Enabling Innovation from the Edges
As it developed, the Internet would have seemed ill suited for a commercial life later. The NSF managed an Internet that lacked market-oriented focusing devices and/or economic inducement mechanisms—namely, a setting that induces market actors to direct efforts toward the most valuable innovative outcomes. There were contracts for carrier services between government buyers and commercial suppliers, for example, but no general market orientation toward the pricing of the exchange of traffic between carriers. The applications were not for sale, nor did anybody design the applications with mass-market use in mind. They developed inside research laboratories that met the needs of a niche researcher and student community. This community desired technically advanced functionality and tolerated considerable imperfections to achieve those advances. That hardly seemed like the propitious beginning for a new transformative mass-market communications network.
Looking back on these events through a wide lens, it is now possible to ask and answer one of the core questions of this book: why did the privatization of the Internet—turning into a private asset a network designed to suit the needs of researchers and students—unleash a wide and profound set of economic outcomes as it grew into a widely used commercial network? How and why did the commercialization of the Internet bring about structural change? Why was this structural change coincident with an economic boom? The answer involves innovation from the edges—multiple perspectives originating from multiple places in an industry with little or no concentrated decision making. Innovation from the edges played an important role in all the key events:
• Multiple perspectives. The commercialization was shaped by a diversity of viewpoints. The design of the Internet met the needs of several different groups who funded it and who participated in its design. The commercialization of the network enabled participation from entrepreneurs with a variety of perceptions about the value of the economic opportunity, and with a variety of perceptions about how best to build businesses to address those opportunities. The opportunities also were larger than any single firm could pursue or would have pursued, so a variety of approaches thrived for a sustained time period.
• Originating from multiple places. Resources from many locations were directed toward building the commercial Internet. These were researchers in laboratories, administrators in a variety of organizations, students at universities, engineers at equipment-making firms, and suppliers of complementary services, to name a few. A set of institutions was adopted to encourage inventions arising from many participants—both before and after privatization of the Internet. The commercial network also accommodated applications from multiple participants in many locations. Especially after privatization and the invention of the web, commercial efforts extended the scope of the network far away from what any single firm would have pursued had it performed R&D by itself.
• Lack of concentrated decision-making power. A number of factors enabled dispersed decision making. Authority for technical improvements was not centralized within one firm or organization. No single entity or decision maker coordinated the Internet for an extended time with a single economic interest as motive. Even the standardization committees developed unrestricted access to their designs and did not restrict their uses. These same committees also maintained processes that did not restrict who could contribute additional improvements. Especially after privatization, competitive markets fostered commercial activities by more than one firm, and virtually every opportunity, from the largest to many small applications, received attention from many suppliers.
This book also has highlighted the many ways in which innovation from the edges encouraged exploratory activity. Demand for new value created different opinions about the appropriate actions to take. Not every participant perceived the value of the opportunities in the same terms, or possessed the same set of assets for exploiting the opportunities. Decentralized decision making was better at permitting a variety of experimentation than a central decision-making process. There were minimal barriers coming from the limitations affiliated with departing from the prevailing view, competing with an established firm, or approaching an opportunity that others had not perceived. All these factors reinforced one another, and the sum total of all this market activity addressed uncertainty about value by innovating much faster and with greater success than any single organization ever could have.
More concretely, innovation from the edges helps tell the economic story of how the Internet got built and why. It helps explain how the commercial Internet had an unexpectedly large impact as it became widely available and adopted as a commercial service. That explanation addresses one of this book’s motivating puzzles: “Unexpected” and “large impact” and “innovative” does not usually happen at the same time. Here is a summary of the answer:
The economic explanation starts with the privatization of the network. As chapters 2, 3, and 4 described, prior to privatization, the creators of the Internet and web had sufficient time to refine their inventions, accumulate suggestions from many corners, and routinize processes for delivering services to users. Institutional support for the Internet was also close to routine, as the IETF was a functioning organization, designed to enable the accumulation and evolution of new functionality. The World Wide Web Consortium, though a new organization at the outset of privatization, governed a set of technologies that had already been tested and refined in a wide set of circumstances. As described in chapter 7, neither the IETF nor the World Wide Web Consortium restricted how technology could be used or who could use it. Both organizations eventually made all information available to any taker—albeit, with early access given to participants at the IETF, or to consortium members at the W3C. That helped to support a wide potential breadth of businesses after privatization, and many of these came from edges.
Very quickly after privatization, participants in commercial markets found new opportunities for the Internet and web. As chapters 4 and 6 stressed, many of these opportunities came as a surprise, and it caused a rush of economic activity. As chapter 6 explained, large parts of the economy would benefit from the commercialization of the Internet and the web, and had not acted in advance of its deployment. In part, this was due to the origins of the Internet and web, as a working prototype inside of universities and laboratories, one generated outside of commercial markets. In part, this was due to the behavior of the leading firms in communications, who had focused their attention elsewhere, and in part, as chapter 7 explained, the value chain behind the commercial Internet and web did not resemble existing value chains, and established players did not recognize how it would behave. Moreover, as chapter 5 analyzed in detail, in part, this was because bulletin board systems had helped grow firms who could enter as ISPs, and these firms held a special status within the industry as outsiders. Soon the opportunities extended to households and especially to businesses, touching virtually any process in which computing played an important role. At first this was not recognized, and BBSs were the best-placed private actors to see the opportunity.
