[Write] [Home] [Up] [Previous] [Next]


A Critical Examination of Adoption Rates

by Gisle Hannemyr

It is widely believed that the adoption rate of the Inter­net has exceeded that of earlier mass communication techno­logies by several magnitudes. This paper reviews the historic data related to some of these techno­logies, draws on actor-network theory as a framework for interpreting such data, traces the transformations and translation of this data in the public, political, and scientific discourse, and discusses the use of «facts» in modern society.

actor-network theory, diffusion of innovations, Internet, technology policy

In both popular media and scientific publications, the idea that the use of the Internet has grown exceptionally fast, at a rate exceeding that of radio and television by almost an order of magnitude, has been circulating for some time. The following quote, taken from the bestseller Successful Cybermarketing in a Week is a typical example:

Cyberfact: It took 38 years for radio to attract 50 million listeners. 13 years for television to attract 50 million viewers. In just 4 years the Internet has attracted 50 million surfers! Those figures can hardly be balked at, especial­ly when you consider the Internet's beginnings. (Gabay, 2000)

In January 2001, I started to look for this particular idea through various Internet search engines. I also asked colleagues and fellow participants in a Usenet news­group and an Internet mailing list if they had seen this particular idea. The response was overwhelming. Apart from assorted newspaper interviews where various gurus of the new economy have celebrated it, spottings cover a broad range of genres, encompassing advertise­ments (e.g. Money Making Ideas, 2001; SunSpring Properties LLC, 2001), politics (e.g. Margherio et al., 1998; Gerlach and Kohn, 1998; Mallet, 1999; Schjøtt-Pedersen, 1999; Patel, 1999; Asmal, 2000) and science (e.g. Braa and Sørensen, 1999; Biukovic, 2000).

Most of these publications do not cite any source. If they do, the reference is usually to a named individual, not a published source. In my collection of sightings, these particular figures for media adoption are attributed to more than a dozen different individuals. E.g.:

A friend of mine, Dale Cordell […] passed along this quote you may have heard from Bill Gates:  «The Internet is the greatest marketing tool since radio, telephone, or television. Internet growth substantiates this.  It took 37 years for radio to reach 50 million listeners, 34 years for the telephone to have 50 million customers, 13 years for television to reach 50 million view­ers and 4 years for the Internet to reach 50 million subscribers». (Zapoleon, 1999)

The absence of any specifics (i.e. time, place, and occasion) is typical[1].

Another characteristic is that geographical coverage of the data is usually omit­ted. When it isn't, the region varies while the numbers stay the same. E.g.:

A recent Goldman Sachs report on the Internet in Asia notes that while it took radio 38 years to attract 50 million listeners in the region, and tele­vision took 13 years to draw the same number of viewers, the Internet has already pulled in its first 50 million Asian users in just four years. (FGtA, 2000)

During a not very systematic bout of statement-spotting carried out by my­self and associates in January 2001, several hundred indivi­dual instan­ces like this were found, and while these individual instances vary widely in context and con­tent, they all seem to share a set of common characteristics: to affix the qualities of importance and urgency to the Internet phenomenon, and to make it imperative to adapt or risk being left out.

At the core of these narratives lies a set of data points: a series of quantifications that, it is claimed, reveals a particular measure of the adoption rates of specific media technologies.  This essay is an attempt to understand the meaning of these data points. In order to do so, I first present a theoretical framework in which the interpretation can take place, then I re-examine the actual data, and finally, I attempt to synthesise the findings into a revised narrative about what we can learn from this.

Innovation, Invention, and Adoption

The theoretical framework used in this essay is mainly taken from diffusion studies, social construction theory, and actor-network theory. It is presented in this section and the subsequent one.

An innovation is an idea, practice, or object that is perceived as new by some relevant social group which may consider adopting it. Invention is the process by which the innovation is discovered or created, while adoption is the process through which individuals belonging to a relevant social group take action to make use of the innovation (Bijker, 1995, p. 45ff).

The process of invention is characterised by enormous interpretative flexibility (Bijker, 1995, p. 76f). Conflicting theories and standards, laboratory contri­vances, parallel and partial prototypes, demonstration machines, and various types of experi­mental usages abound.

Sometimes, invention leads to failure or rejection (Bijker, 1995, p. 14f; Winston, 1998, p. 7; Rogers, 1995, p. 171), and eventually those that are adopted may have a long way to go before they bear fruit. The beneficial effect of citrus fruit as prevention for scurvy was established experimentally as early as 1601, but the British Navy did not adopt the practice of eating oranges and lemons on long sea journeys until 1795. In the intervening years, a vast number of different scurvy remedies were competing for attention (Rogers, 1995, p. 7f). Another example is the prototype radio transmitter demonstrated by David Hughes in 1880. It never became more than a prototype because his contemporaries could not see how it could be of any practical use, and interpreted Hughes' device as a mundane demonstration of the well known phenomenon of electrical induction (Winston, 1998, p. 68). Adoption can also be delayed by the absence of what Brian Winston calls a «supervening social necessity» (Winston, 1998, p. 6f). For example, Francis Ronalds demonstrated a fully working wire tele­graph for the British naval authorities in 1816, but the Royal Navy could not see how this device could replace semaphore in communication between ship and shore (Economist, 1999; Winston, 1998, p. 7). Only later in the 19th century, when commercial railway traffic created a need for a long distance signaling system to prevent trains from colliding on single track railway lines and Samuel Morse had perfected his coding system, did the process of inventing the wire telegraph transform into the process of adopting it (Yates, 1989, p. 23f).

