From hierarchy to open configurations: decentralization and user-generated content

Computer utility to the internet

In the 1960s there was much talk of a “computer utility” wherein computing power would be “piped” into individual homes just like electricity, which does not require each household to own its own generator (Campbell-Kelly and Aspray, 1996; Dunlop, 1970).

The economics of computing seemed to dictate the creation of these computer utilities. According to the then widely accepted Grosch’s Law, the “power of a computer varied as the square of its price” (Campbell-Kelly and Aspray, 1996: 216). Since the economies of scale appeared to favor more expensive computers, it seemed to make sense to share a centralized mainframe among many users rather than to install a computer in each household (Campbell-Kelly and Aspray, 1996; Parkhill, 1966). Based on this logic, the general expectation was that the time-sharing computer networks would eventually evolve into computer utilities.

By 1967 time-sharing services had grown to a $20 million industry and new systems were popping up all across the country. Then, suddenly, the computer utility concept became passé.… The main reasons behind the collapse of the computer utility paradigm were the unexpected difficulties in producing software to run large systems, combined with rapidly declining hardware prices.… Now, it made more sense for organizations to buy a minicomputer outright and set-up an internal time-sharing system than subscribe to a time-sharing service and pay per hour of use. (Campbell-Kelly and Aspray, 1996: 222)

Despite the time-sharing crash, the idea of mass computing lived on. In the 1970s the concept reappeared as “videotex,” where the idea was to provide individual homes and offices access to information via the television set or a dedicated terminal.

Organizationally, videotex’s development as a “vision” project was animated by converging agendas of myriad players. Broadcasters were keen on making use of the “extra” capacity in the bandwidth separating channels and in vertical blanking intervals (Case, 1994). Telephone companies wanted to increase traffic over their networks (Cats-Baril and Jelassi, 1994). Equipment manufacturers wanted to sell more sets and terminals. Newspapers wanted additional outlet, especially for “excess” news that did not get printed (Compaine, 1985). Banks and other firms were interested in reducing transaction costs via electronic transactions (Tydeman et al., 1982). In this way, videotex became a vessel for a mingled potpourri of opportunistic actors and interests.

Consumers, on the other hand, were reluctant and showed little interest in the new technology. The British Post Office launched Prestel in 1979 with a forecast that it would have 100,000 users in two years. After ten years, it had only 90,000. The results were the same in most countries. The videotex projects were stuck in the classic chicken and egg problem: “costs too high to develop a market – the market too small to bring down the cost” (Witcher, 1985: 19).

To create the critical mass, although the simple robust terminals, Minitels, cost $165 each, in early 1980s the French telephone company started distributing them at no cost (Nohan, 1986). Its calculation was that the increased traffic would recover the terminals’ cost in four years. By May 1987, 1.6 million Minitel terminals were in operation generating 40 million calls per month, about 50% of which were for the telephone company’s electronic directory, 30% were generated by business use, and the remaining 20% by services provided by entrepreneurs. The surge of entrepreneurial activity was no accident. The telephone company gave two-thirds of its fee, about $5 per hour per customer, to the entrepreneurs. Entrepreneurs rushed in to provide services like dating services, electronic shopping, legal services, and games – cutting-edge online services that had never been seen before (Peterson, 1986).

Minitel’s success attracted much attention the world over. For a while it seemed that Minitel had established the dominant model for online services for the foreseeable future. Just then the development of the PC set into motion a paradigm shift towards the internet. The emergence of the internet did create a computer utility of sorts but its character was very different from what had been originally envisioned, as it has no center per se.

At first the French did not pay much attention to the internet. They saw the internet as a “competing technical (and US-dominated) standard, and not as the architecture for the information society” (Brousseau, 2003: 49). Later, as the realization dawned that the internet was far bigger than just a competitive videotex, there was a reassessment of French thinking (Mailland, 2009). In 1996 the French government set up a commission to study the implications of the internet’s ascent. The commission concluded that the internet represented not an incremental change of an evolutionary type but a radical shift in paradigm (Fletcher, 1997). Similarly, another report warned of “in-depth upheaval of our ways of thinking and operating; a centralised, vertical approach is having to be replaced by flat, horizontal and decentralised directions” (French Government, 1998: 15). Accordingly, Prime Minister Jospin asked France Telecom to start migrating Minitel services to the internet (Strode, 1999) and the Minitel started to be phased out.

Wireless internet to open cell phones

The wireless internet projects skipped the hierarchy phase because their objective from the very beginning was to extend the internet into the wireless arena. In other words, the internet, an open system, was to be the primary source for content. However, the system architecture itself was hierarchical. In effect, the wireless internet projects were open-hierarchies. They took two different routes around a similar time – the late 1990s: WAP (Wireless Application Protocol) in most parts of the world and i-Mode in Japan. We will first briefly discuss the WAP projects, which struggled, and then focus on successful i-Mode and subsequent developments.

