The Global Diffusion of the Internet: Patterns and Problems

S.E. Goodman
L.I. Press
S.R. Ruth
A.M. Rutkowski 

If the Internet were a stock it would be considered a market phenomenon, with sustained double digit growth and no apparent end in sight to the upward spiral. Recent Internet numbers are stunning. Between January 1993 and January 1994, the number of hosts grew from 1,313,000 to 2,217,000; an impressive 69% increase[7]. Over 70 countries have full TCP/IP Internet connectivity, and about 150 have at least e-mail services through IP or via more limited forms of connectivity, e.g., UUCP or Fidonet. Monthly traffic on the US NSF backbone alone is about 10 terabytes[1].

But behind these statistics of overwhelming successes there are some sobering concerns about the international and other forms of diffusion of this exciting technology. Surprisingly little is known about this diffusion beyond the basic macro-statistics. We do know that about one fifth of the world's population, estimated only on the basis of the countries they live in, has access to far more capability than the rest. There is a direct correlation between measures of national development and the quality of network services available to advanced industrial and less developed countries (LDCs) [10]. The connectivity to many LDCs is often only a Fidonet link to a few PCs with less than a dozen regular users. Even within the most advanced, well connected countries the majority of the populations have little or no participation. For example, despite the much proclaimed connectivity of US universities, where access is almost a free good, it appears that only a small fraction of professors are users at most schools, and there may be significant differences in levels of use across academic disciplines.

In spite of the magnitude of the phenomena, relatively little has been written about the patterns and problems of network diffusion, especially in relation to LDCs. We describe several factors that facilitate or retard network diffusion, and conclude with brief statements of several other apparent disparities.

Patterns of Network Diffusion

We have identified a few models which, taken either in pure form or in hybrid combinations, appear to cover most of the forms of international proliferation.

Building national backbones

This pattern is characterized by active support from governments to build and maintain national backbones, and to help provide gateways to other nations. This model is most often pursued in advanced industrial countries, and frequently through the creation of research and academic networks, e.g., in Australia, Chile, Japan, South Africa, and Switzerland. France has notably used this model with a more general backbone serving the larger public. These backbones are usually augmented in various ways over time (see the other models below), and several support gigabit monthly traffic loads.

For most countries, such national backbones have been outside the range of financial resources or of national priorities. Nevertheless, modest variants of backbone building (or at least extensive local networking) have been made possible through the initial efforts of some enthusiasts in a few LDCs, notably Costa Rica and Zambia. In Costa Rica's case the diffusion process was characterized by what could be called a "demand pull" trajectory. As new connections were made, grants and other donations were sought to build necessary infrastructure. Costa Rica's development involved a series of steps that have required incremental improvements in national capacity before diffusion to partner institutions. In Zambia, the deputy minister of health and all the provincial hospitals have been linked together through an initiative at the University of Zambia. This net has been extended to other users, and is one of the more successful efforts to build a networking constituency in sub-equatorial Africa.

International and regional initiatives

Many international agencies invest in regions or in specialty areas that can gain from networking capabilities. For example, the United Nations Develolpment Programme has robust Internet connectivity in over one hundred nations. The Organization of American States (OAS) sponsors a variety of low-cost programs that encourage the development of networking in Latin America. There are also national agencies which sponsor international initiatives to promote networking. One such agency is USAID, which spends many millions on improving networking for specific applications and on telecommunications more generally, e.g., energy and hunger studies, population research, and building telecommunications infrastructure. The World Health Organization and UN Disaster Relief Organization, among others, also support specialized networks.

Non-Government Organizations (NGOs), e.g., non-profit organizations providing technical or other forms of aid to LDCs, are also creating new initiatives in networking. But most of the nets used by NGOs are fairly primative.

Another important model for regional diffusion is that of a local leader emerging and drawing neighbors into its orbit. There are noteworthy examples in Africa and Central America. CERPOD, a center for applied research on demography and development in Bomako, Mali, works with nine Sahelian countries in the Permanent Interstates Committee for Draught Control. CERPOD is attempting to be the leader in setting up e-mail capability for itself as well as its colleague countries. Costa Rica has become a regional hub offering services to other countries in Central America, most recently Nicaragua. In time, with improved local technical skills and increased use, the "orbited" counties tend to wean themselves from the regional leader.

Grass root nets

Many networks have emerged in LDCs where IT conditions are characterized by poor telephone lines, unsympathetic PTT officials, a paucity of hardware and software, and many other problems (see below). Yet people in these countries have still found it possible to put together network links of various kinds. For example, some of these connections are made through packet radio. Even in the middle of a desert or jungle, a radio with an inexpensive terminal node controller can connect to a low orbiting satellite and send messages on the Internet.