For a successful innovation, the process of invention eventual­ly leads to what Wiebe E. Bijker calls closure (Bijker, 1995, p. 84ff)[2]. Closure means that the inter­pretative flexibility of the innovation diminishes, i.e. that the relevant social group reaches some sort of consensus about the dominant meaning of the inno­vation, inclu­ding such things as usages, characteristics, qualities, and standards.

Inscription, Convergence, and Irreversibility

It should be noted that the two processes of invention and adoption are not totally disjointed. The process of inven­tion involves many instances of adoption by at least some social group, and innovations are fre­quently subject to reinter­pretation and transformation during the adoption process[3].

This has indeed been the case with all the technologies discussed in the present paper (i.e.: telephone, radio, television, and Internet). The early telephones could only operate over a range of about twenty miles (Aronson, 1977, p. 27) and were not able to sustain two-way communication. There­fore, Bell first proposed that the telephone be used for one-way communication to transmit music, drama, and news to a listening audience. Even after the tech­nical pro­blems with two-way telephony had been resolved, one way transmission of news and entertainment conti­nu­ed to be a popular application of the telephone (Briggs, 1977; Marvin, 1988, p. 209-231). For the telephone to develop into today's global network service, it had to undergo numerous transformations. In the case of television, the development of color and the persistence of a plethora of non-compatible color systems (e.g. PAL, SECAM, and NTSC) also caution us against taking Bijker's notion of closure too literally. For the Internet, examples of recent transformations include the introduction of the World Wide Web in 1990, the current turmoil regarding broadband services, and the fracas that is likely to accompany the transition from the current version of the core protocols to IP version 6.

In addition to these and other technological transformations, political, and social entanglements frequently interfere with closure. In the United States, it took the government the best part of a century to sort out how to regu­late interoperability and competitive practices between long distance and local tele­phone carriers[4]. After the first public radio station in the United States commenced operation on November 2, 1920 (Gelman, 1995, p. 80), chaos follow­ed as there was no regulation in place to prevent broad­casters from interfering with each other, and this was not remedied until the Radio Act of 1927 set up the Federal Radio Commission (FRC) to oversee the industry (Smith et al., 1995, p. 37). A similar interference problem impeded the deployment of commercial television stations, and caused the Federal Freq­uency Commission  (FFC) in 1948 to institute a four year freeze on new trans­mitters (Winston, 1998, p. 119f).

The perpetual process of transformation and translation that apparently accom­panies any technology in-use has led to some controversy surrounding the con­cepts of stabilisation and closure. The words «stabilisation» and «closure» have too many connotations of finality and immuta­bility, and therefore fail to take into account the fact that technology keeps changing even after the point in time Bijker's «closure» occurs.

In actor-network theory, the notions of «closure» and «stabilisation» are there­fore replaced by a set of less definitive terms. Actor-network theory considers tech­ni­cal objects as scripts or programs of action co-ordinating a network of roles. These roles are played by the objects them­selves (telephones, radios, contracts, etc.), their supporting infrastructure, regulatory and financial frame­work (FRC/FFC, Department of Defence, venture capital, etc.) and the humans (inventors, entre­preneurs, users, sales­persons, performers, technicians, etc.). Bijker uses the term «sociotechnical ensemble» to denote this network of objects, infrastructures, and humans and the roles they play (Bijker, 1995, p. 273f). The central idea is that the innovation takes the part of what Susan Leigh Star (Bowker and Star, 1999) calls a boun­dary object[5] that is shared between these actors, who attempt to «inscribe» on it their visions of the object's mean­ing in the world (Akrich, 1992, p. 208; Latour, 1991). These scripts, which are a product of domination, negotiation, and mutual adjustment are mediated, translated, and even changed as time passes. At certain points in this process, there is some degree of agreement in this process. Michel Callon uses the term «convergence» (Callon, 1991) to measure the degree of agreement, and then proceeds to introduce «alignment» and «co-ordination» as two dimensions of convergence. Alignment measures the extent to which actors can agree on the translation. «A successful process of translation then generates a shared space, equivalence and commensurability» (Callon, 1991, p. 145). Co-ordi­nation measures the degree to which the interpretative flexibility is restricted by rules or conventions. «Strong co-ordination [refer to a network where the] universe of possible translations is relatively restricted, and network behaviour is relatively predictable» (Callon, 1991, p. 147). When a strongly aligned and co-ordinated network emerge, we may have a translation of the object's script where:

Callon (1991) calls this «irreversibility,» but hastens to add: «It is also a matter that is never fully resolved: all translations, however apparently secure, are in principle reversible» (p. 150).

