Unleashing Commercial Iconoclasts
In IBM we frequently refer to our need for “wild ducks.” The moral is drawn from a story by the Danish philosopher Søren Kierkegaard, who told of a man who fed the wild ducks flying south in great flocks each fall. After a while some of the ducks no longer bothered to fly south; they wintered in Denmark on what he fed them. In time they flew less and less. After three or four years they grew so lazy and fat that they found difficulty in flying at all. Kierkegaard drew his point: you can make wild ducks tame, but you can never make tame ducks wild again. One might also add that the duck who is tamed will never go anywhere any more. We are convinced that any business needs its wild ducks. And in IBM we try not to tame them.
—Thomas J. Watson Jr.1
Wild ducks has been a colloquial term in computing for decades. By the early 1990s few remembered that it originated with Thomas Watson Jr., the charismatic chief executive officer (CEO) who inherited IBM from his father, and who built IBM into a computing powerhouse in the 1960s.
Watson Sr. encouraged social conformity in his firm because he believed it made his sales force more effective. Salesmen had to wear blue suits, for example. Watson Jr. continued with that practice, but also came to a different understanding with his technical talent, which he labeled wild ducks. Watson’s wild ducks had permission to be diverse—that is, they did not wear blue suits—so long as they invented.
Wild ducks are a particular group of technically adept innovators, often considered social outsiders by those controlling funding. In this sense “outsiders” is a relative term, defined against another norm for routine behavior. Outsiders could exhibit a range of behaviors and social differences, which are regarded as potentially disruptive and costly to the regular operations of an organization. The reverse also often holds—that is, wild ducks regard the practice of those involved in regular operations as interfering with their inventive activity.
In Watson’s world wild ducks also were avant-garde thinkers, often chasing visions they saw as aesthetic, potentially running contrary to received wisdom. Yet they were essential for keeping an organization vital.2 The value of their ideas could defy evaluation ex ante, and in many technical cases it was hard to evaluate even after a prototype was developed. In addition, wild ducks in a laboratory setting could be willing to take a distinct labor contract from others, such as those in sales. A wild duck might take a cut in pay in exchange for (what was regarded as) a more favorable working environment, such as one permitting lab workers to have more research freedom.3
Watson Jr. reiterated these points publicly with a purpose. Wild ducks could serve a useful purpose, but only if an organization made explicit policies to accommodate them. While IBM had made an arrangement to accommodate its wild ducks, their behavior raises a broader question: what arrangement had society made? As this chapter describes, to the extent that the United States had a policy for its wild ducks in computing, the absence of policy was its most salient feature. On first glance, the absence of policy arose as part of a long-standing social compact to live and let live, and not interfere with freedom of expression.
The symptoms were readily apparent in public conversation. A large and vociferous debate preceded and accompanied the NSF’s privatization of the Internet. In contrast, no loud national policy debate informed or shaped the emergence of the commercial Internet access market built around a browser. Why the difference? The short explanation is straightforward: the wild ducks of private industry had lived without public scrutiny for years and wanted it to stay that way.
But that misses something deeper that is not widely appreciated. At the time, activities in the commercial industry seemed unworthy of debate or notice to many managers in leading firms. That is more than just a statement about the fickleness of managerial attention. Rather, the different participants in the nascent Internet access market—many of them wild ducks—took actions that they viewed as both technically incremental and straightforward business decisions, but not as big news. The New York Times, Wall Street Journal, and Washington Post might have written about the Internet when NSF managed it from Washington, DC, but these leading news outlets did not write about the emerging situation in markets, nor was this an obviously wrong decision by editors. For years computer systems had evolved and broadened the scope of their activities into reservations, banking, logistics, and other network services, but rarely did the launching of a new service deserve headline in national news organizations. It merely reflected the normal parochialism of a longtime participant in high technology communications markets.
The full and longer explanation might start on the western side of the Great Plains, in Colorado, in a suburb just outside of Denver. That was the unassuming home for Boardwatch Magazine, the primary trade magazine for firms that offered bulletin board services (BBS). Each BBS existed merely to supply its customers with services for which they were willing to pay. No government agency directly subsidized the BBS industry in anticipation of its crucial role in the commercial Internet, nor did its suppliers get into the business with that eventual goal in mind. Many in the mainstream treated them as merely another breed of wild duck, eccentric marginal actors who satisfied a few fanatics with their services. At best, BBSs were viewed as harbingers of a far-off electronic future.
The most common BBS was a comparatively simple technology by the norms of frontier Internet computing, it was the result of years of exploration in online information services, and it was largely invented and used by a community loosely connected with those involved in the TCP/IP-based Internet. Despite differences in background, the BBS community shared a passion with the community that had grown up around TCP/IP—namely, a defining interest in networking technology and in its uses.4
The name, bulletin board system, suggests a metaphor for what a bulletin board did. It was the electronic equivalent of the corkboards of classrooms, where various articles were tacked to the wall, ready for others to examine; it was essentially an electronic message center that catered to people with a specific hobby or interests. The messages accumulated and generated a conversation. Sometimes this conversation took the form of a “chat,” where multiple users typed in their text, responding to one another. It also offered additional services, such as forums for news groups, shareware sites for articles, and electronic mail among members. Many BBSs also had a large menu of online games. Some even attempted to set up marketplaces where users could shop.5 For many participants it became an addictive habit and the means through which they expressed their social existence.6
With a BBS, a set of users would phone a particular telephone number, usually a local one, and plug their computer into the call. Using (typically) a PC, or (among the technically adept) a workstation, the users would then establish a connection with a server at that phone number. The server contained files for downloading. Seen from the perspective of the Internet, these BBSs would give users information from only a local server and contain no connection to the Internet.
In more advanced configurations, one BBS could call another in serial, sending messages in an architecture called “store and forward,” reminiscent of classic telegraph systems. These were BBSs, however, which made them more effective than any telegraph ever was. One received the message, automatically sent the message to another, who received it and spread it along. Through such architectures every system could get the same message, and rather quickly. While admired by the wild ducks when they first came out, seen from the perspective of later developments, this was a clunky and cumbersome method for connecting many servers.