Once the prototypes for the entire system were demonstrated, especially as embodied in the operations of a Netscape browser, innovative firms from many parts of the economy perceived many areas in which the Internet and web could apply—back-office enterprise information technology, customer-facing information technology, and operations-enhancing enterprise computing. That reaction occurred at households trying web-based applications, ISPs and access providers selling subscriptions and data services, entrepreneurial entrants growing web-based businesses, established firms modifying existing operations, and more firms with perceptions from the edges. The rush of entry into supplying web-based applications reflected a real economic opportunity for many entrepreneurs and established firms, simultaneously perceived by many of them. Contemporaries called that a gold rush, and, indeed, as chapter 6 explained, initially it resembled one.
The boom in investment and entrepreneurial entry persisted; many reasons account for that persistence, and these were distinct from the factors that started the gold rush. As chapter 7 stressed, the commercial Internet and web had a supporting value chain behind it, and those institutions had been built over a long time and would sustain more than just a transition of the technology out of the university and into commercial use. Those institutions possessed all the attributes affiliated with a successful commercial value chain. That enabled the support of value creation over a long period. One additional feature of the experience sprung from the openness of value chain: it permitted radical change to reach the market when it otherwise might have encountered roadblocks at private firms.
Unlike the initial rush, however, this radical change came about due to more entrants than just the earliest entrants that made discoveries of value. Chapters 8, 9, and 10 analyzed how that new opportunity provided potential value to a distributed set of participants. It provided new value in a dispersed set of circumstances across the economy—financial transactions, retailing and wholesaling, and logistics. That was a large-scale commercial opportunity, and it generated a concomitant scale of investment, again, coming from many edges of the economy, as well as from its core established firms.
Each participant’s actions simultaneously raised the value of the other, and positively reinforced the economic incentive of the other to continue to invest. That was a symptom of the presence of a network effect at an economy-wide level, and by itself, would have generated a greater scale of investment. Moreover, in many settings the activities of one set of participants taught others lessons that they could use, so the simultaneous activities of so much experimentation at a large scale also increased the extent of learning. The breadth and number of opportunities in the late 1990s extended across virtually every sector of the economy, which meant that every important actor in the economy was investing in the Internet at virtually the same time. That placed enormous strains on the ability of firms and labor to supply inputs and services. The symptoms of impatience were everywhere and manifested themselves as resource shortages in IT labor markets and entrepreneurial managerial markets.
Said succinctly, the factors that started the boom differed from the factors that sustained it. A network effect manifested as a virtuous cycle in the late 1990s. That it is distinct from a gold rush—namely, the factors that catalyzed investment in the middle of the 1990s.
All the impatient activity encouraged another important feature of the economic situation, the seemingly high tolerance for economic experiments, many of these coming from the edges. Part of that tolerance was unsurprising in light of the scale of the new opportunity and in light of the background of many of the entrepreneurs—many had come from computing, where technological races were common. The late 1990s were extraordinary because the experimentation involved such an extensive list of participants and continued for so long. Many questions remained open about the value of different aspects of the Internet in the long run. Firms that initially explored approaches to creating value immediately after privatization continued to have incentive to explore new opportunities to create value several years later. Technically oriented entrepreneurialism thrived, and perceptions that had been regarded as from the edges were taken seriously by firms that had previously ignored such views. It made for strange bedfellows. Firms that had little economic relationship to one another prior to the Internet’s deployment found themselves making deals and basing growth projections on those deals. Entrepreneurs began to talk with each of these firms, and established firms had to assign managers to follow developments that they had previously ignored to make thorough assessments.
The contrast between chapters 8 and 11 illustrates another lesson. These chapters highlighted how two established firms, IBM and Microsoft, profited from the commercial Internet and web. IBM eventually sold more services as a technological intermediary, while the commercial Internet and web raised the value of the PC value chain, described in chapter 7, and into which Microsoft sold its products. Their similarities ended there. As chapter 11 described, Microsoft also sought to control the value chain that emerged and cement its exclusive position. That led it to put up resistance to the emergence of alternative platforms, and it led to one of the dramatic confrontations of the era, between platforms built around Netscape’s browser and Microsoft’s.
For a time, especially near the end of the boom in growth, Internet exceptionalism became the prevailing view, a point stressed in chapter 12. This philosophy defied the economic archetypes of business history and boldly rejected traditional approaches to building and valuing businesses. This view partly arose as a product of the impatience of the era and the confusing nature of the new opportunities. Yet that alone does not explain how this perspective overwhelmed the voices of most experienced entrepreneurs and investors and analysts, such as those who had experienced technology-led prior booms in PCs and client-server networks. Internet exceptionalism was reinforced by a financial sector that reaped enormous short-term profits from encouraging the activity consistent with Internet exceptionalism, and because those with a skeptical view could not gain influential positions. Many participants overinvested or invested in directions whose merits eventually proved fleeting, and for some participants, the ideology of Internet exceptionalism lost creditability only long after the end of the boom.