For our practical purpose (i.e. to define the base year for the adoption process), there is little difference between the notions of stabilisation and closure as described by Bijker, and the notions of convergence and irreversi­bility as described by Callon.

What we are looking for, is a point in time with significant decrease in interpre­tative flexibility, increasing alignment, and strong degree of co-ordination. For all technologies, several candidate dates exist.

For example: Alexander Graham Bell and his financial backers formed the Bell Patent Association on February 27, 1875 and were awarded its first patent (called «Improvement in Telegraphy») the next day. On March 10 the following year, Bell was able to demonstrate a working device. In April 1877, Bell sold his first two telephones to businessman Charles Williams, Jr., who also owned the workshop where Bell conducted his experiments. To use the phones, Williams strung a private line between his home in Somerville and the workshop in Boston (Winston, 1998, p. 53). Under Bell's initial lease agreement, the subscriber was only allowed to use the telephone to connect to exactly one other party (Aronson, 1977, p. 23). The service was advertised as follows:

The terms for leasing two telephones for social purposes connecting a dwelling house with any other building will be $20 a year; for business pur­poses $40 a year, payable semi-annually in advance. (Bell 1877 adver­tisement, cited in Winston, 1998, p. 53)

The first telephone exchange was set up in Boston on May 17, 1877, but it did not provide the customers it connected with the opportunity to talk to each other. Instead, it was used to receive orders at a central location that were retransmit­ted to a general express agency (Aronson, 1977, p. 24). The first commercial telephone exchange where subscribers could call each other on request opened at New Haven, Connecticut on January 28, 1878 (Brown, 1991).

Eventually, a service over distances greater than 20 miles was introduced.  After technical obstacles were overcome, there were problems in the business model. In 1885, the Bell affiliated Southern New England Telephone Company announced it was shutting down its 200-mile segment of the long-distance toll line between Boston and New York City, even though the line was a technical success. The regional company did not make enough money on this segment to justify its operating costs (Brown, 1991).

As innovations go, radio is even more complex than the telephone. Without going into the same level of detail as the discussion on telephony, it can be argued that radio was invented in 1894, when Oliver Lodge demon­strated his wireless telegraph (Winston, 1998, p. 69).  However, the adoption of radio as a medium of mass communication in the United States did not take place until November 1920, when KDKA, the first public radio station in the United States commenced ope­ration (Schramm, 1949, p. 547-552; Winston, 1998, p. 77). In the meantime, radio had seen a number of appli­ca­tions, such as wireless telegraphy between ship and shore, communica­tion between military units in the field during WWI, and as an alternative to the telephone for person-to-person calls.

Similarly, the first patent for «television» was filed in 1911, but prior to that, a number of ingenious devices for «telephotography», «telescopy,» and «teleauto­graphy» had been demonstrated (Winston, 1998, p. 94). Eventually, electronic proto­types started to appear, but early television was plagued by standards «wars» that significantly delayed public adoption. Then, shortly after the FCC approved the NTSC standard in July 1941, the United States entered WWII, which immediately put further development of commercial television in a hiatus. This postponed the start of commercial television in the United States until 1945 (DeFleur, 1973, p. 83).

It can be argued that the Internet came into existence in 1964 with the invention of packet switching by RAND researcher Paul Baran; or 1969, with commence­ment of operation of the ARPAnet; or in 1983, with the introduction of the Internet Protocol; or in 1989, when commercial ISPs (Internet Service Provid­ers) such as AlterNet and PSInet started to provide services to the general public (Abbate, 2000, p. 197f); or in 1990, when the World Wide Web and the first Internet web browser debuted, or in 1991 when the Commercial Internet eXchange (CIX) Association, Inc.  was established (Zakon, 2001), or in 1993, when versions of the Mosaic graphical web browser for home computers running the popular Microsoft Windows or Apple Macintosh operating systems was released by NCSA (Gillies and Cailliau, 2000, p. 241).

From the above choice of dates, which one best reflects the point in time when, in the USA, these technologies reached the stage where their inscriptions may be said to be irreversible? There is, of course, no single «right» answer, the following dates are those I believe best meet Callon's criteria of irreversibility.

For telephone, radio, and television, this is also when the technologies has almost no «real users» (i.e. less than ten thousand individuals, almost all of them entrepre­neurs, experimenters, and implementers).

In 1989 there were already around 400 000 Internet users in the USA. While an Internet connection was not commercially available prior to 1989, a sizeable user community connecting to machines located at universities and research laboratories existed. In addition, between 1994 and 1995 a number of online services that predate commercialisation of the Internet by several years were connected to the Internet, bringing their huge, pre-existing user bases to the Internet. CompuServe, for instance, was established in 1979 and had in 1994 grown to 3.2 million subscribers, rival America Online had at the same time 3.5 million subscriber, and Prodigy 1.4 million subscribers. (Winston, 1998, p. 333). This totals to around 8 mil­lion subscribers[6].