In other words, in the early 1990s BBSs occupied a position quite distinct from the Internet. Although the personal computer, modem, and related electronic software played essential roles in both the Internet and the BBS, for many Internet users, the BBS was an afterthought, an alternative they occasionally used, but not the center of their activity. For many outside the research communities with access to the Internet, in contrast, BBSs offered the primary channel for receiving frontier commercial services, such as community discussions, schedules and lists, and a variety of electronic documents. As this chapter explains, the blurring of those lines in the early 1990s had enormous consequences for the emergence of innovation from the edges.
The Internet and BBSs shared one similarity: how they took advantage of local telephone calling rules and regulations. Users typically paid for the service, but not the telephone call, because state regulators in virtually every state compelled local telephone companies to provide unmeasured pricing for local calls over short distances (with some exceptions in some areas in which the local caller incurred small per minute charges). In urban areas these distances ranged from a ten- to fifteen-mile radius, but were larger for rural areas. Local calls involved extremely low costs per minute, if any, except to those who called from outside the local area. While the specific features of these policies differed from one state to another, their general features were similar everywhere. States required low prices for local calls as part of a broad program to encourage universal service in telephone use.
The BBS operators were operating under the existing regulatory setting, made by distant regulators. Like the citizen-band radio community and the early PC market, the community of bulletin board providers drew from a wide swatch of eclectic wild ducks that could be found all over the country. Some BBSs had purely commercial motives and others were hobbyists. Some operated their bulletin boards as part of a technical obsession or hobby, others as part of their upstanding commercial business where they took professional pride in the delivery of their services. Still others—indeed, a substantial number—operated the BBS as part of the pornography business and preferred to remain out of the center of the public eye in communities that regarded such activity as salacious and unsavory.
When the Internet became available, there were thousands of BBSs throughout the United States, and there was no national policy debate about what they should do when the NSFNET privatized. Many did not even participate much in the Internet at that point, and those that did, such as those who cooperated with UUNET’s nonprofit activities, were not central to the debate about the NSFNET’s privatization. Yet virtually all of them eventually became active in the commercial Internet by the mid-1990s. Its services were ideal for hosting content and providing other services related to Internet-supported browsers.
With nonchalance at first, beginning in the early 1990s, and becoming visible around 1993, many BBSs decided to become torchbearers for the accelerated arrival of the commercial browser-using Internet. These participants in the BBS industry became familiar with HTML, URLs, and the technical methods for making PCs accept TCP/IP software. This chapter explains how, in a short time, they joined the commercial-oriented Internet. As far as most of those in the BBS industry were concerned, it was not a big deal. They were wild ducks, they were used to being ignored by the mainstream, and they were accustomed to sharing their passions among themselves. They did not see any point in trying to bring more attention to their actions than necessary to generate advertising that attracted customers. They would propel the commercial Internet forward quickly, and, importantly, they did so almost everywhere in the United States.
The Last Mile
The facts on the ground provided no reason for optimism about the Internet’s prospects in the early 1990s. All prior attempts to develop electronic mass markets had crashed on the shores of what was perceived to be “the last mile problem.” The last mile concerned the supply of infrastructure services for delivering data between the national/global data grid and end users.
In the early 1990s, US national regulatory policy focused on growing infrastructure—namely, it focused on a solution employing telephone companies. These companies would deploy technologies that allowed for higher datatransfer rates over telephone lines, such as ISDN (Integrated Services Digital Networks), which supported bandwidth speeds of 128K. Regulators debated how changes to access and interconnection policies altered investment incentives for incumbent local exchange providers. Considerable momentum built up inside telephone company bureaucracies to support the deployment.
By 1994 it was clear that ISDN had a problem, and it was simple to describe: There was not much demand for it. It was too expensive on a monthly basis for most households and too cumbersome to install. Few households or businesses wanted the service. Inside the industry a joke had emerged about ISDN, that it stood for “I Still Don’t Need it.”
Odd as it may seem in retrospect, many contemporary managers within telephone companies did not recognize that there was no need to upgrade the speed of data transfer with an expensive technological solution. The existing telephone service was more than adequate for most TCP/IP-based applications, such as electronic mail and simple browsing. The last mile issues did not need new infrastructure or new technology at this time. Rather, the situation needed new business processes and a viable commercial plan for raising revenue to meet the operational costs of offering e-mail and a browser and supporting those applications.
A commercial service over regular dial-up telephone services based on a resource-sharing protocol, TCP/IP, had not been on this list of solutions to last mile problems until the NSF finished its privatization. With the newly privatized industry demonstrating its feasibility every day, the advantages to the Internet were becoming obvious to those who understood how it worked. TCP/IP worked with minimal installation at a home. Due to long experience with NSF and the military, long-distance data carriers used compatible digital processes; so too did data-exchange carriers, and equipment providers of routers, such as Cisco. Units from MCI, Sprint, BBN, and IBM, as well as entrepreneurial spinoffs such as UUNET and PSINet had operations and provided Internet service.7 Every regional carrier knew how to operate the underlying protocols as well, since it merely involved moving digitized data.
In addition, the Internet provided some (but limited) international communications abilities. While many other developed countries had their own networking technologies (for example, France, Germany), and some countries had oriented themselves to international efforts (such as those at OSI), there were operating TCP/IP networks in many parts of Europe. The working prototypes of the TCP/IP Internet worked robustly now and allowed for vendor and user participation.
It just needed one thing: commercially viable Internet service providers in most locations. That explains why there were many potential disadvantages to the commercial Internet, especially in 1994. Norms for providing Internet service to the business enterprise market were not settled. The servers were usually Unix based, and these offered the most technically straightforward route to TCP/IP compatibility. Using them, however, also required hiring technically oriented staff, which only the largest installations tended to do.
The software issues were complex. Because local area network (LAN) equipment had to be made more usable, some standard retrofits had emerged, but better software upgrades were anticipated. Novell’s Netware, the most commonly employed network and server software, did not blend with TCP/IP-based networking very easily.8 In addition, Microsoft’s server software, Windows NT, was slated to have TCP/IP compatibility, but that product’s release date was slipping, not due until the middle of 1995.