The narrative of the Internet’s evolution did not end with the dot-com crash. As chapters 13 and 14 describe, in, respectively, search engines and wireless local Internet access, the factors that shaped the next wave of investment would resemble those that shaped the first wave, while also differing from it. Further exploration took on an air of renewal in light of the lessons of prior experience. Thus the origins of the Internet, its privatization, and it diffusion into private use, connected to the next wave of investment by commercial actors. As it turned out, the creation of value from search engines built on the interplay between advances made in a university by a couple of graduate students and its commercialization, while the advances made in wireless Internet access resulted from the interplay among government policy for spectrum, the design of a standardization committee, and the aspirations of many private firms. As with prior episodes of commercialization of innovation, markets remained open to a variety of perspectives and contained institutions that permitted value to migrate to its highest value.
It added up to a remarkable set of events, one of the greatest episodes of technologically led growth in the two-hundred-year history of American capitalism. The growth and deployment of the Internet was unexpected, innovative, and impactful. It is not an exaggeration to say the deployment of the commercial Internet changed the way everyone lives, works, and plays, changing US economic activity in irreversible ways. Most astonishing of all, it brought about these changes at an extraordinarily fast pace.
Economic Archetypes
If Internet exceptionalism does not provide a valid explanation for events, then what should replace it? This book has argued that many well-known economic archetypes shaped the emergence of innovation from the edges. The rate, direction, and contour of events could not be understood without placing commercialization at the center of the explanation. Moreover, the economic forces that did matter for the Internet were not unique to the Internet. All of these archetypes had appeared in other technology markets, and, more specifically, other computing and communications markets.
A list of many key economic archetypes is provided below. No single economic archetype was solely responsible for the economic experience affiliated with the deployment of the privatization of the Internet, or with the experience with commercialization. A summary of important economic archetypes can illustrate the point:
• Competitive core encouraged competitive complementary markets. Backbone markets were competitive because, as explained in chapter 3, the privatization of the NSF backbone created competitive conditions. Chapters 5 and 8 discussed how competitive backbone markets helped grow competitive complementary markets in access for complementary infrastructure markets, such as ISPs. Those conditions enabled quick and low-cost entry of ISPs in many locations, which facilitated creating a large national network to support commercial applications.
• A killer app could be a catalyst for adoption and investment. The application with the highest value for the early Internet was e-mail, a point made repeatedly in chapters 2 and 3. The most valuable application for the privatized Internet was the web browser, a point first raised in chapter 4 and followed in many later chapters. The demonstration of the browser-based Internet motivated additional adoption by businesses and households. It also motivated the development of new applications that shared digitized pictures, digitized sound, and eventually digitized video. In other words, the web was the killer app of the privatized Internet.
• Economic experiments in the marketplace generated valuable lessons about coinvention. The privatization of the Internet motivated a number of ISPs to experiment with developing the access business and applying it to new areas and unserved applications, an observation central to chapters 5 and 8. The deployment of the browser-based World Wide Web motivated a number of programmers and entrepreneurs to start new firms that applied the browser in new applications, an observation central to chapter 9. The actions of entrepreneurs collectively amounted to a large economic experiment in learning how to adapt the Internet to user needs in a variety of locations and commercial settings. As chapter 10 observed, established firms also incurred coinvention expenses as they adapted the Internet to their business needs.
• Network effects in demand accelerated investment in all parts of the value chain. The value of participating in the TCP/IP network—as either a supplier or a user—rose as the number of participants increased. Chapters 8, 9, and 10 explained how increasing household use, business use, and the provision of networking services reinforced the economic gains from investing in Internet markets. This was a network effect at an economy-wide level, and it shaped the activity of every participant. It increased incentives to invest sooner rather than later. It pushed many firms to invest simultaneously, exacerbating the boom.
• Platform governance shaped the character of contributions from the periphery. Different types of platform governance have different strengths and weaknesses; this was the key observation of the analysis in chapter 7. Well-managed modular platforms, such as the IETF’s management of TCP/IP, supported many contributors and inventive specialists. Well-managed proprietary platforms, such as Microsoft’s Windows, also could efficiently manage contributions from many specialized application developers. Open platforms, such as TCP/IP, allowed technical change to come from many contributors by making information accessible and from not restricting what got built for the platform, while proprietary platforms resisted changes that reduced the value captured by a platform leader. That led the open platform to support radical change that the proprietary platform resisted, and it led a well-managed proprietary platform to be superior at implementing a large coordinated improvement in the platform.
• Creative destruction had many commercial determinants. Just as Joseph Schumpeter wrote many decades earlier, creative destruction involved more than just the mere invention of new techniques or new technical knowledge, it also involved the creation of distinct new processes for delivering commercial services. Creative destruction also involved new combinations of business processes for organizations that support new services. This theme appeared throughout chapters 8, 9, and 10. The market-based setting of the late 1990s favored the commercialization of innovation, involving a wide variety of firms with distinct approaches to addressing the market opportunities.