Counting Users

Identifying the year of irreversibility is not the only difficulty we face when comparing telephone, radio, television, and Internet data. Equally problematic is the aspect of finding and extracting reliable data about users from historic records.

Ideally, estimates would be based on comprehensive and identi­cal surveys conducted on a regular basis over many years. Unfortunately, such data does not exist. Instead there exists a variety of data from many different sources (government agencies, market research companies, industry associa­tions, annual reports, directories, etc.) using different methodologies and measuring different things. Some of these data are good, some are bad, and some are downright misleading. It takes a major effort and knowledge to determine what is good information and what is poor data and should be discarded.

In this paper, the method for counting users is to first collect all avail­able data from various sources, then to carefully consider each data set and see how it fits with the others. A consistent and comparable set is selected and de­sig­nated authoritative. The remaining datasets are designated supplemen­tary.  The present study only considers data from the United States. The United States is the only major geographical region for which sufficient data sets are available to make this of type exercise possible.

Whenever possible, I have tried to rely on data from U.S. Bureau of Census (BoC) decen­nial census of housing tables. These are collected every ten years from all households in the U.S. by skilled statisticians. The supplementary data sets are used to interpolate trends between the data points that make up the BoC data.

To supplement the BoC housing data on radio and television usage, I've extract­ed historic data from a number of different sources, based upon surveys and industry statistics (Schramm, 1949; DeFleur, 1973; Froehlich and Kent, 1991; Smith et al., 1995; Gelman, 1995; Winston, 1998; Famighetti, 1999). These sup­ple­men­tal data are used to confirm the BoC data, and to interpolate the years between the decennial tables.

The BoC decennial housing data does not measure telephone ownership prior to 1960. Early telephone figures are difficult to find and the available data sets of telephone ownership from the last century are not broken down into residen­tial and business phones, by demographics, or by region. As long as the Bell Tele­phone Company was a monopoly, the number of subscribers of its services was a good indicator of number of users, but after the main Bell patents expired in 1893 a number of small competitors sprang up making it difficult to keep track of developments. The following sources have been used to extract figures for num­ber of households in the United States with a telephone prior to 1960: (Iardella, 1964; Pool, 1977; Marvin, 1988; Froehlich and Kent, 1991; Winston, 1998).

The BoC does not include Internet use in its decennial census, but it regularly conducts a survey (Rohde et al., 2000) where approximately 48 000 sample households selected from the 1990 decennial census files are inter­viewed about computer and Internet use. My number for households with an Internet connection in the United States is primarily based upon these surveys. As supplementary data, I've used a number of different market surveys (summarized on NUA Ltd., 2000) of Internet usage in the United States. Another supple­mentary source for this data is the quarterly updates published by Matrix Information and Directory Services (e.g. MIDS, 1998) and the Network Wizards domain survey (Network Wizards, 2000). The latter indicate the number of host computers connected to the Internet (not users), which can be construed as an indicator of growth rate.

All the authoritative statistics mentioned above resolve into households, not individual users. This means that we need to convert house­holds into individuals. The conversion is car­ried out in the simplest possible manner, by multi­plying the number of house­holds with the average household size (as reported by the BoC) for the year in question. This method counts everyone in a house­hold (actual users and non-users) as a «user.» It also ignores any skewed distribution of devices between large and small house­holds. This means that the resulting figures should only be understood to be rough indicators of the rela­tive levels of adoption at various times, not exact measures of actual number of individual users. It is difficult to estimate exactly the bias introduced by the method's inability to distinguish between users and non-users in a household, but it obviously results in some over­estimation. Assuming that it is likely that there are many more non-users of a difficult-to-use medium (e.g. the Internet and the telephone) compared to a simpler-to-use medium (e.g. radio and tele­vision), this method will overestimate the adoption rates of the Internet and the telephone more than those of the radio and the tele­vision. It could be argued that more educated households have fewer members and until recently were the most likely to have access to the Internet, further skewing the numbers.

The estimated figures are shown in table 1 below. For each of the four techno­log­ies, there is designated a base year B, which is the year we start counting uses.

B+0 0 0 0 0.4 million
B+5 0.6 million17 million 16 million 6.1 million
B+100.9 million56 million 80 million 79 million
B+151.2 million86 million 142 million 
B+203.8 million99 million 161 million 
B+2510 million 115 million179 million 
B+3019 million 133 million   
B+3532 million 137 million   
B+4042 million 149 million    
B+4559 million      
TABLE 1: Number of users at year B+x

The table shows that of these technologies, only the telephone requi­red consid­erable time (approximately 43 years) to acquire 50 million users in the United States. All the others reached that benchmark in less than ten years.

Figure 1 plots the adoption rate in the United States for the first ten years for radio, television, and the Internet.  The graph indicat­es that the early adoption rates for all three media are roughly of the same order of magnitude.