The workstation was the most common client for accessing the Internet, but that was not a general solution. Internet connection software was not easy enough to figure out for a novice user. As with the servers, TCP/IP-based traffic and web-based traffic worked easily on Unix-based computers in a client role, but that appealed only to technically adept users. Many participants awaited the arrival of Windows 95 for the PC, which promised better compatibility with the core plumbing of TCP/IP.9
Nor did the Internet’s proponents have a sexy value proposition to support its widespread use. In early 1994, before Clark and Andreessen formed their company, there was no palpable buzz in Silicon Valley about Mosaic. It had diffused throughout the research and university user community. That working prototype had not yet turned into a recognizable commercial industry of web services. Aside from an abundance of web pages from researchers and university users, there were no start-ups or provocative initiatives. The biggest news in early 1994 came from the venerable Encyclopædia Britannica, which had released a web-compatible version of its text in January. That was interpreted as a notable technical accomplishment, and potentially a competitive response to Encarta, a CD-ROM encyclopedia from Microsoft, but not a motivation for starting any entrepreneurial Internet businesses.10
Household service for novice users also was a big question mark. Some of the existing national bulletin board firms, such as Genie, Prodigy, CompuServe, and AOL, had adopted Internet e-mail by 1994, and retrofit it into their services with varying degrees of success.11 Some firms, such as MCI, had tried offering proprietary electronic mail services but had not enjoyed much commercial demand and had made efforts to offer TCP/IP-based e-mail. No norm had emerged for how services for households should operate in a bulletin board that also provided access to the Internet. For example, there were open questions about whether to charge per time period, per traffic flow in bits, or simply for the connection. There were also unanswered questions about how to charge a premium for increasing the quality of services. Nobody was quite sure what novice consumers would want and what they would do in the face of different prices.
How would the commercial Internet get out of this position? A most unlikely coalition of iconoclasts and visionaries took the first step, and many came from the BBS industry.
From BBS to ISP
Even prior to the privatization of the NSFNET some BBSs began developing a commercial Internet service. A few of those tried commercializing a browser. They all acquired the new label, Internet service provider, or ISP.
Operating an ISP was similar to operating a BBS. It involved much of the same equipment to handle phone calls, as well as the same software to handle password authentication and monitoring. In addition, the modem bank and the arrangement with the telephone company remained the same. With an ISP, however, there were the added features of software that enabled a caller to use the Internet and a connection to the Internet that allowed the ISP to carry the traffic. For most BBS operators, additional features to enable Internet traffic were not technically out of their reach, though not technically simple. That also had an implication for the ISP business. Billing the customer could remain the same, or the ISP could reconsider whether to bill for time or not. Most of the additional investment would go into software development and customer support.
Watching the changes at Boardwatch Magazine were a good barometer for noting what happened to the commercial Internet access market. The earliest advertisements for ISPs in Boardwatch Magazine appeared in late 1993, and the magazine’s staff thought a few of these developments were worthy of news stories. As it turned out, advertisements grew slowly until mid-1995, at which point the operation became so large that Boardwatch Magazine began to organize the presentation, showing many pages of tables in the format of a directory.
Table 5.1 presents the number of ISPs listed in Boardwatch for several different years. The data gives a good sense of the timing behind the growth of dial-up ISPs. In November 1993, only twenty-four advertised; in January 1995, there were thirty-five. By May 1996, however, those numbers had exploded to over two thousand. It would become over 4,500 by January 1999.
The advertisements in Boardwatch were not the idiosyncratic result of something only the magazine experienced. They accurately reflected the trends seen in other publications.12 By 1996, however, the situation had changed in books, just as it was changing for Boardwatch Magazine.13 In other words, the industry began to explode with an entirely new wave of participants.
An Unlikely Partnership
The BBSs operated as part of an unlikely, uneasy, and asymmetric partnership with the telephone companies. The partnership shaped innovative behavior in both good and distortional ways, yet it was not a voluntary relationship, but rather one born from necessity and from reasons idiosyncratic to the United States, founded on years of lawsuits between AT&T and regulators. Local telephone firms and bulletin board operators’ relationships were defined by years of lawsuits, regulatory disputes, and distinct differences in the attitudes of managers and employees about how to employ technology and operate a firm. Such a legally enforced partnership contained multiple recriminations and tedious disagreements over lines of law, and it was a relationship marked by frequent suits, countersuits, and new suits.
Partnership itself was not a barrier to innovation. Partnerships were not unusual in technology markets. The history of commercial innovation contains many famous teams in which tension between the members fostered creativity and rivalry, fueling outcomes that neither individual would have achieved on his or her own. Such teams inside one firm were a staple of high technology lore: Wozniak and Jobs at Apple, Gates and Ballmer at Microsoft, or Bechtolsheim and Joy at SUN.
Fruitful partnerships among firms were also not unusual in spite of the complexity of managing a (possibly) contentious relationship across firm boundaries. For example, Microsoft and Intel had a partnership that dominated the PC industry in the 1990s. Bill Gates at Microsoft and Andy Grove at Intel each had distinct upbringings, ages, ethnic backgrounds, and managerial philosophies about what was best for their firms. In spite of those differences, though, they found common ground to make the situation work profitably for both.
To be precise, however, in all of high-tech lore, there was no precedent for an innovative involuntary partnership among firms resulting in much of anything. In such a partnership, loud bickering and lawsuits would have been expected to lead to little or no innovative performance. In the case of the telephone companies and the BBS industry, that would have been reasonable to expect as well.
First, consider the telephone company. Telephone companies in the United States performed an enormous and essential economic activity.14 These firms primarily had one responsibility: to operate and maintain the world’s largest and most sophisticated landline telephone system. Regulators had long ago found numerous ways to place demanding performance requirements on telephone companies, requiring build-outs into expensive low-density areas so the entire population could receive service. Regulators had also demanded capacity decisions that supported completing calls at all times under all settings, even in the face of extraordinary demand, while also requiring low expenses for the service.
Over the years, the largest telephone companies had evolved to accommodate these stipulations. They were accustomed to routine market demands, predictable investment trajectories, and long regulatory review prior to any change. They had evolved into stable utilities and had grown bureaucratic organizations suited to stable investment under regulated conditions, with little competitive threat.
Telephone companies and BBSs could not be more different in their organizational structure or market orientation. There was no such thing as a typical BBS, but three types of bulletin board operators were predominant in the 1980s and early 1990s: national online service providers, specialized online service providers, and operators who supported online pornography. Each had a distinctly different type of relationship with telephone companies and regulators.
