1. Goodman et al. [1994] took an early look at Internet diffusion. See
Table 11, section IV, for a list of GDI studies and references. Check Wolcott
[2001] for new GDI work. 2. Archibugi et al. [1999] and Edquist [1997] provide
excellent introductions to the national systems of innovation (NSI) literature.
The GDI methodology can also be applied to a region larger than a country or to
sub-units within a country, and throughout this paper readers may substitute
region for country with no loss of generality. This is in keeping with similar
extensions to the NSI approach [Howells 1999]. We return to the NSI approach in
section III. 3. A much more complete map of the information industries is
provided by Houghton [1999], who defines 16 segments divided on the horizontal
axis by conduits to content (form to substance), and on the vertical axis by
products to services. 4. When data are available to do so, we exclude those who
have only UUCP, Fido, or any store and forward accounts. 5. Health, commercial,
and academic were initially selected because they corresponded to categories in
the United Nations Development Programme (UNDP) human development index. The
public sector is obviously a very important user, potentially comparable in size
to the others. Seehttp://www.undp.org/. 6. For each network, they derive the
cyclomatic number and the alpha, beta, and gamma, which are all ratios derived
using various formulas from the number of nodes and connections between nodes
(edges) in a graph. This is the underlying structure that comprises a computer
network. Gorman and Malecki [2000, pg. 120] provide more precise definitions. 7.
Van Slyke [1998] traces the concept of technology clusters to Silverman and
Baily [1961], whereas LaRose and Hoag [1996] cite Rogers [1986] as the primary
source. Technology clusters in information technology have been examined for a
least a decade [cf. Chin and Moore 1991]. King et al. [1994] discuss clusters in
the broader context of institutions and their role in innovation diffusion.
Prescott and Van Slyke [1997] recommended treating the Internet as a cluster and
summarize some evidence in support of this approach. Hahn and Schoch [1997]
emphasize that a cluster entails adoption of some, but not necessarily all, of
the constituent technologies. 8. Institutional views have been taken by Daly
[1999] and the "Internet Counts" project with which he has been affiliated
[Wilson et al. 1998]. The Leland Initiative Telematics for Africa project at the
Center for International Development and Conflict Management (CIDCM) at the
University of Maryland is developing an instrument for Internet assessment that
is highly focused on institutions and uses qualitative data gathering techniques
[bridges.org 2001]. 9. The GDI work grew out of a similar approach to measuring
the information systems capability of a nation, developed by Wolcott et al.
[1996]. Sharif [1988] provided an important example of technology assessment
measurement, and Liff et al. [1993] provided an important example of devising
dimensions. It is beyond the scope of this paper to formulate an overarching
theory of Internet diffusion and to relate it to other general models of
Internet diffusion which have already been developed. Daly [2000] has proposed a
conceptual framework that interrelates Internet penetration, utilization, and
impacts. His model includes many of the same dimensions and determinants as
contained in our model. Bazar and Boalch [1997] also put forth a general model
of Internet diffusion that has some similarities with our model. Abramson [2000,
pg. 70] characterizes our methodology as a "sophisticated adjunct" to
"country-by-country Internet user counts," which of course does not capture the
full richness of our dimensions and determinants. 10. A somewhat different focus
was taken by Basalla [1988], who related evolutionary analogies to the history
of technology. The growing field of evolutionary economics represents another
application of evolutionary analogies [cf. Nelson and Winter 1982]. 11. Their
work differs from ours, however, in that they were trying to characterize
information systems themselves. We are trying to find characteristics that
reflect the extent of adoption of an innovation (not the innovation per se). In
examining properties, functions, and the timing of the introduction of new
systems, Ein-Dor and Segev [1993] adopted a combination of phenetic,
teleological, and phylogenetic approaches that mirror the traditional methods of
evolutionary taxonomy. See Ridley [1986]. 12. An easily accessible summary of
his theory may be found in Rogers and Scott [1997]. 13. Larry Press provides a
web page with links to many of the Internet measurement efforts. See
http://som.csudh.edu/fac/lpress/GDIFF/GDIFFprojects.htm. Hal Varian also
maintains a page about Internet metrics at
http://www.sims.berkeley.edu/resources/infoecon/Accounting.html. An extensive
list of references to all aspects of Internet measurement is maintained by
Martin Dodge [2000] at http://www.cybergeography.org (also see Dodge [1998]. 14.
Buselle et al. [1999] discuss various approaches to Internet diffusion studies,
but all involve adoption as the dependent variable. One typical example of such
a study is Teo and Tan [1998]. 15. For example, the IDC/World Times Information
Society Index (ISI) aggregates available national-level indicators by category
into four measures: computer infrastructure, information infrastructure,
Internet infrastructure, and social infrastructure. The social infrastructure
consists of "Civil liberties, Newspaper readership per capita, Press freedom,
Secondary school enrollment, and Tertiary school enrollment." Sources include
IDC, UNESCO, ITU, World Bank, and Freedom House. These are aggregated to form a
single index. See World Times/IDC [2000]. This is a good example of a method
that does not directly measure Internet diffusion. 16. Sudweeks and Simoff
[1999] provide a means for integrating quantitative and qualitative approaches
in Internet research that is similar to the GDI methodology. Both ultimately
make interpretation central. 17. The GDI methodology has been used by other
researchers studying Iran [Adibi et al. 1999] and Uganda [Minges et al. 2000].
Numerous master's level papers have been written as well by various students of
some of the authors. 18. Teledatos mainly offered UUCP connections over an X.25,
but was upgrading their X.25 backbone in late 1997 so that they could run IP
over X.25. 19. Editor's Note: The following reference list contains hyperlinks
to World Wide Web pages. Readers with the ability to access the Web directly or
are reading the paper on the Web can gain direct access to these linked
references. Readers are warned, however, that 1. these links existed as of the
date of publication but are not guaranteed to be working thereafter. 2. the
contents of Web pages may change over time. Where version information is
provided in the References, different versions may not contain the information
or the conclusions referenced. 3. the author(s) of the Web pages, not AIS, is
(are) responsible for the accuracy of their content. 4. the author(s) of this
article, not AIS, is (are) responsible for the accuracy of the URL and version
information.