• No established firm willingly encouraged creative destruction against itself. No established firm knowingly agreed to reduce its own value by letting another firm provide a substitute or cannibalize a product or service. No established firm planned for all contingencies and market events, and, therefore, planned to support the entire range of commercial services and applications enabled by new technology. Thus lack of cannibalization and lack of planning made it likely that established firms did less than market entrants in taking actions that fostered creative destruction. This theme arose with special salience in chapter 9, when discussing how IBM had failed to anticipate the commercial web, and how it responded to entrepreneurial entry with new services. It also arose in chapters 7 and 11, when discussing how Microsoft failed to anticipate the commercial Internet and web, and how it behaved during the browser wars by attempting to protect its lucrative position in the value chain for PCs.
• The presence of monopoly and the exercise of market power discouraged innovation from the edges. As chapter 3 discussed, IBM tried to control the privatized Internet backbone as a monopoly, while chapter 11 discussed how Microsoft resisted threats to reduce the value of its monopoly control over personal computer operating systems, and if federal prosecutors had not intervened, they might have pushed even further. Every instance of opposition to innovation from the edges came from established firms who had reasons to resist the competitive outcomes.
• Unrestricted markets supported technological competition between distinct perceptions. The history of the Internet contains several crucial moments when competition between organizations with distinct perceptions supported distinct approaches to creating market value. As described in chapters 2 and 3, in the early 1990s PSINet and UUNET were entrepreneurs with a distinct viewpoint about how the Internet would create value, and they invested more aggressively in the data-carrier business than many established firms. In the middle of the 1990s, as chapters 4 and 7 stressed, Microsoft failed to make aggressive investments in supporting web technology, and Netscape, a start-up with a distinct viewpoint about how to create value, challenged it by trying to grow the browser-based market for new applications. As chapter 13 highlighted, at the end of the 1990s many of the earliest portals failed to anticipate the importance of search, and Google, a young firm with a distinct approach to generating search results, grew a business that supported an advertising-based model of search. In other words, entry into markets fostered technological competition, not as a contest between similar firms with similar visions, but, rather, as the result of unrestricted market entry by young firms competing with a different perception about how to create value.
There was nothing inevitable about innovation from the edges. Some economic behavior encouraged it, but not all did. One important aspect of the historical experience was the way events reinforced one another at a market-wide level. For example, lack of concentrated decision making and dispersed technical leadership fostered a variety of inventions, contributed to greater competitive conduct from more suppliers, and nurtured a great variety of economic experiments. The Internet platform fostered decentralized, independent decisions, which nurtured a variety of inventive specialists and, as it happened, enabled the invention of a killer app that spawned even more follow-on innovation. Each of these actions made the other more valuable and encouraged many actors to take a small action that reinforced the direction of change.
Three surprising observations emerge from this summary at this point. The first has to do with the symbiosis between several coordinated collective efforts—in particular, the design of standards—and the emergence of innovation from the edges. For example, the design of TCP/IP came from a very coordinated process, and that enabled inventive specialists. Upgrading TCP/IP in the IETF was much more of a collective effort, and with the additional investments in applications, it evolved into something no single designer could have imagined. As another example, the initial design of the World Wide Web also came from a single vision. It too became something bigger than any one individual could have imagined, especially once it grew into a large platform upon which others built their applications. Lastly, Wi-Fi also initially emerged from a coordinated collective effort at designing a standard. It too evolved into a range of valuable applications no individual could have designed in advance, building on a range of experiments by wireless access providers.
The second surprising observation has to do with the lack of coordination behind creative destruction. There was no invisible hand automatically guaranteeing that market events would become either creative or destructive. On the one hand, events were consistent with the classic Schumpeterian framework: entrepreneurial outsiders innovated, and existing firms innovated in response to entrepreneurial innovation from outsiders. Yet that is too simple a characterization. The same events that motivated innovative behavior also motivated defensive resistance and attempts to control competitive events. The process of creative destruction emerged only after the failure of attempts at defensiveness by established firms and failures in the attempts to control competitive processes. Competition triumphed if it became overwhelming, involving a wave of entrants and the sum of many actions. In short, creative destruction emerged endogenously because established firms were ineffective at manipulating the competitive process to slow down entry.
Another surprising observation about frictions emerges from this summary. During the deployment of the commercial Internet, low frictions shaped exploratory activities, and that affected multiple participants simultaneously. A friction refers to two related activities: the nonmonetary cost of designing and setting up procedures to deliver a new service or innovative product, and the nonmonetary cost of executing a set of proscribed processes and procedures for delivering services to users—specifically, the nonmonetary costs of employing personnel to gain technical knowledge and the hassles of addressing and managing delays, lost output, and other events for which no secondary market exists. In other words, low frictions encouraged innovation.