The national online service providers (for example, AOL, Prodigy, Genie, CompuServe) had entrepreneurial roots, but, unlike the other two groups, had experience in state and national regulatory arenas. They viewed the phone companies as cooperative partners in some respects. In part this was due to their size. They regularly negotiated contractual agreements with high-level executives from many firms, and, at least in principle, had their concerns addressed directly. Moreover, as national firms they would not hesitate to hire lawyers to lobby the federal regulatory apparatus to settle disputes.
These firms also had been in business for years. Some, like CompuServe and Prodigy, had corporate parents with ambitious aims and long-term visions for their services. All offered service using extensive proprietary content, as well as proprietary features underlying the standards and protocols. Some also managed to resell proprietary content from other online services, such as LexisNexis. Others sold access to news and other features, such as airline flight schedules.
In 1992, these BBS providers and their standards and protocols were very different from the Internet. Their managers experienced a balkanized world of electronic commerce, one in which users signed up for proprietary services. Communicating between different firms would require bilateral arrangements between firms, and those did not exist as often as they did exist. The national providers encountered difficulties getting their e-mail clients to work with each other, and users could only get content for which their sponsoring firm had arranged in advance.
Prior to the mid-1990s, this group of BBSs had difficulty widening their appeal beyond technically sophisticated home PC users. Their services were cumbersome to set up and operate, and only a technical sophisticated user was willing to learn the precise steps and diagnose the issues. Yet due to their corporate sponsorship, these BBSs tended to have the largest customer bases and the most experience with large-scale marketing and operations. Most important, because their outlooks tended to differ from those found in academic computing and networking, they viewed commercial opportunities in a more competitive light.
For some time, mainstream computing had treated the online providers as prototypes for a distant revolution. For example, when Microsoft sought to establish its own online service (launched in 1994 and named MSN for Microsoft Network), Microsoft’s management thought the revolution was far off in the future. It first approached AOL to buy a piece of it, then after being rebuked, it explicitly modeled its service after AOL, believing it had plenty of time to catch up.15 Because Microsoft also saw the online services as prototypes for futuristic online services, its management concluded that the biggest growth for MSN was many years into the future, most likely coincident with the diffusion of high speed services to homes, delivered by fixed wireline broadband more advanced than ISDN and much cheaper. The expected time of arrival was sometime in the next millennium.16
Few forecast that the demand for, and ultimate arrival of, the browser-based commercial Internet access market would come so much sooner. Nevertheless, just a few years later as Internet demand boomed, the national BBSs attempted to become national ISPs. As early as 1995, several of these providers had well-functioning connections to the Internet for most of their user base. By this point, AOL had gotten into its “carpet-bombing” campaign, blanketing households with free start-up disks. It was the brainchild of Jan Brandt, AOL’s marketing director, and Steve Case, its CEO.17 It simplified installation and gave free demonstrations before it began billing, and user response was positive.
AOL’s method for developing an installed base came in for considerable attention, and ridicule in technical circles. Part of its approach addressed an actual problem—lack of user experience with online services made many potential users reluctant to adopt—by giving them a free look, and, charged them only if they stayed with the service. Some of the ridicule reflected one generation feeling superior to another, with the earlier adopters pointing at their own sophistication. It also reflected something else. Part of its approach involved hustling and garish marketing, distributing the disks by a variety of means—direct mail, newspaper insert, and any other manner that put the disks in the hands of potential users. This type of activity focused on commercial goals, flaunted its unconcealed marketing aims, and appealed to nontechnical users. Such activity had not been part of the NSF-sponsored Internet, and to many longtime participants in the Internet it looked novel, and foreign. To longtime participants in bulletin boards, however, it looked like a further extension of years of the commercial hustling, long found among the spreaders of online services.
The second set of BBSs resembled the first group in technical operation, but tended to be much smaller, focusing on one or a small number of locations. These were specialized organizations that attempted to accomplish a specific function, such as product support or sales (for example, information sales or product manuals), club support (for example, airplane hobbyists), or games (for example, groups devoted to Dungeons and Dragons). Some provided limited user-group functions, such as daily updates of news, organized Usenet group postings, or periodic e-mail. Many of these smaller organizations operated BBSs as a side business, or, perhaps, as one of many activities related to a hobby, a community, or as a service to the local friends.
Some of these, such as the WELL, acquired a loyal following of users who regularly phoned in to take part in long discussions, a precursor to the emergence of electronic communities.18 The WELL achieved a cult-like status among its users, as well as considerable coverage from general media.19 Indeed, many of the online service providers took notice of WELL’s success and began to imitate its practices, viewing community building as a source of loyalty for their national services.20
Most important, the participants from this second group had perspectives that differed from the NSF descendants. They might not have cared as much about stretching frontier computing for its own sake, or about high-minded public debates about the use of public assets. Instead they focused more on making their systems easy for a nontechnical user. Many of them also focused on a local area, providing service to a niche set of customers or community.
Their leaders also differed from many mainstream executives in computing, taking on the role of both a user and supplier of services. They also tried different pricing scenarios. For example, when a small BBS supported another activity, such as local PC repair service, the operator might only have cared about covering their costs, not necessarily about finding ways to maximize revenue.
Because they were present in many cities nationwide, the entrepreneurs among them offered views that differed from technology business leaders in places like Silicon Valley, New York, and Boston. For example, many were often content to supply service to a small town that would appear unprofitable to a major corporation. Many would focus on providing local news or specialized information that national outfits would otherwise ignore.
Some of these smaller bulletin board operators did later become ISPs.21 As was previously mentioned, supporting such groups led to the founding of some commercial ISPs, such as UUNET. However, the vast majority became smaller ISPs, often with a geographically local focus, just as they had done with their BBS.
The third type of BBS, and arguably the most numerous and geographically dispersed, was the operator who supported pornography. Although any individual firm tended to be small, these operators were present in every city in the United States. Boardwatch Magazine listed thousands of them among their BBSs in the United States.
It is an open secret of US communications policy that pornography proliferated on the frontiers of information technology. Suppliers of pornographic material had a history of crashing through convention in order to exploit a new technical opportunity, and they had a history of experimenting with a variety of pricing and marketing models on the frontiers. Moreover, their experiments provided a forecast. Not infrequently, commercial mainstream suppliers would come along a little later with less salacious applications of the same technical and business elements a pornographer had successfully pioneered.