The best known and most venerable enabling technology for grass roots connectivity is Fidonet, which has been in existence for about ten years. Fido sites are found in the smallest towns and in the poorest countries. The basic idea is that of a simple protocol for storing messages and a powerful error checking mechanism in software for assuring that the desired message is transmitted correctly. Fido can even cope with the local PTT's not-so-tender clutches by operating in a mode that allows messages to be polled from a site in another country. Volunteers in Technical Assistance (VITA), a Washington-based NGO, polls messages each day from offices and project sites in Africa. The Institute for Global Communications, a San Francisco-based non-profit, uses a similar service for UUCP nets in Latin America, Eastern Europe, and elsewhere. Both NGOs find that this method works better than fax, offers cheaper and more reliable traffic, and encourages more persons to connect.

Yet even this most appropriate techniology for LDCs is spreading more extensively among the richer countires. In mid-1993, there were 12,641 Fido nodes in North America (up 33% from the previous year) and only 95 (up 21%) in Africa[9].

Commercial carriers and resellers

Commercial carriers can be important catalysts or significant burdens to the diffusion process. The national PTTs are the major arbiters of the costs of networking services, and in many countries the cost of network equipment and other infrastructure has been made very high (see below).

On the other hand, some value added network (VAN) providers, like CompuServe, have done much to promote and extend networking to substantial groups of users around the world. In a few cases the connectivity of countries is tied more to being able to piggyback on the networking infrastructure of businesses than to national policy agendas. The recent explosion of connectivity in Ecuador is as much due to the Banco Pacifico's investments in infrastructure as to a government policy. This commercial effort, spearheaded by one of the bank's vice presidents, Xavier Baquero, assured not only a commercially viable connection to US links through Miami and Homestead, Florida centers, but broad Bitnet and Internet opportunities for academics as well as other commercial enterprises. Another example is Relcom, whose customer base in Russia stood at 391 nodes in 70 Soviet cities at the time of the attempted coup in August 1991 (during which Relcom was the object of international attention for maintaining e-mail links between the then-USSR and the rest of the world), and then expanded to 2,601 nodes less than a year later. An important aspect of Relcom's growth has been its openness to all classes of customers, including private enterprises[8].


There are important examples of hybrids of these patterns for the international spread of networking. The combination of grass root networks and regional leaders is particularly noteworthy. For example, Mike Lawrie of Rhodes University in South Africa took it upon himself to help establish Fido connections to the South African national research and academic Internet backbone for Botswana, Lesotho, Mauritius, Mozambique, Namibia, Swaziland, Zambia, and Zimbabwe. These nets frequently evolve from Fido to UUCP to IP connectivity in conjunction with increased demand and experience.

Impediments to Network Diffusion

Most of the barriers to the wider distribution of networking that we have observed tend to fall into three categories: (1) government policies, laws, and practices; (2) technical impediments; and (3) local and "cultural" factors.

Government policies, laws, and practices

Government policies can clearly cut in both directions. On the positive side, all of the national backbones discussed above would have been impossible without government support, starting with the US Department of Defense's role in the initial development of critical networking technologies and building the ARPANet, and extending to NSF's support for the backbone mentioned earlier. The US National Information Infrastructure initiative prospectively offers every school in the country access to the resources of the Internet. On the negative side, problems range from strong government controls on all aspects of electronic communications for security, political or revenue generating reasons; to national bureaucracies and local bureaucrats who can effectively cripple networking through ignorance, bureaucratic politics, or not-so-benign neglect.

The strong forms of externally imposed or internal government controls for political or national security reasons that characterized the Cold War and certain extreme rightist or leftist regimes, are disappearing. The United States has relaxed many of its national security export controls on computing, telecommunications, and other dual-use technologies to many former communist countries, and CoCom has been disbanded. Some of the former communist countries, including Estonia, Hungary, Slovenia and others, are now essentially without internally or externally imposed political or security controls. Others, like Russia and Cuba, are now more limited by cost considerations than by political or security controls, although some persist. However, such controls, while not always explicit, still seem to exist in China, parts of the Middle East, and a few other places. China appears to have an explicit "P.R.C. Regulations on Safeguarding Computer Information Systems" that, among other possible concerns, may make it illegal to send e-mail without filing a customs declaration[5].

Although IT in general, and networking in particular, have often been praised as democratizing technologies, we note that strong internal security regimes and heavy constraints on Western-style civil rights have not necessarily precluded use. Singapore is becoming a prosperous "intelligent island" based on IT; Iran is greatly expanding its domestic telephone system while controlling the private use of satellite dishes; and under the repressive Pinochet regime networking in Chile expanded dramatically, and was an early example of the regional leader model.