WAP was created as a cross-platform solution by an industry consortium. It stripped down internet web pages, especially graphics, to make them viewable on mobile screens (Leavitt, 2000). It floundered because of quality of service and usability problems. It took too long for users to access the desired information, for example, two minutes just to check the weather forecast. In addition, they had to dial in every time they needed access (Digital 4Sight Corporation, 2001). Furthermore, the content creators had difficulty working with WAP, which required a different set of skills than the internet’s HTML (Kelly, 2006). The lack of user enthusiasm dampened investments by the content providers and we had the classic chicken and egg problem.

In 1999 DoCoMo launched i-Mode, which made transactions easy by enabling content providers to collect micro payments via the phone bills (Lemon, 2001). In marked contrast to WAP projects, DoCoMo targeted youth for entertainment services and offered “always on” capability (Fransman, 2002; Ratliff, 2002; Tanikawa, 2000). Furthermore, it used C-HTML (compact HTML) markup language (Ratliff, 2002; Sandoval and Bradley, 2000).

i-Mode has two types of content providers: official and unofficial. Those charging micropayments are official ones. They have gone through an approval process and are listed on the menu. DoCoMo’s control over service providers and subscribers, as both can be denied service, creates a secure environment for e-commerce (Fransman, 2002; Natsuno, 2003).

The unofficial ones, on the other hand, are not listed on i-Mode’s browser. Subscribers can access them by typing URLs and using bookmarks. These sites feature advertisements or charge fees via credit cards or third-party billing services (Lemon, 2001). They can provide adult content. i-Mode was a huge success. Its success attracted competitors: KDDI and Softbank Mobile. While there are variations among the competing systems, they are cast in the same mold. Henceforth the discussion will focus on the archetypal i-Mode model rather than DoCoMo’s i-Mode specifically.

Initially i-Mode was skewed towards official providers. It was easier to access their websites and pay for their services. However, two recent developments have changed the dynamics: search engines and micro browsers.

By 2001, Google had a search engine for i-Mode. Its integration into proprietary browsers, however, was cumbersome. That changed when KDDI developed micro browsers that enabled access to internet websites in 2004 and then, in a strategic partnership with Google, started featuring its search window on the new browser in 2006 (KDDI, 2006). The ensuing traffic pattern is noteworthy: while 22% of the queries lead to visits to official sites, 47% and 31% lead to unofficial and internet sites respectively (Billich et al., 2007). DoCoMo and Softbank followed suit. Now less than 10% of i-Mode revenues come from the commissions paid by official sites (KDDI, 2007).

The open cell phone enthusiasts are bent on pushing the process towards un-order. The Homebrew Mobile Phone Club, modeled after the famous Homebrew Computer Club, describes itself as “a physical and virtual club to do to mobile phones what Jobs and Woz did to computers.” ² Suraj Patel of the Tuxphone Project, another such initiative, notes that it is quite taxing to put together a “homebrew” cell phone and the final product is clunky. “So what’s the motive?… the urge to play around with devices that are an integral part of all our lives,” which he predicts will lead to unexpected innovations (quoted in TMCnet.com, 2006, online). Similarly, Casey Halverson, a Seattle-based open phones enthusiast, thinks that as more people use mobile devices, “they will be running into more and more limits with closed systems” (quoted in Borland, 2005: online). The hacking activities may go beyond the geeks to the general public and the result could be an open platform.

While technical openness does not necessarily result in the generation of user-generated content, it does open up the possibility. If past experience is anything to go by, we could expect the chances of seeing that happen to be pretty high.

Discussion and conclusion

The product of our analysis – stages in system decentralization – would not have been borne of either the “technology” or “technologies” approach. The former would not have registered the details specific to a particular type of information system. The latter would not have registered the commonality that runs through such systems. Our “technology-technologies” approach, on the other hand, yields a framework that will enable us to approach the next new technology with increased analytical sophistication. Even if the next new technology does not have the same development pattern, the framework will prompt us to study it in a more nuanced way.

With regard to the specifics of the information systems we studied, we were particularly interested in understanding the transition from hierarchical configurations to flatter ones in these systems, especially the intermediate stage we call open-hierarchy. One of our objectives was to demonstrate this intermediate open-hierarchy stage. We see the open-hierarchy stage quite clearly in Minitel (hierarchical delivery system, open content creation) and i-Mode (hierarchical delivery system, open content creation). We will instead focus on our second objective – the overall dynamics of the transition process.