FIG. 1: Technology adoption, number of users

This becomes even more obvious if we, instead of plotting the absolute number of users, plot usage as a percentage of the total population. Because radio and television was adopted at times when the population of the United States was lower than it is now, this indicates that the adoption rate for the Internet actually is lagging behind these technologies. The adoption rate for the telephone, as already noted, is considerable lower (figure 2).

FIG. 2: Technology adoption relative to population

Revisiting the Idea

It seems clear from these data that the Internet did grow as rapidly as several other mass technologies shown, but this still doesn't bear out the popular conception that the Internet growth was somehow exceptional or significantly outside the bounds of prior experience. How, then, can such an idea rise and spread?

The earliest example in my collection is a graph appearing in an extensive report about «the future of web-based retailing» from investment banker Morgan Stanley (Meeker and Pearson, 1997, p. 2-2). It is reproduced as figure 3 below.

FIG. 3: Media adoption, reproduced from the Morgan Stanley report

The report cites «Morgan Stanley Technology Research» as the source[7] for this data. The graph shows radio as starting in 1922 and growing more or less line­arly. Television is shown starting in 1950 with a steep initial adoption rate, and then tapering off from 1956.  According to this graph, in 1960 there were around 50 million radio users and 45 million television users in the United States.

One staple trick of the engineering trade is the «back of an envelope» calcula­tion. This refers to using a trivial computation as a means of checking if some data makes sense. If we take the Morgan Stanley graph and re-compute its 1960 data points as population percentages, we find that it asserts that in 1960, 28% of the U.S. population used radio and 25% used tele­vision.  This is at odds with all other statistic available on media use at that time E.g.: the biannual Roper Organization Surveys (cited in Castells, 1997, p. 313).

Another puzzling feature in this graph is that the birth of the Internet is set to somewhere around 1994, and the number of users at that point is indicated to be very close to zero. According to MIDS, the number of U.S. Internet users at that time was more than 6 million (MIDS, 1995).

I have no way to explain the huge discrepancies between the adoption rate data presented in the Morgan Stanley and all other data I have about the number of users of these technologies at these times. To cut a long story short: I believe that it is very obvious that the adoption curves for radio, television, and the Internet presented by Morgan Stanley are balderdash. I have not checked up upon their figures for cable.

In April 1998, U.S. Department of Commerce (DoC) translated the graph in the Morgan Stanley report into the following statement:

Radio was in existence 38 years before 50 million people tuned in; TV took 13 years to reach that benchmark. […] Once it was opened to the general public, the Internet crossed that line in four years. (Margherio et al., 1998, p. 4)

The Morgan Stanley report (op. cit.) is cited as the source for the radio and television adoption rate estimates. As for Internet, the DoC report shortens the time span from five years to four years, and changes the base year from 1994 to 1993.  This is explained as follows:

In 1993, the alpha version of Mosaic, the graphical user interface to the WWW, was released, giving non-technical users the ability to navigate the Internet. This report uses 1993 as the date when the Internet became truly open to the public. […] No exact figures exist on Internet usage worldwide, but different sources point to 1997 as the year when Internet usage approaches/crosses the 50 million mark. For instance, NUA, an Internet consultancy and developer, compiles figures from different research analysts and finds the following ranges of Internet usage: 1995: 8-30 million, 1996: 28-40 million, 1997: >100 million. (Note: some research groups report U.S. figures only. Global figures for 1995 and 1996 were derived from NUA estimates on U.S. Internet usage as a percent of global Internet usage.) (Margherio et al., 1998, p. 53f)

Within days of the publication of the DoC Report in April 1998, adoption rate data with the signature 38, 13 and 4 years for radio, television and the Internet started to appear in media in the United States.

Then, in October 1998, the data jumped the Atlantic. In a report prepared for the European Commission (EC). This report copies a lot of material, uncredited, from the DoC report (op. cit.), including the following:

Radio existed for thirty-eight years before it reached a penetration of 50 million listeners; television took thirteen years to reach 50 million viewers. […] In contrast to these other successful technologies, after becoming available to the public, the Internet required only four years to reach 50 million users. (Gerlach and Kohn, 1998, p. 10)

Having established itself in government-sponsored reports on both sides of the Atlantic, the idea now starts to mutate. For example: In January 1999, in an attach­ment to the Observer newspaper, the following statement appeared:

The net is the most profound change in our communications environment since the invention of print. Just compare it to other media: Radio took about 37 years to reach its first 50 million listeners, while television took 15 years to reach the same target. The world wide web took just over three years. (Observer, 1999)

The most interesting change here is the transformation of the «Internet» into «the world wide web» and the year that is consequently subtracted from the time it took it to reach the benchmark 50 million users. Of course, as evident from the note accompanying the DoC report, the release of the Mosaic web browser in 1993 is the event used to identify the base year of the Internet (i.e. the Internet and the world wide web is considered identical for the purpose of computing this adoption rate). But here, somebody who is aware that there is a difference has noted that while the Internet became commercially available in 1989, the World Wide Web did not appear until 1990, and deducted the intervening year to make it «right» before presenting the story to the public.