Why did low frictions matter? They helped make it cheaper to learn in markets when the most valuable direction of change remained unknown. By experimenting in markets, value creation emerged from novel combinations of demand-fulfillment and operations from multiple vendors combining their complementary services in new ways for mass users. That led to insight that no expert could cull easily from an experiment in a lab or an interview of a few survey takers.
More to the point, low frictions supported creative destruction many entrepreneurs led. Low frictions helped reduce the maddeningly difficult activities of establishing new firms at early stages, speeding up the emergence of a Schumpeterian wave. Any reasonably thorough case study of the processes behind entrepreneurially led innovation at the time emphasized the frustration, confusion, and utter plethora of loose ends an entrepreneurial enterprise experienced, even one with ample funding and adequate managerial experience. A very similar set of observations shaped established firms engaged in exploratory activity. Low frictions behind exploratory activity allowed economic archetypes to work devastatingly quickly, at a large scale across many firms at nearly the same time. That helped propel entrepreneurial activity into a catalyst for innovation from established firms.
Government Policy
Internet exceptionalism also overlooks the role of government policy, or merely mischaracterizes its importance. The US government alone could not have created the commercial Internet, nor could the commercial Internet have arisen in the absence of sagacious policy. The government’s interaction with commercial firms played a crucial role in the development and growth of the commercial Internet. Policy either shaped incentives or placed bounds on the range of outcomes. Sometimes these policies were a matter of choice, reflecting the preferences, wisdom, or foolishness of those in positions of governance. Sometimes the policies were adopted with foresight and the intent to shape the Internet. Other times the policies grew out of unrelated concerns and issues and were adopted for reasons only loosely connected to their consequences for the Internet’s growth.
Policies for moving federally funded technology from universities to private ownership played an enormous role in setting the stage for the Internet’s commercialization. This theme appeared in many chapters, such as chapters 3, 4, 7, 13, and 14, which, respectively, discussed the growth of the backbone, the web, ISPs, Google, and Wi-Fi. Many chapters, especially chapters 5, 8, and 11, referred to one or more elements of the US government’s competition policy for monopolies, and, once again, events cannot be understood properly without understanding how competition policy shaped outcomes. Sometimes the effect of policy was subtler. As chapters 5 and 8 stressed, some policies for the first amendment protected BBSs, and de facto acted as commercial policy to encourage ISPs.
To illustrate the broad point consider how several important policies shaped events prior to privatization:
• Policies mediated the tension that arose when university technologies moved into commercial markets. One policy decision mattered more than any other—the decision to privatize the NSF backbone, as chapters 2 and 3 described. Closely related was the decision to support nonexclusive interconnection through NAPs. Seen with the benefit of hindsight, those who initiated and managed the privatization of the Internet had some notion that procompetitive interconnection policies could lead to good outcomes, but they did not foresee all the effects that would result. The licensing policies of universities also played a role. Sometimes they led universities to assert property rights, as chapter 4 described, and, though done with good intentions, these could interfere with moving invention into commercial firms, such as taken by the University of Illinois during the commercialization of the browser. Sometimes the policies were light-handed, as chapter 13 described, and fostered use of university patents by private firms founded by former students and professors. Such policies ended up fostering the growth of Google.
• Limiting the uses of the research-oriented Internet reduced the range of inventions for the early Internet. The NSF’s managers focused the Internet on the needs of their niche user communities—students, faculty, university administrators, and researchers. As chapter 3 observed, this had salutary effects and also distorted later events. Due to restrictions on participation and acceptable uses, many commercial applications for the Internet were unexplored prior to privatization. There was no possibility for tailoring new products and services to every potential new set of users outside of universities and research settings. Restrictions had an important impact on the applications built prior to privatization, and the large unmet potential available to commercial firms after privatization. The lack of working prototypes of commercial applications for the Internet also played a role in fostering misunderstanding about the potential for new opportunities in the Internet after privatization, as chapters 4 and 6 documented.
• Government actors selectively intervened in network design at early moments. The NSF and DOD played a crucial role in starting and sponsoring organizations that managed and standardized the operations of the Internet, such as helping to establish the IETF, an observation made in chapters 2, 3, and 7. Yet government employees did not have exclusive influence on design choices, nor did they assert a dominating voice. Most of those decisions were left to the Internet community. That discretion fostered the growth of independent Internet governance.
• Pragmatic government policies nurtured institutions for the long run. The NSF’s managers foresaw the need to establish a standards organization to maintain the life of the TCP/IP platform. As chapters 3 and 7 stressed, government managers aimed to have the community survive past the era of exclusive government stewardship. NSF allowed standards development to embody academic preferences for a technical meritocracy, unrestricted participation, and unlimited access to the results. Those policies fostered unrestricted entry as the commercial Internet grew and helped to give discretion to a wide set of innovators.