A number of countries have legal constraints, other than those in the traditional national security domain, affecting the use of networks. These include restrictions on carrying third party traffic, intellectual property rights, transborder data flow, privacy, and censorship. Singapore, for example, seems to be having some trouble reconciling its anti-pornography laws with making the Internet widely available.

There is a great disparity in pricing policies among many of the world's government controlled or regulated PTTs. The reasons for this range from the inefficiencies commonly found with large monopolies, to the desire of governments to control and increase hard-currency revenues. Factors of 10 differences in prices have been cited between providers in the US and Europe[3]. PTTs in Africa, Pakistan, the Philippines and elsewhere are well-known for their attempts to acquire hard currency from users. For example, in China the cost of receiving a thousand word message can be more than the receiver's monthly salary, and network users from outside China are given advisories to limit inbound traffic.

Technical impediments

One of the most fundamental impediments is poor physical telecommunications more generally. However, the increasing recognition of the importance of telecommunications to national and business infrastructure has resulted in its proliferation to newly opened societies and markets, most notably in Eastern Europe and the former Soviet Union, and to rapidly expanding markets in LDCs, e.g., Egypt and Iran. In some places, this trend is further accelerated by the privatization of telecommunications services. Many international telecom companies, especially giants like Siemens, Alcatel, NEC and others, have targeted these markets. Furthermore, various international and national organizations such as the World Bank, USAID, and certain national banks are providing billions in credits and other forms of financial support. The overall result is an enormous expansion and upgrading of the telecommunications capacity in LDCs.

Another inhibitor to the diffusion process among LDCs is the lack of technical expertise and training programs for both system administrators and end users[8]. Some organizations, like the OAS, are trying to address this on a regional basis through activities intended to sharpen system operators' skills and increase the number of uses. However, the nets themselves provide an important vehicle for assistance, especially to users with basic working knowledge and e-mail access. Help is often only an e-mail message away.

One more important technical factor is user (un)friendly interfaces. Most of the human-computer interfaces available for networking favor the "technically initiated" who are comfortable in the English language. That leaves the vast majority of people in the world unfavored. The availability of simple-to-use interfaces in local languages is a necessary condition before extensive use of networks is possible in most of the world. Furthermore, the unwieldy nature of the interfaces needed in most countries to connect to the Internet often requires substantial training, a luxury not common in LDCs. So the people with the best access, including a terminal in their office or building, get the chance to practice and overcome the quirks and nuances of daily network use. There are emerging user-friendly interfaces like Mosaic, and services like America On Line that are making major strides in easing the user into the inherent quirkiness of network use. But these are exclusively available in the "have" nations, the ones that already contain the most Internet users. A particularly important need is for improved interfaces for places where connectivity bandwidth is low -- it will be a long time before network users in most African countries will be doing much with tools like Mosaic.

Local and cultural factors

Networks are often perceived by many in LDCs (the "south") as intended to "talk to people in the north." There may not be much perception of the need for local-to-local communications. Furthermore, there may be little in the way of a culture for sharing information locally, or of doing business in an impersonal way.

An extensive example of limited diffusion is associated with the nearly fifty predominantly Muslim nations, and especially those in the Middle East[4]. As a group, these countries are less active and well connected in proportion to their national wealth and populations than other parts of the world. Tunisia, one of the more active, has pursued a strategy like the demand pull approach of Costa Rica, developing incremental infrastructure, particularly among the scientific research institutes, to promote sustained growth. But even here, where there are relatively good results, there are questions of network ownership and management. As the director of a major science institute commented: "information is seen as a source of power in Tunisia; by sharing it you concede that power."[2] As Iran and even Lebanon take tentative steps to enter the network mainstream by establishing Internet nodes, it will be interesting to see if a wider diffusion of message traffic will occur. But in early 1994, of the nearly 15,000 nets on the global Internet, the total for all the Muslim nations was 42, and 29 of those were in Turkey and Indonesia.[6]


In many of the cases of network diffusion considered here, there are several underlying issues that are troubling. The most important is an inherent bias toward those countries, organizations, individuals and even academic specialties that are traditionally the first to receive any valuable technology, and a {it}de facto{it} bias against those who are usually the last. Although in some senses there is an almost umlimited amount of the Internet that can be created and used, and it is diffusing both rapidly and extensively, more capability is going faster to the "haves" than to the "have nots" of the world.