Many of these operators were comfortable with commercial practices the NSFNET descendants and national firms did not employ. They also had two key differences. The pornographers were social pariahs in many cities, protected under the First Amendment, and defended by such groups as the ACLU (American Civil Liberties Union) in light of the perceived larger principles. Yet, mainstream computing firms, as well as many other BBSs, shunned open associations with them. Suppliers monitored the experiments of the pornographers and learned from them, and typically did so without giving credit to them as a group or attributing the lessons to any specific firm.
In addition, many pornographers operated profitable businesses outside of the spotlight. Profitability was important for these prospective ISPs. They found it easy to get the cash from their own operations to become ISPs. That gave them the freedom to invest according to their entrepreneurial instincts, without interacting with banks or other formal financial institutions whose oversight might have altered their investments. In some cases, that led them to develop other functions in the commercial Internet, such as becoming content providers and hosting sites. Unsurprisingly, as the Internet commercialized some of them moved into the lucrative online pornography business.
Like most service industries, the BBS support staff was local, with employees who lived nearby. The infrastructure for supporting the market already existed in the early 1990s due to the spread of the PC. Surveys showed that more than half the US adult population used a computer regularly by 1995, with the percentage among college-educated young men approaching more than three-quarters, albeit, the number with a PC at home was considerably smaller.22 Nonetheless, that was sufficiently large to support repair services for PCs in most cities of any significant size. And there were always retail outlets for peripherals, such as modems, printers, and software. National mail order catalogs also supported specialty purchases.
That BBSs supporting pornography could be found everywhere in the United States is a literal statement about simple supply and demand. The supply of technically skilled entrepreneurs throughout the United States who could operate a BBS for pornography responded to a demand for the service virtually everywhere, and the business made enough revenue to cover the costs, even in a small isolated town. As a result, any location that contained sufficient density of human settlement to justify a hair salon, a grocery store, and a street light at a busy corner also had at least one BBS providing pornography, and usually more.23 Said another way, because of the pervasiveness of pornography BBSs, every small town had at least one potential BBS that could convert to supply ISP service.
In contrast, geographic pervasiveness was not a goal of the national firms. They did not find it profitable to establish local telephone service in geographically isolated small cities, because in small cities the subscription levels just did not generate sufficient revenue to cover operating costs. For the same reasons, the hobbyists and national providers also focused on serving most medium-sized and large cities.
Many of the local BBSs brought with them a hacker ethic, and in one area it contrasted with the commercial online services. Many hackers believed information should be free, and they eschewed the institution of copyright licensing.24 They did not hesitate to download software for sharing. On their bulletin board systems they made available news items for sharing. Their attitude toward paying for content was almost cavalier, if it was present at all. Even if they did not hold this attitude for ideological reasons, many small BBSs turned a blind eye to this behavior among their users because it helped business. They had no incentive to police it.
The hacker ethic had existed prior to privatization, but the NSF had held it in check with the rules for acceptable use. Just as with enforcing restrictions on commerce, some universities had turned a blind eye to violations of copyright by its students, while other universities had policed such activity, preventing it from blossoming. The NSF would be gone after 1995. Without oversight, where would these attitudes take sharing of information and copying of copyrighted material? This was an open question as the BBS providers became ISPs, melding their actions with the newly privatized Internet.
In summary, as the NSF privatized the Internet in 1994, the BBS industry had outlooks that differed from telephone companies. There was no single typical response from these firms. They had widely different backgrounds. They did not speak with a single voice on any of the key questions regarding forecasting, strategy, or public policy, which is what made the innovations that evolved from their unlikely partnership with the telephone companies so unpredictable at the time, and incredible and surprising in retrospect.
Rules, Regulations, and Restrictions
The partnership between BBSs and telephone companies got much of its animus from prior disputes over regulatory rules. Many local telephone companies tolerated the BBSs because they had to, not necessarily because they wanted to. Many BBSs did not trust the telephone companies and remained wary of any action they viewed as a betrayal of precedent.
The FCC compelled local telephone companies not to “discriminate” against BBSs, or any others that were classified as “enhanced service providers.” These complex regulations grew out of years of antitrust and regulatory lawsuits and inquiries at the FCC, focused on preventing AT&T from blocking development of new services complementary to the telephone system. They governed the designs for communications equipment firms and the operations of information service firms, where competitive supply conditions prevailed. They also governed how both types of firms interacted with monopoly telephone companies and aimed to prevent the monopoly from shaping competitive conditions in distortionary ways.
The set of regulations changed over time, and became known as Computers I, II, and III. By 1993 and later, only Computers II and III were in effect.25 Computer II would play a profound role during the commercialization of the Internet. Yet Computer II’s adoption and design arose from many motivations, and some of those were only loosely connected to the Internet. This is worthwhile to understand in detail, except that the legal nuances would put to sleep all but a couple dozen wonkish addicts of telephone regulation. What follows is a compromise between a brief synopsis and a full explanation.
Computer I began in 1966, resulting in an order in 1971. The issues motivating Computer I predated the design of TCP/IP and concerned issues more fundamental to the regulation of telephone systems. It began to put in place rules for mediating the relationship between the telephone company (who had rights because it “owned” its equipment), and others with computing equipment (who had to have legal “permission” to attach to the network).
Computer I would accommodate computer modems and did not anticipate what those modems would be used for. That would prove awkward for the FCC in the 1970s, especially as the earliest experiments in commercial packet networking became available for private users. Computer II, issued in 1980, was designed to resolve these tensions and was viewed as an improvement to Computer I. Should the telephone companies have monopoly rights to these packet network services? (No). If multiple firms provide such services, then should they be regulated like monopoly local telephone service? (No). If these are not regulated as monopolies, then should there be a bright regulatory line between them and the telephone? (Yes). What are they to be called to distinguish them from telephone service? (“Enhanced Services”). Can an enhanced service get access to a business line from the telephone company, then resell access to it as a BBS? (Yes). Once again, notice the timing: Computer II was put into practice long before DARPA transferred stewardship over part of the TCP/IP network to NSF. Its motivation is connected to the rise of packet networking and services using digitized data over telephone lines, which loosely connects it to the rise of the same technical factors that yielded the Internet.26
Computer II had a seemingly straightforward purpose: to preserve and nurture competitive markets in services and products that worked with the telephone system. It did so by tightly describing the obligations of telephone providers and circumscribing their ability to escape these obligations. Computer II achieved this objective with two interrelated requirements.27 First, it permitted telephone firms to enter these same markets only if the telephone firm contained a structurally separate division offering services that competed with new entrants. Second, Computer II designed the form of the protection in a way that made it easy to monitor. Telephone firms were compelled to offer a competitor what it would offer to its own division.28 The first provision permitted phone companies to use their expertise in new markets, while the second provision limited the ability of the phone company to put a potential rival at a disadvantage.