Then, the data in the Observer article is given another spin by a group of scientists. Neither the EC report nor the Observer articles give any clue about what geographical region the findings apply to, but the Morgan Stanley report where it all started stated explicitly that its data referred to the United States. When the Observer article is translated into scientific prose, its precision level is «improved» and its geographical coverage changed to be appropriate to the global scope inherent in the title of book, Planet Internet.

It took radio 37 years to gain 50 million listeners world-wide. Television only required 15 years to gain 50 million viewers. The World Wide Web acquired 50 million surfers within 3 years. (Braa and Sørensen, 2000, p. 21, citing the above mentioned Observer article as source)

Science is Politics by Other Means[8]

I am amazed by the places this particular idea has managed to go after being released to the public in 1997, astounded by the sheer rate of imitation and amused by the number of individuals who have seen fit to work it into their conversation.

More interesting, however, is how these data are enrolled[9] in political discourse in support of various political agendas. The DoC and the EC reports are examples of such usages, that argue that the alleged exceptional adoption rate of the Internet, makes it, and the affiliated phenomenon «the digital economy» into something that needs special conside­ration and political support.

In a parliamentary debate in Norway in 1999, representative Karl Eirik Schjøtt-Pedersen emphasised the importance of government being «technology-friendly», citing the exceptional adoption rate of the Internet to lend support to this notion (Schjøtt-Pedersen, 1999).

A presentation for the House of Com­mons in the United Kingdom by expert Alpesh Patel follows the same pattern:

E-commerce presents an exceptional opportunity to benefit the nation through job creation, competitiveness, tax revenue, wealth creation. […] There is ample evidence to suggest the significant impact of e-commerce on indu­stry, trade and the economic health of Britain. However, the impor­tance is still underestimated by many. Estats reckoned a year ago that global internet traffic doubles every two days. It took radio 38 years to reach 50 million listeners, television took 13 years to reach 50 million users. The internet achieved 50 million users in four years. […] If we are to create e-commerce giants of the future to benefit Britain, urgent action needs to be taken now. (Patel, 1999)

Policy, however, does not usually emanate from facts alone. Instead, concerns, ideologies, and prejudices (among other things) shape policy, and facts are then enrolled to support whatever solution that emerges.  Therefore the recurrence of the idea that Internet has grown exceptionally fast in the public policy debate should not be taken as evidence that it has actually informed the policies in question.  It is only an indicator of its popularity as an argument that is perceived to be self evident.  This popularity may in part explain why it is repeated so frequently, and has been able to survive in the public discourse unchallenged for so long.


In this paper I've examined, in some detail, a popular statement that suggests that the adoption rate of the Internet is exceptionally high compared to that of radio and television.

  1. I have demonstrated that the data used to support this argument are tenuous, and that the interpretation of this data is further complicated by the fact that neither invention nor adoption are clear-cut events (in fact, they are processes).
  2. I have shown, through careful compilation and examination of the historical record, that there appear to be no major differences between the adoption rate of the Internet and the patterns of adoption we know from radio and television in the past.
  3. I have identified the discrepancy in adoption rates as possibly resulting from trans­lations of actual data into carefully phrased re-statements of fact in support of a particular financial or political agenda.
  4. I have argued that this particular misconception was able to ingrain itself in popular, political, and academic discourse because of our trust in, and reliance upon, media-constructed reality.

As a final paradox, I would like to state that I probably would not have been able to write this essay without the Internet. The majority of the examples used to illustrate this case study were located through Internet search engines or with the help of individuals that I corre­spond with through the Internet. In most cases, I was also able to find the reports and articles cited online, or to order paper copies by means of online order forms.

The Internet has merit; it just can do without the hyperbole and ballyhoo.


I have not been able to verify that Bill Gates ever said this.
Bijker points out that a semiotic component, called «stabilisation», accompan­ies the social component of «closure». He also says that these are «two sides of the same coin» (op. cit).
Rogers (1995) uses the word «re-invention» to emphasise the «invention» aspect of transformations that results from adoption attempts.
The Bell System was finally divested on January 1, 1984 (Brown, 1991).
An object that inhabits several communities of practice and satisfies the informational require­ments of each of them.
There was probably some degree of overlap between the subscriber bases, so the total number of distinct subscribers is underestimated. On the other hand, it was not uncommon for several members of a household to share a single subscription, so the number of distinct users was probably greater than the number of subscribers.
I've tried to contact the authors to learn more about this source. There has been no response.
This particular twist on Prussian general Karl von Clausewitz' «War is politics by other means» was originally coined by Bruno Latour (Latour, 1988, p.  229).
The notion of «enrolling» an actant (such as a particular data set) in a particular campaign is borrowed from of actor-network theory. For a review of ANT vocabulary see Akrich and Latour (1992).


Abbate, Janet, 2000. Inventing the Internet, MIT Press, Cambridge,  Massachusetts.

Akrich, Madeleine, 1992. The de-scription of technical objects. In: Shaping technology/building society, (Eds. W. E. Bijker and J. Law), MIT Press, Cambridge, MA, pp. 205-224.