Government policy continued to play a role after privatization, and no single statement can characterize government policy, nor did postprivatization government policy arise as a fully coordinated attempt to foster the Internet. Perhaps more surprisingly, despite the absence of explicit coordination across policies, many of the commercial policies tilted in the same general direction: toward increasing competitiveness, fostering more entrepreneurial entry, and reducing the role for a dominant decision maker in commercial markets. All of those encouraged innovation from the edges. This general tendency is evident in specific policies, including:
• Common carrier policies encouraged competition in communications networks. US policies for telephony evolved in an era when a single firm dominated, and this theme emerged in chapters 2, 5, 8, and 14. Computer II or federal policy for customer premise equipment continued to evolve after AT&T’s divestiture. They applied to local telephone companies during and after the privatization of the Internet and encouraged competitive markets in complementary services and equipment with importance to the Internet, such as modems, personal computers, local area networks, and plenty of software applications. The key elements would differ from one example to another, but the general principle applied: regulations limited the ability of carriers for voice and data to alter their service for different types of traffic and loads from distinct complementary business services. Said another way, regulations prevented carriers from having any incentive or ability to refuse service, hold up complementary services, or discriminate their pricing and menu of services offered to different complementary firms.
• Antitrust and regulatory policy protected and enabled entrepreneurs. A range of competitive policies in the United States preserved the potential entry of entrepreneurial firms in computing and information services. Some were the outgrowth of policies to limit the reach of the monopoly phone provider, which chapters 2 and 8 described. Some of these were the outgrowth of other concerns, such as First Amendment rights to protect free speech, which protected a BBS with salacious or seemingly unsavory content from censorship, which chapters 5 and 8 described. Some resulted from standard antitrust concerns with the blockading of distribution by established firms, and this observation was crucial for understanding the origins of the competition policies described in chapter 11, which covered the browser wars.
• Antitrust policy fostered less consolidated decision making. Several chapters—notably 2, 8, and 11—discussed many antitrust decisions that moved toward less consolidation—for example, the breakup of AT&T, the continuing monitoring of local telephone firms after divestiture, and the review of the merger to prevent consolidation of communications markets, such as that attempted by WorldCom. These policies encouraged dispersed decision making, reduced the potential that a single prevailing view determined outcomes, and raised the potential for a variety of perspectives to shape the Internet.
• Antitrust policy remained present when the name of the dominant firm changed. The list of dominant firms changed, but not the concerns about dominance. Before the privatization of the Internet the concerns about dominance focused on AT&T and IBM, as chapter 2 stressed. At a later time those concerns focused on WorldCom, Microsoft, and AOL, as chapters 8, 9, 10, and 11 stressed. As many chapters discussed, antitrust policy played an essential role in limiting unfettered governance of key assets by a dominant firm. Such concepts also were applied broadly. A leading software firm, Microsoft, chose to respond with aggressive actions toward Netscape, a firm Microsoft viewed as a competitive threat to its lucrative platform. As chapter 11 described, the US government made a crucial decision when it chose to intervene in browser markets in order to limit anticompetitive behavior. It not only shaped events in those markets, it also made a statement that antitrust could be used to protect distribution of commercial innovation.
• FCC policy could enable the migration of assets from low-value to high-value use. By permitting ISPs to interconnect with the phone system, assets designed for one use, such as voice telephony, found value in another use, such as transferring data to support Internet applications. Chapters 5 and 8 showed that processes designed for one purpose, such as a BBS service, later served another, such as dial-up ISP service. Chapter 14 showed that designs aimed at one need, such as wireless transfer of data for baby monitors and garage door openers, served alternative needs, such as wireless Internet access in enclosed spaces. In some cases, such as the development of wireless access, new forms of property rights were salutary for developments.
• Many policies encouraged innovation from the edges, even though this was not their primary or intended purpose. As chapters 5 and 8 described, the FCC imposed Computer II on telephone companies to encourage growth in information services, and this was implemented long before the Internet privatized, and the prior decisions were reaffirmed to support growth of the commercial Internet rather than recreated anew. As another example from those chapters, the First Amendment protected a set of bulletin board firms in the pornography market, helping to provide a set of iconoclastic entrepreneurs for the young ISP market. And as another from chapter 14, the spectrum was available for the development of wireless Internet access, though the policies that made it available had a long and tortured political history and came into being for reasons only loosely connected to the rise of the commercial Internet.
Once again, nothing about these policies was inevitable. The backbone of the Internet almost became captured by IBM during privatization, as chapter 3 described, but this undesirable outcome was avoided due to the actions of several entrepreneurs, who established CIX. It also was avoided due to the modification of the NSF charter, which required congressional action. Several antitrust lawsuits pushed against the tide of consolidation, but none of these—AT&T, IBM, Microsoft, or even WorldCom—had a preordained outcome, as chapters 9, 10, and 11 stressed. Telephone monopolies could have interfered with the growth of the ISP industry, but rules prevented that from occurring, as chapters 5 and 8 stressed. Government policy makers had to decide to pursue these legal cases, and do so competently.