Most relevant to what follows in later chapters, by the early 1990s, every manager at a local telephone company had lived with these rules and knew them well. They required the telephone firms to treat a competitor the same as its own division, and they applied to BBS operators, just as they applied to any other electronic service, such as ISP lines to university networks. More to the point, the FCC would face questions around the time of privatization about whether to revisit these policies as BBS operators considered carrying Internet traffic, and the FCC would largely choose to leave the policies unchanged. That left all potential parties in a familiar regulatory environment with known regulatory obligations.29
These rules were incredibly important to BBS operators. The long fights to develop the rules left an uncomfortable legacy with many BBSs. Many believed they could not order lines from telephone companies and resolve technical issues to make their modem banks operate efficiently without the routine cooperation of the local telephone firm. Many managers also believed those local telephone firms would not be cooperative without being compelled to do so.
If the Internet did not originally motivate the design of these rules, what motivated them in the first place? It is easy to conceive of them as a by-product of specific legal cases, though that is too narrow. Seen in a broader light, however, they can be interpreted as reflections of two longstanding tendencies in US antitrust law and regulation. For one, US antitrust law enforcement agencies have a long history of attempting to prevent a dominant incumbent telephone company from using its dominance in one market (for example, local telephone service) for commercial gain in another where competitive entry and innovative activity might emerge (for example, selling data or selling services that use digitized data). Another distinct motivation arose out of common carrier regulation, which had been applied to telephone companies for many decades. It required monopoly utility carriers to refrain from using their monopoly to benefit one business partner at the expense of another market supplier, that is, not to discriminate against any potential business partner.30 The phrase “common carrier” embodies the principle: carriage has to be provided as a common good, available to all on the same terms.
Summarizing, the broad principles of antitrust law and common carrier regulation were widely appreciated, and the details of each underwent frequent reapplication to suit new circumstances and technologies. These rules were regarded as a nuisance for local telephone firm operations in the details, but, as a technical and operational matter, they were also a fact of life in the business. Any such engineering details had long ago been worked out, and new services in the later part of the 1980s and early 1990s tended to raise incremental issues, at most.31 The daily issues behind compliance were routine. Because the bulletin board volumes were comparatively small in relation to voice telephony, most local telephone firms and local regulators thought the burdens were a manageable nuisance. The timing and features of calls to bulletin boards looked distinct from voice calls, placing peculiar demands on capacity during peak load times, but most often at nonpeak times, after business hours. They were also a small source of revenue, since they led users to demand second lines.
A slight fast-forward can also illustrate a related point. The first major restructuring of US communications law since 1934 took place during the writing of the 1996 Telecommunications Act. The act borrowed much of its wording for the sections regarding enhanced information services from Computer II and embedded it in the legislation’s language. The act retained the architectural logic of the regulations—its seemingly bright line between the monopoly provision of local telephone access and competitive provision of enhanced services.
Additional evidence of its acceptance can be found in the priorities of the White House after commercialization. A memo from Ira Magaziner became well known, as it argued that many aspects of the network’s commercial evolution was too complex and contained too many unknowns, and did not lend itself to government intervention.32 Despite that stance, the Clinton administration was busy negotiating with Congress on many legislative fronts related to the Internet and would choose to focus its attention on issues such as provisions for the Communications Decency Act, the Digital Millennium Copyright Act, and the establishment of ICANN. It did not return to the regulation of ISPs until the Internet Tax Freedom Act, passed in October 1998, which placed a federal moratorium on taxing the provision of Internet access.33
Summarizing, the commercial market for dial-up Internet access emerged at a time with comparatively favorable regulatory environment for ISPs, due to Computer II. The presence of a monopoly provision of local telephone access did not delay the rollout at most ISPs, nor did the need to adjudicate and tailor a new set of regulatory rules put new frictions in front of an entrepreneur selling ISP service. Entrepreneurs everywhere focused, instead, on something other than the debates inside Washington, DC; they focused on developing their Internet service.
An Implicit Policy
In the research-oriented Internet, users employed dial-up services operated by their universities. In 1994 the open question was whether a similar service would arise for commercial users of the newly privatized Internet. Initially, a few commercial ISPs forged ahead, presuming that they would absorb the regulatory norms from the BBS industry. At first, telephone companies did not deter them. Those that tried did not succeed in suppressing all the experimentation. That lack of opposition did not change for most telephone companies until the mid- to late 1990s, when the ISP business started to become large. In other words, neither US telephone company executives nor regulators initially assumed that the telephone companies should be the only providers of Internet service.
Was that policy the outcome of a great, deliberate procedure? No. It was a policy based on historical inertia over a set of regulatory rules adopted in an earlier era for a related but distinct set of market circumstances. Why did that policy matter? Later events in other countries would illustrate why. Regulators in many other countries did not inherit any tradition of separation between carriers (telephone companies) and content providers (BBSs/ISPs). As the Internet commercialized in their countries, they just handed Internet service to their telephone companies without debate. Pricing and service norms evolved in the context of the distribution of service from that monopoly provider.
Not so in the United States. In principle, BBSs were not precluded from turning themselves into ISPs and trying to offer Internet service if they so wanted. This leeway gave US ISPs the opportunity to act and, crucially, it gave them some discretion with which to act without interference from their business partners, the telephone companies. That is, ISP managers perceived that they could explore commercial possibilities without telephone companies constraining their actions—a situation that was protected by the ample precedents embodied in Computer II.