Yoneji Masuda, writing in 1981, could see way beyond the videotex, which had then captured everyone’s imagination. He saw four stages in the development of the information utility: (1) Public Services Stage: Information is generated by the service provider, (2) User Production Stage: Information is generated by the user for his/her own needs, (3) Shared Utilization Stage: Information generated by individual users is shared with the public at large, and (4) Synergetic Production and Shared Utilization Stage: Information is produced by groups of individuals and publicly shared. Masuda noted that first stage is represented by the videotex. But he could not use terms that would be familiar to us for the remaining three stages because they did not exist then. He was essentially referring to the internet, worldwide web, and Web 2.0. What is interesting for our current purpose is that even Masuda, who was correct about almost everything, thought that the system would be enabled by “central facilities equipped with large scale computers capable of simultaneous parallel processing, connected with large capacity memory devices, a large number of program packages and extensive data bases” (Masuda, 1981: 75). In effect, his imagination could not go beyond open-hierarchy.

What the above reading of Masuda indicates is that it is much easier to visualize the transition from hierarchy to open-hierarchy than from open-hierarchy to un-order. We will first examine the first transition and then the second one.

Reflecting on the lack of success of videotex, Hooper (1985), chief executive of Prestel, placed electronic publishing services in two categories: “trigger services” and “incremental services.” The former had the power to attract potential customers to subscribe. The latter lacked this power but were likely to become popular after the customers subscribed to the service. According to Hooper:

The weather forecast epitomizes the notion of an “incremental service,” a service that would be used by existing subscribers but which would not by itself induce people to become subscribers. The “trigger service,” by contrast, is sufficiently powerful by itself to motivate acquisition of the necessary receiving equipment – to trigger new subscribers. (1985: 182)

Accordingly, there was frenetic hunt for a “trigger service” at Prestel and other videotex projects. But it remained elusive.

The breakthrough came not in the form of a “killer app” but a new form of architecture – open-hierarchy. The reason why the Minitel was a huge success was that it opened content provision to third parties, who were not limited to select contractors but included almost anyone who was interested in offering content. This bottom-up approach led to development of the very popular messageries conviviales service, which allowed subscribers to use pseudonyms and anonymously participate in conference calls. The most famous of them were the messageries roses for those looking for romance (Kirrmann, 1990). Looking back, Mr Grezes, of France Telecom, says, “We had assumed people might use Minitel just for administrative business – paying bills and so on. In fact, people developed all sorts of surprising ways of getting the most out of the system” (quoted in Arnold, 2003, online). This experience suggests that open-hierarchies work because they make room for bottom-up surprises.

After years of frustrating lack of success with the hierarchy, in spite of frenetic tinkering, the resounding success of the open-hierarchy was impressive. Many countries unsuccessfully tried to emulate Minitel. What is important is that in the pre-internet era Minitel was considered an exemplar. In a striking parallel, i-Mode was for a period considered an exemplar and many countries sought to emulate it.

The emergence of the internet was a radical departure from previous models. It was a surprise, especially so for the French. Even then, in 2001 Nicholas Duforcq, a divisional director of France Telecom, went so far as to say, “Our strategy has been to promote Minitel as a complementary service to Internet – not in direct competition to it” (McGrath, 2001). The grassroots Wi-Fi networks were similarly a surprise (Lehr and McKnight, 2003; Sawhney, 2003, 2005). Today the idea of an open cell seems preposterous but it may come to pass with all the hacker activity directed in that direction.

Why do we see this pattern?

Here’s the rub. Hierarchical design was the archetypal configuration during the era of mass production because of two primary reasons. One, the technological constraints were such that hierarchical design minimized the need for processing, which was difficult and expensive in the pre-digital environments. Two, the hierarchical design supported the institutional need for control. In the digital era cheap processing power has greatly mitigated the need for hierarchical design on technological grounds. Thus, as Nora and Minc noted, “if there is centralization, it is a matter of will, not of constraint” (1980: 55). The reality is that management, in many quarters, retains the desire for hierarchical control even though technological constraints have been greatly relaxed. But the dilemma management faces is that hierarchical control has its limitations, especially when there is the need for production of creative content on a large scale. The resulting lack of success with hierarchical systems induces the move to open-hierarchy and with it comes success. Open-hierarchy, however, marks the outer limits of the loosening up that can occur within a hierarchical configuration and correspondingly the limits of managerial imagination, especially when newfound success via open-hierarchy creates a sense of self-satisfaction after repeated failure with hierarchy. The next great advancement requires abandonment of hierarchy to un-order, which is brought about by bottom-up user activity and the resulting success, which is manifold bigger than that of open-hierarchy, is a total surprise.