Much of this article has been concerned with disparities between nations. These disparities run the spectrum of of quantity (e.g., number of users per capita), quality (e.g., the existence of no more than marginal e-mail services in many LDCs), and cost (virtually free to most American and Canadian college students in contrast to the unacceptable equivalent of the entire salary for a professional person in parts of Africa).

Although we have not dealt with them explicitly in this short essay, there are substantial inequities within countries as well. Some of the most pronounced are the urban-rural differences in most LDCs. We know of very little network presence outside of a few main cities in most LDCs. Another deficiency may well be the lack of network communications between peoples who are members of "nations within nations," e.g., Indian tribes in the U.S., Canada, and Mexico. This is potentially a very charged subject within many countries, e.g., Mexico or Russia. Even within urban areas of advanced industrial countries, there may be great differences between what is available to inner city and suburban schools.

Furthermore, in all of the countries we have been able to examine at close range -- including dozens ranging from the most industrially advanced to the poorest -- the astronomers, mathematicians, physicists, computer scientists, and engineers are the dominant users from the beginning. It could be argued that the Internet actually provides more services that could interest humanists, arts and letters faculty, etc., yet these specialties are usually poorly represented among users. The vast resources of USENET probably have more to offer a social sciences professor in Lagos or Moscow, than to a mathematician in these locations. In many countries favored groups also do not include many women, although research indicates that women, once on the networks, are likely to more frequent users and to be be at least as productive as men.[11]

We have tried to add a layer of analysis as a counterpoint to the healthy network statistics that are frequently reported. Our perspective is that beneath the statistics is a widening gap between the networking haves and have nots, a gap that many providers of Internet services are not empowered to change. Unless there are concerted efforts to reverse this situation, we believe the gap will continue to grow.


[1] Aupperle, E. NSF backbone trends. _Internet Society News_ 2, 4 (Winter 
1994), 13.

[2] Ben Hanada, E., Cracknell, M. Feasibility of a sustainable development 
network - Tunisia. UN Development Programme, New York, 1992, 30.

[3] Carpenter, B.E., Williams, D.O.  Data networking for the European
academic and research community: Is it important?  _ Electronic Networking-
Research, Applications and Policies_, 2 (1992), 56-65. 

[4] Goodman, S.E., Green, J.D. Computing in the Middle East.   _Comm. ACM_
35, 8 (August 1992), 21-25. 

[5] P.R.C. Regulations on safeguarding computer information systems.
Beijing XINHUA Domestic Service in Chinese, Feb. 23, 1994. Circulated, in 
English, with anonymous commentary, via several routes on the Internet. 

[6] Internet global statistics: 1993.  _Internet Society News_ 2, 4
(Winter 1994), 7. 

[7] Lottor, Mark. Report-9401. Directory pub/zone,
January, 1994.

[8] Press, L. The net: progress and opportunity.  _Comm. ACM_ 35, 12
(December 1992), 21-25. 

[9] Rickard, J., Funk, G. The international Fidonet -- 22,067 bulletin
boards with an attitude. _Boardwatch_ (August 1993), 66-68. 

[10] Ruth, S. The network revolution - A cui bono perspective.  _Higher 
Education Policy_ 6, 4 (1994), 39.

[11] Ruth, S., Goet, R. Must invisible colleges be invisible? An approach to 
examining large communities of network users. _Internet Research_ 3, 1 
(Spring 1993), 47.

Notes and Pointers

Anthony Rutkowski is Executive Director of the Internet Society and Editor-in-Chief of the _Internet Society News_, a major source of statistics on global networking. Stephen Ruth is Professor of Decision Sciences and MIS, and Director of the International Center for Applied Studies in MIS at George Mason University. Lawrence Press is Professor of Information Systems at California State University, Domingues Hills, and contributing editor for personal computing to the _Communications_. Seymour Goodman is Professor of MIS and Policy at the University of Arizona, Carnegie Science Fellow at Stanford University, and contributing editor for international perspectives to the _Communications_.

Statistics on the Internet are also available through anonymous ftp. In addition to Mark Lottor's ftp site[7], Larry Landweber maintains an extensive and verified "International Connectivity Table" that is ftpable from in the connectivity_table directory. Landweber's table is also regularly updated and published in the _Internet Society News_.

Readers are encouraged to send comments, suggestions, anecdotes, insightful speculation, raw data, and articles on subjects relating to international aspects of IT. All correspondence should be addressed to:

Sy Goodman
Center for International Security and Arms Control
Galvez House
Stanford University
Stanford, CA 94305  

Disclaimer: The views and opinions expressed on unofficial pages of California State University, Dominguez Hills faculty, staff or students are strictly those of the page authors. The content of these pages has not been reviewed or approved by California State University, Dominguez Hills.