As the commercial Internet began to develop, some BBSs began experimenting with offering service, a few as early as 1993 according to the ads in Boardwatch. By 1994 all the national BBSs were experimenting with some form of Internet access, albeit, often it was limited in its capabilities. Clearly, however, retail services had started sooner than that for some customers, especially in business. For example, PSINet and UUNET sold service as early as 1990 and offered backbone services to others comparatively early as well.34
The date of the first BBS turning into an ISP is less relevant than the fact of its transformation as a precedent. Even in 1994, few ISPs existed, and their total national revenue amounted to little. Prototypes do not have to be large, nonetheless, to demonstrate the viability of their commercial service.
Events in 1994 helped crystallize the viability of the BBS as an ISP. A BBS needed to add two things to become an ISP, and neither was out of its technical reach. First, a BBS had to arrange to make a connection to the Internet. That would involve establishing contractual relationships with another firm to carry data, and it usually involved setting up an additional server to handle the management of data traffic between the Internet and the user.35 That was feasible to do in most major cities in the United States, because by 1994 the NSFNET was in the midst of privatizing the Internet in many locations around the country, and its plans had fostered a competitive supply of backbone providers. Sprint, AT&T, PSINet, and others offered backbone interconnections to any ISP, and a market for interconnection between ISP and backbone had begun to spring up in virtually every major city.36
Second, a BBS/ISP had to add software to enable the browser to work. That was not trivial because the conception of the business changed. Previously, the bulletin board provided all the information in proprietary services. A browser transformed the BBS into an ISP with extraordinary capabilities. An ISP gave its users a window on visual information provided by sites all over the Internet.
In 1994, at a time before many web pages existed, the change in business begged questions about where content would come from. While universities’ websites tended to be the primary providers, e-mail was another common early use. The questions soon ended as new providers of content flooded in from a variety of sources. As it turned out, Tim Berners-Lee’s software was easy to use, and self-publishing web pages became one catalyst for growth of the web.
It also was possible to simply be an ISP without any content. For example, like many ISPs offering retail access accounts, Netcom began business well before Internet privatization was finished. Established in San Jose, California, by Bob Reiger, an information systems engineer at Lockheed, Netcom grew into a provider of Internet access to students off campus. Reiger never had a business as a BBS and never felt he had to. Instead, he always focused on improving the quality of the ISP service. He became particularly well known for making tools so the service worked well with Windows 3.1. Furthermore, Netcom’s business grew with the web. By the mid-1990s, Netcom became one of the leading providers of Internet access, at its peak in early 1996 claiming close to half a million worldwide users.37
Overall, in 1994, the browser was an incremental change to a BBS already offering Internet service. That is a key observation and worthwhile to highlight: the browser depended on a set of standards, the World Wide Web, which was a layer of software on top of the TCP/IP stack. An ISP enabled this to work if it downloaded the appropriate shareware software and sent one piece to its user.
Any BBS could easily learn the additional steps necessary to make it work with the World Wide Web. Many technically adept engineers with experience with PCs already knew the basics and could open an ISP from scratch, once again, tweaking it to work with the World Wide Web.
As ISPs began to grow in number and size, an additional requirement became more salient: the need for more telephone lines. When BBSs were small, this had not been an issue, but the growth of ISPs soon generated a need for many phone lines. Most ISPs could not arrange for that capacity from their phone companies overnight. Soon, getting these lines would become a contentious issue in many areas.
An identity began to coalesce at about the same time. Some self-aware ISPs characterized their place as outsiders, as rebels within a bottom-up movement. Their labels said it all. They began to cast themselves as “Net-heads,” in contrast to the “Bell-heads” who did not perceive the Internet on the same terms.38
Enabling Vital Visionaries
The first entrepreneurial ventures in Internet access were largely undertaken by commercial iconoclasts. These entrepreneurs occasionally rejected conventional wisdom, maintaining a view about the future that differed from the majority. Often, as in this case, the majority treated these visionaries as eccentrics to be ignored, or, perhaps, lone wolves that are allowed to howl at the moon as long as nobody else loses money on behalf of their idiosyncratic views. Mainstream industry considered the iconoclasts to be peripheral participants, living on the edges.
The academic community was one source of potential iconoclasts for the commercial Internet, but it did not become a major source. Although that community had developed many prototypes, it did not seed many firms. Only a trickle of firms showed up with this background, such as PSINet, UUNET and Netcom. Had that been the only source of entrepreneurs for the commercial Internet access business in the United States then the access market probably would have grown, but not as fast as it did, and not in so many places.
The BBS industry provided another source of potential entrants, and, as later events would show, an abundant one. Many were far more optimistic about the business prospects than many administrators with academic experience.
The commercial Internet gave commercial iconoclasts the discretion to act in the Internet access market. That occurred for numerous reasons: The NSF backbone connected them all at low cost; the presence of multiple backbone providers largely supported multiple points of interconnection in major urban areas; the end-to-end architecture of the Internet gave potential ISPs discretion; US regulatory rules reduced the frictions of setting up modem banks; the shareware norms of academic computing gave Apache to everyone; and many BBSs decided to act, and act in ways they perceived to be incremental.
For a time these iconoclasts did something they perceived to be so incremental that at first these actions hardly received notice from those to whom it would later appear to be radical. Only as it started to gain scale in the market—by signing up customers all across the country—did other participants in the Internet realize that these innovators had come from the edges, and had operationalized all the elements of a new and potentially mainstream business.
1 Watson (1963).
2 Thomas Watson Jr. reiterated the point in multiple occasions. Consider the following:
I talk a lot around here about wild ducks, and people kid me a good deal about my wild ducks. But it takes a few wild ducks to make any business go, because if you don’t have the fellows with the new ideas willing to buck the managerial trends and shock them into doing something new and better, if you don’t have those kind of people, the business pretty well slows down. So I would tell a 21-year-old I.B.M.’er what I’ve told a lot of 21-year-old college people … that is, that the priceless ingredient that a youngster has when he starts in business is that sense of not compromising beyond a certain point.
Quoted in http://www.nytimes.com/1989/05/07/business/l-wild-ducks-048289.html, accessed May 2010.
3 This theme extends into other industries as well. For broad views on this topic, see Christensen (1997), Stern (2004), Rich and Janos (1994), and for its application to the early Internet, see Greenstein (2010a).
4 For a history of the earlier generation of experiments, see, e.g., Hauben and Hauben (1997), Banks (2008), Turner (2006), or Campbell-Kelly and Garcia-Swartz (2013).