Akrich, Madeleine and Bruno Latour, 1992. A Summary of a Convenient Vocabulary for the Semiotics of Human and Nonhuman Assemblies. In: Shaping Technology/Building Society: Studies in Sociotechnical Change, (Eds. W. E. Bijker and J. Law), MIT Press, Cambridge, Mass., pp. 259-264.

Aronson, Sidney H., 1977. Bell's Electrical Toy: What's the Use? The Sociology of Early Telephone Usage. In: The Social Impact of the Telephone, (Ed. I. d. S. Pool), The MIT Press, Cambridge, Massachusetts.

Asmal, Kader, 2000. Speech by Minister of Education to the American Chamber of Commerce of South Africa, In: American Chamber of Commerce, Houghton, Johannesburg, October 27.

Bijker, Wiebe E., 1995. Of Bicycles, Bakelites and Bulbs. Towards a Theory of Sociotechnical Change, MIT Press, Cambridge, Massachusetts.

Biukovic, Ljiljana, 2000. Unification of Cyber-Jurisdiction Rules: Just How Close Are the EU and the U.S.?, In: Regulating the Internet: EU & US Perspectives(Eds, R. Kielbowicz and A. Ramasastry) University of Washington, Seattle, Seattle, WA, April 27-29.

Bowker, Geoffrey C. and Susan Leigh Star, 1999. Sorting things out. Classification and its consequences, MIT Press, Cambridge, Massachusetts.

Briggs, Asa, 1977. The Pleasure Telephone: A Chapter in the Prehistory of Media. In: The Social Impact of the Telephone, (Ed. I. d. S. Pool), The MIT Press, Cambridge, Massachusetts, pp. 40-65.

Brown, Charles L., 1991. The Bell System. In: The Froehlich/Kent Encyclopedia of Telecommunications, Vol. 2 (Eds. F. E. Froehlich and A. Kent), Marcel Dekker, Inc, New York.

Braa, Kristin and Carsten Sørensen, 1999. The Internet Factor, Scandinavian Journal of Information Systems, 10(1&2):235-240.

Braa, Kristin, Carsten Sørensen and Bo Dahlbom, 2000. Changes. From big calculator to global network. In: Planet Internet, (Eds. K. Braa, C. Sørensen and B. Dahlbom), Studentlitteratur, Lund, pp. 13-39.

Callon, Michel, 1991. Techno-Economic Networks and Irreversibility. In: A Sociology of Monsters. Essays on Power, Technology and Domination., (Ed. J. Law), Routledge, London, pp. 132-161.

Castells, Manuel, 1997. The Power of Identity, Blackwell Publishers, Oxford.

DeFleur, Melvin L., 1973. Massekommunikasjon og sosial endring. In: Massekommunikasjon, (Ed. H. F. Dahl), Gyldendal, Oslo.

Economist, 1999. Talking to the World, The Economist, December 31.

Famighetti, Robert, (Ed.) 1999. The World Almanac and Book of Facts 2000, Primedia Reference Inc., Mahwah, New Jersey.

FGtA, 2000. Internet grows faster than ever in Asia, [2001-01-08], last updated: October 23, Financial Gateway to Asia, (newsletter) <http://www.financialgatewaytoasia.com/articles/2000-10/2000-10-23-internet.html>.

Froehlich, Fritz E. and Allen Kent, 1991. The Froehlich/Kent encyclopedia of telecommunications, Marcel Dekker, Inc., New York.

Gabay, J. Jonathan, 2000. Successful Cybermarketing in a Week, Hodder & Stoughton, London.

Gelman, Morrie, 1995. 75 Years of Pioneers: A Personalized History of the Fifth Estate from Frank Conrad to Rupert Murdoch, Broadcasting & Cable.

Gerlach, Charles and David Kohn, 1998. Content and commerce driven strategies in Global Networks. Building the Network Economy in Europe, October, Office for Official Publications of the European Communities, Luxembourg.

Gillies, James and Robert Cailliau, 2000. How the Web was born: The story of the World Wide Web, Oxford University Press, Oxford.

Hannemyr, Gisle, 1999. Begynnelsen på en historie om Internett. In: Nettsamfunn, (Eds. K. Braa, p. Hetland and G. Liestøl), Tano-Aschehoug, Oslo, pp. 11-27.

Iardella, Albert B., 1964. Western Electric and the Bell System. A Survey of Service, Western Electric Company, New York.

Latour, Bruno, 1988. The pasteurization of France, Harvard University Press, Cambridge, Mass.

Latour, Bruno, 1991. Technology is society made durable. In: A Sociology of Monsters. Essays on Power, Technology and Domination., (Ed. J. Law), Routledge, London, pp. 103-131.

Mallet, Robert, 1999. Keynote Address, In: APEC (Asia-Pacific Economic Cooperation) Ministers Conference, Christchurch, New Zealand, April.

Margherio, Lynn, et al., 1998. The Emerging Digital Economy, April, U.S. Department of Commerce, Washington, DC.

