post to that site. These mutu­ally reinforcing growth trends would inevitably lead to a robust durable monopoly. These investors found a chimeric “Internet barrier to entry” that mirrored Microsoft’s highly effective applications barrier to entry. This belief in widespread network effects guided investment strategies throughout the bubble. But the circumstances surrounding Microsoft’s dominance were unique, and grounded in the peculiarities of platform software.³⁵ The misapplication of network economics, and the attempt to exploit growth without lock-in, was doomed to fail—as indeed it did.

And so, while Microsoft rose, the dot-coms sank. The lessons of IO and antitrust—playing themselves out as the information sector paraded across the front page—no longer seemed quite as abstract as they first appeared to be.