5 These shopping sites largely did not succeed prior to the rise of the commercial web. See Banks (2008), or Campbell-Kelly and Garcia-Swartz (2013).
6 See especially Hauben and Hauben (1997) and Banks (2008).
7 For an examination of the technical and business issues of developing services, see, e.g., Marcus (1999).
8 Forman (2005) describes these issues and offers statistical evidence about their consequences for how fast enterprises adopted the Internet.
9 Numerous retrofits and add-ons had enabled users of Windows and DOS-based machines to get functional service over dial-up networks and local area networks, but these technical solutions were ad hoc.
10 See Greenstein and Devereux (2009).
11 See the extensive discussion in Banks (2008).
12 In one of the earliest Internet “handbooks,” Krol (1992) lists forty-five North American providers (eight have national presence). In the second edition of the same book, Krol (1994) lists eighty-six North American providers (ten have national presence). Marine and colleagues (1993) lists twenty-eight North American ISPs and six foreign ISPs.
13 For example, Schneider (1996) lists 882 US ISPs and 149 foreign ISPs.
14 In 1992 the telephone industry generated 171 billion dollars of revenue, with the three largest components going to local service ($43B), long distance ($69B), and network access ($31B). Annual Survey of Communications Services, US Census. http://www.census.gov/services/sas/historic_data.html, accessed July 2009.
15 Ferguson (1999), 53, reports a story that commonly circulated at the time. He says Gates approached Case, CEO of AOL, on the recommendation of Russ Siegelman, the head of the Microsoft Network, MSN. He then reports,
With his characteristic delicacy, Gates said to Steve Case, “I can buy twenty percent of you, or I can buy all of you, or I can go into this business and bury you.” Displaying considerable courage, Case replied that AOL was not for sale, so Gates decided to bury him. That same day, Gates authorized Siegelman to proceed with development of the Microsoft Network.”
16 For a longer explanation, see Greenstein (2008a), Haigh (2008), or Bresnahan, Greenstein, and Henderson (2012).
17 Banks (2008), 135, reports that the first experiment with carpet bombing began as a direct mail campaign to households, in which Brandt included free time with the disks. Steve Case expressed skepticism, but authorized a small experiment with 200,000 people at a cost of $250,000. After an astounding 10 percent response rate, AOL began expanding the approach with more direct mailings and other approaches. AOL’s campaign was so large that in some weeks of 1998 it bought the world supply of disks. Stated Brandt, “At one point 50% of the CD’s produced worldwide had an AOL logo on it.” See Siegler (2010).
18 In 1985 Stewart Brand and Larry Brilliant founded the Whole Earth ’Lectronic Link (WELL, for short), starting a virtual conversation between the writers and readers of the Whole Earth Review.
19 See, e.g., Markoff (2005), or Turner (2006).
20 See, e.g., Swisher (1998).
21 This transition is very apparent in the directories of the 1997 and 1998 editions of Board-watch Magazine.
22 Table 1138, Adult Computer and Adult Internet Users by Selected Characteristics: 1995 to 2005, Statistical Abstract of the United States. 2007. The National Data Book.
23 See the discussions in Aspray (2004), chapter 2, and Cronin (2006). As further evidence, the 1996 edition of Boardwatch Magazine’s directory makes it apparent (from examination of the names of many of these BBSs) that BBSs that specialized in pornography were geographically dispersed across the country.
24 For longer discussions, see, e.g., Markoff (2005), or Turner (2006).
25 There is a very long history behind these events, and this chapter reviews only a part. See, e.g., Noll and Owen (1989), Werbach (1997), Oxman (1999), Cannon (2001), Owen (2002), Hogendorn (2005), and Nuechterlein and Weiser (2005).
26 This should be read as a statement about the origins and adoption of these rules. Their merits were revisited periodically, as circumstances and the ideology of policy changed. In particular, these rules were revisited just as the commercial Internet was developing, and retained. The rule about reselling business lines was key to developing a competitive dial-up ISP industry. See Oxman (1999).
27 This is a necessary oversimplification. Cannon (2001) provides a history of these rules and how they changed over time.
28 Under Computer III, a telephone company could have an integrated division as long as it also had a very detailed and approved plan for supplying interconnection to other providers. The first Computer III order was issued in 1986 and underwent subsequent revision and court challenges. At the dawn of the commercial Internet Computer III was still undergoing change.
29 This is a necessary oversimplification. See Cannon (2001) and Oxman (1999).
30 In addition, Judge Green administered the modified final judgment of the divestiture of AT&T with bright lines regulating local telephone firm involvement in enhanced service markets. See Noll and Owen (1989) or Nuechterlein and Weiser (2005).
31 This is a simplification for the sake of brevity. Vendors did learn what was possible, but the engineering challenges turned out to be manageable. See Werbach (1997), Oxman (1999), or Nuechterlein and Weiser (2005).
32 Brian Kahin, private communication, September 27, 2013.
33 The market’s young status justified the law, according to supporters, who worried that excessive local taxation could deter growth for the new nationwide commercial applications of electronic commerce.
34 It became something of a point of pride to claim to be the first provider in a local area. From quick examination of the Boardwatch advertisements, the earliest entrants tended to be in major urban areas, such as Chicago, Boston, New York, San Francisco, and so on. Of the national online service companies Banks (2008), 165, reports that Delphi first found a way to get TCP/IP e-mail in 1992.
35 The protocols for doing so improved over this time. ISPs first employed SLIP, or serial line Internet protocol, which was written by Rick Adams, founder of UUNET. See, e.g., RFC 1055. This was later replaced by PPP, or point-to-point protocol. The latter was first proposed in the IETF in 1989, and improved over the early to mid-1990s. See, e.g., RFC 1661, from July 1994.
36 Seen through the historical lens of frequent interconnection disputes in telephony, the presence of multiple backbone firms left no backbone firm in a position to be a monopoly provider, except in low density areas, where it was possible to charge high rates for access to lines. Said succinctly, lack of monopoly in urban areas eliminated the viability of using interconnection as a strategic tool to advance business aims in retail markets in urban areas.
37 This included customers all across the United States, as well as in Canada and the United Kingdom.
38 See, e.g., Friedan (2002) for further discussion.