Policy also could lower frictions—in the sense that government policy could minimize or eliminate an entrepreneur’s need to “ask for permission” of either a large dominant firm or an overbearing government regulator. This was especially evident in discussions about ISPs in chapters 5 and 8. If an ISP interacted with a telephone company that held to minimal government regulations, such as Computer II, then the ISP’s entry could not be delayed. That permitted a private firm to pursue new commercial directions and new opportunities for creating value without giving advanced notice to any other firm or government decision maker. Another example arose in chapters 3, 7, and 14, when discussing standards. If a new entrant conformed to existing standards, such as those defined by the IETF, or comply with rules, as defined by the FCC’s Part 15 rules, then it could move forward with its commercial aims. Thereafter these firms could focus on their commercial aims.
In a related sense not all government policy had salutary consequences. There were some examples of frictions caused by government policy or by the confusion in formulating policy. The privatization of the Internet became delayed by confusion about government policy for privatizing assets, and only after those plans were issued in 1994 could private actors move forward with certainty. As another example, the change in compensation for competitive local exchange providers in 1999 had a key role in fostering the competitive crash in telephony. Finally, as chapter 14 described, the FCC took a long time to designate some spectrum as unlicensed for short-range uses. The timing of those decisions shaped the timing of commercial experiments in wireless applications.
These observations lead to a surprising conclusion. Government actors are inherently limited; at most, policy can solve a specific issue, and suggest a direction, and not plan or mandate a long path for product development. And it was not the standard mantra of regulatory and antitrust policy to lower frictions, and only in a few circumstances—such as the formulation of Computer II—did government policy come close to framing the policies as aiming to foster entry from outsiders and entrepreneurs. Yet, in retrospect, it appears as if many policies aimed at reducing frictions for innovation from the edges, as if many policy makers made small moves in the same direction, and the policies sprung from a nearly unified vision and a persistent and coordinated policy effort, albeit, implemented and executed in flits and spurts.
How could such coordination arise over time, or across such a wide set of government policies and economics episodes? That brings this summary to the role of influential institutions.
Influential Institutions
Internet exceptionalism is deficient in one other respect. It does not orient analysis toward the role of influential institutions—that is, practices and social norms that shaped behavior and outcomes. Despite arising from noneconomic origins, many had an important role in economic outcomes, giving them similar impulses, embedding events with tendencies toward similar directions. As with economic archetypes and government policy, no single influential institution caused the economic experience of the late 1990s. Yet durable institutions last a long time and operate on multiple events, giving them that sense of coordination.
As illustration, here is a short list of several crucial institutions:
• Democratization of standards development. TCP/IP incubated in DARPA and was a centralized and top-down standardization effort. In the 1990s the IETF began to control more of the standardization efforts for new protocols. The computer science community favored nonhierarchical practices, and was notable for what they were not: beholden to the managerial auspices of AT&T or IBM or any other dominant firm, such as Intel, Microsoft, or Cisco. This was an important observation in chapters 2, 3, and 7. Participants preferred not to have a single dominant view control development. Multiple viewpoints were encouraged, and information about outcomes was not restricted to a small number. Similar observations could be made about other crucial standards organizations, such as the IEEE and World Wide Web Consortium.
• Respecting transparency. Transparency was considered an unalloyed good feature of institutions that developed standards. The transparency of processes at the IETF, World Wide Web Consortium, and the IEEE reduced frictions with using standards. Respect for transparency facilitated the development of processes for disseminating information to any participant. As chapters 2 and 7 described, transparent processes allowed participants—ISPs, equipment firms, content firms, and big buyers of networking software—to know that change was imminent and to adjust their plans accordingly. That process helped develop an Internet platform with many interdependent decision makers, on which many other specialized firms built their business.
• Embodying norms that respected independent decision making. Academic norms shaped rules for making information accessible to all comers at the IETF and placed no restrictions on use. As chapters 3 and 7 stressed, these norms presumed a technical meritocracy would select among options invented by specialists, who had the discretion to address problems as they saw fit. Innovative activity was given the discretion to pursue any direction it desired. Once again, similar observations held for the IEEE and the World Wide Web Consortium. There was a pervasive attitude to enable others to build their prototypes and try to distribute them in the market.
• Pursuit of purposeful idealism motivated action. There were many examples of events where the pursuit of an ideal played an essential role. Perhaps no better example illustrates the role of purposeful idealism than the actions of Tim Berners-Lee, who chose to make the pieces of the World Wide Web available as shareware without restriction. As chapters 4 and 7 discussed, a few years after making the shareware available, he chose to establish a consortium to govern the evolution of the web and coordinate its improvement. At no point did he pursue diffusion through a more conventional commercial business strategy from which he could have profited directly. A second example comes from privatization of the Internet. It got the industry off to a competitive start, in part, due to the actions of honest policy wonks. As chapter 3 described, policies favoring competitive interconnection reflected the idealism of a large community, which sought to foster a new communications medium. A third example came from the attempt to release unlicensed spectrum, as chapter 14 described. The fights within the Federal Communications Commission reflected distinct ideologies about ideal policy. The spectrum was released once all political parties found something to like in the program and agreed that unlicensed spectrum put their reformist ideals into practice.