Marvin, Carolyn, 1988. When Old Technologies were New. Thinking About Electric Communication in the Late Nineteenth Century, Oxford University Press, Oxford.

Meeker, Mary and Sharon Pearson, 1997. Morgan Stanley U.S. Investment Research: The Internet Retailing Report, May 28, Morgan Stanley, New York.

MIDS, 1995. Matrix Maps Quarterly, July 1995, July, Matrix Information and Directory Services.

MIDS, 1998. Matrix Maps Quarterly, January 1998, July, Matrix Information and Directory Services.

Money Making Ideas, 2001. Free Internet Money Making Ideas and What You Can Sell, [2001-03-02], Money Making Ideas, (advertisement) <http://www.money-making-ideas.net/index2.htm>.

Network Wizards, 2000. Internet Domain Survey, July 2000, Network Wizards, (web page) <http://www.isc.org/ds/WWW-2000index.php>.

NUA Ltd., 2000. How Many Users, [2000-07-31], NUA Ltd., (web page) <http://www.nua.com/surveys/how_many_online/>.

Observer, 1999. Guide to the Internet, Supplement to The Observer, London, January 17, p. 31.

Patel, Alpesh B., 1999. Appendix. 40. Memorandum submitted by Mr. Alpesh B. Patel. Trade and Industry - Seventh Report, March 10, House of Commons, London.

Pool, Ithiel de Sola, (Ed.) 1977. The Social Impact of the Telephone, The MIT Press, Cambridge, Massachusetts.

Rogers, Everett M., 1995. Diffusion of Innovations, The Free Press, New York.

Rohde, Gregory L., et al., 2000. Falling through the Net: Towards Digital Inclusion, October, U.S. Department of Commerce, Washington, DC, p. 122.

Schjøtt-Pedersen, Karl Eirik, 1999. Innstilling fra samferdselskomiteen om enkelte regulatoriske spørsmål i telesektoren, June 3, Stortinget, Oslo.

Schramm, Wilbur Lang, 1949. Mass Communications; a book of readings selected and edited for the Institute of Communications Research in the University of Illinois., University of Illinois Press, Urbana.

Smith, F. Leslie, Milan D. Meeske and John W. Wright, 1995. Electronic media and government: the Regulation of Wireless and Wired Mass Communication in the United States, Longman Publishers, White Plains, N.Y.

SunSpring Properties LLC, 2001. How to make money in real estate, SunSpring Properties LLC, (advertisement) <http://www.angelfire.com/ut/bevshome/makemoney.html>.

Winston, Brian, 1998. Media Technology and Society. A History: From the Telegraph to the Internet, Routledge, London.

Yates, JoAnne, 1989. Control through Communication. The Rise of System in American Management, The John Hopkins University Press, Baltimore.

Zakon, Robert Hobbes, 2001. Hobbes' Internet Timeline, [2001-08-23], last updated: 2001-08-23, Robert Hobbes Zakon, (web page) <http://www.zakon.org/robert/internet/timeline/>.

Zapoleon, Guy, 1999. Internet, Satellite Radio & Consolidation: The Effects on Radio, [2001-03-02], last updated: September, Zapoleon Media Strategies, (report) <http://www.zapoleon.com/knowledge/gz/internet_consolidation.html>.


In January 2001, I posted a request for «Internet adoption rate» sightings on a mailing list and a newsgroup I subscribe to. Thanks are due to fol­lowing individuals who were kind enough to respond: Espen Andersen, Nick Austin, Paolo Ceresa, Dalip Dewan, Olav Arne Forbord, Pet­ter Gott­schalk, Ola Fos­heim Grøstad, Håvard Hegna, Knut Hegna, Jo Herstad, Tor J. Larsen, Wolfgang Leister, Knut Lundby, Grethe Mel­by, Bjarne Nærum, Kai A. Olsen, Ragnvald San­nes, Ingjerd Skogseid, Ivar Solheim, Hilde­gunn Sæthre, Morten Søby, Carsten Sørensen, Gro Malnes Øvrebø and Finn Arve Aagesen. Thanks are also due to Bonnie L. Damon of U.S. Bureau of Census, who was kind enough to extract the appropriate historic figures from the bureau's decennial statistics; and Patricia Buckley, of the Economics and Statistics Administration, U.S. Department of Commerce, who helped me locate the report where it all started. And finally, a big «thank you» to Carsten Sørensen, who went out of his way to dig out the lost Observer supplement after my library had given up on locating it, to Hans Fredrik Dahl, Sundeep Sahay and Fred-Arne Øvergaard for interesting discussions, Joan Greenbaum, Eric Monteiro, and Eline Vedel who read an early draft and as usual provided helpful comments and insights, and the three anonymous reviewers who contributed helpful comments and corrections through their reviews.

Creative Commons License First published in The Information Society, 19:2, p. 111-121, April-June 2003
Copyright © 2003 Gisle Hannemyr. Some rights reserved. This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 3.0 License.

[English] [Norwegian]
[Write] [Home] [Up] [Previous] [Next]