• Impatient entrepreneurship became the new normal. As chapters 6, 9, and 12 emphasized, considerable innovation activity was commercialized by entrepreneurs, funded by US venture capitalists, with IPOs that passed through Wall Street’s financial firms. It oriented enormous sums of investment toward developing many applications for the commercial network. It also resulted in many entrepreneurial firms pursuing their strategies with impatient actions that appeared imprudent in retrospect. At the time, very few questioned the impatience. It is possible to use standard economic reasoning to partially explain behavior in the financial institutions, stressing self-dealing and a number of Wall Street practices that benefited insiders. It is more difficult to explain how so many participants willingly abandoned the standard tools of financial assessment and invested in growing entrepreneurial ventures with thin economic foundations. Yet for a short time it became acceptable and normal to act this way, and many adherents resisted calls to return to prior more prudent norms of business behavior.
The sum of these observations leads to a surprising conclusion. Innovation from the edges emerged under the encouragement of several institutions embedded in US commercial markets and government policy conversations. It was not as if one single invisible hand guided every actor, or the price system coordinated innovative decisions, as in classical economic thinking. Rather, many norms and institutions encouraged innovation from the edges, and inside a variety of myriad economic decisions and policy debates.
General Lessons for Economic Growth
This book has stressed that competition fostered growth in a range of economic activity related to the commercial Internet. This conclusion contrasts with the usual economic argument for more competition, and the contrasting implications deserve attention. The default arguments for more competition stress the direct gains to users from more suppliers, such as lower prices and more customer orientation in nonprice activities. While those benefits did arise in the commercial Internet—for example, more ISPs lowered prices for Internet access—the experience in this instance highlights an additional benefit that comes with competitive decision making. Decentralized decision making fostered a diversity of innovative viewpoints.
One benefit from diversity could be labeled “better benchmarking.” Benchmarking was very valuable when trying to discern whether an established firm overcommitted to a single technological forecast. That type of error was likely when executives grew out of touch with frontier developments due to unquestioned hierarchical norms inside their organization. The lack of investment in the Internet by established firms in communications markets, such as AT&T or the Baby Bells, could serve as one such example. Bill Gates’s error in not anticipating the Internet served as another illustration. Gates would not have changed his view had there not been a firm with a distinct commercial outlook directly confronting his plans for capturing value from a value chain. In other words, diversity of innovative approaches gave observers a point of comparison, which helped managers and outsider investors discern whether a particular firm made a good choice, particularly where there was greater uncertainty about the source of value or the right strategy.
The book also highlights a trade-off between uniformity and diversity, most explicitly in the discussions about standards. Standards performed two distinct functions. First, they produced technical uniformity in some aspects of a design. That reduced variety, by definition. Most standards committees did not pursue that goal with the expectation that a new design for a standard by itself marked the end the story. Most committees did it in the service of a second function, to enable variety of additional actions from commercial participants. Many standards could help specialists invent in one corner of a technical universe without worrying about the other layers. Almost paradoxically, therefore, standards could play a positive role by establishing uniformity in the service of diversity.
Another trade-off between diversity and uniformity emerged from discussions about learning and exploration, when there are a variety of viewpoints about the source of value. When value remained unknown, and could not be quickly or inexpensively uncovered in a laboratory, then experience with markets could contribute to understanding how to deploy and implement new innovation in operating businesses. As the experience in ISPs and wireless access markets illustrated, diversity of approaches generated a range of different lessons, and as many firms experimented, firms learned lessons from observing one another. Summarizing, more diversity fostered a greater variety of different lessons, and the accumulated pool of industry common knowledge became richer and more vital for all participants, and bigger than any single organization could have generated by itself.
In the case of the Internet, low frictions, decentralization, and a diversity of views allowed for an especially potent impact from more entry and competition, enhanced as it was by innovation from the edges. Sampling from a variety of perspectives encouraged the appearance of new or previously unimagined combinations of business activities, or, the coordination of complementary actions that fell outside of those established firms supported. Many inventive specialists discovered how to apply the Internet and the web to applications that had previously only been imagined. Other explorations opened up genuinely new opportunities no research had explored in a lab. Some of those discoveries were unreachable with prior prototypes, and the Internet enabled them by lowering costs or by allowing someone to explore a new combination of technology. The combination of low friction and high decentralization unleashed a particularly potent variety of approaches to market-based learning.
These observations also inform a sense in which the commercialization of the Internet was exceptional. Because innovation from the edges played such a large role, competitive markets reinforced the impact of drawing innovation from a variety of suppliers, both entrepreneurs and established firms. Experimenting with a variety of innovations in all parts of the economy, the whole had greater economic impact than the sum of the impact from any its parts.
That observation should not deter a reader of this book from walking away with one major theme, however. In every case, the key question should not be: how is the Internet different from all other technologies, and why does that make Internet economics exceptional? Instead, inquiries should start by asking: why does the commercialization of the Internet resemble the commercialization of other technologies? The key lessons are learned if the question is: how and why did the operations of economic archetypes, the adoption of government policies, and the influence of institutions encourage or discourage innovation from the edges? Addressing those two key questions provides the crucial insight for understanding why outcomes acquired their specific economic characteristics and contours.