First Monday

Refuting objections to 
a Global Rural Network (GRNet) for developing nations

Refuting objections to a Global Rural Network (GRNet) for developing nations by Larry Press

In a previous article, we suggested that it is now time to undertake a Grand Challenge project: providing Internet connectivity for every village in every developing nation. Doing so would require perhaps a decade and billions of dollars for design and planning, procurement, installation and operation. Critics object that such a project would not be worth the effort and investment. This article considers nine objections to such an undertaking.

  1. Internet connectivity would be nice, but it is not a high priority.
  2. Internet service has been offered in rural areas of developing nations, and there is little demand for it.
  3. There are no applications of interest or value to illiterate rural people who do not speak English.
  4. There is no sustainable business model.
  5. Developing nations lack the people and resources to do research.
  6. Even if the world community can justify sponsoring the research leading to a concrete backbone plan, developing nations cannot afford to implement it.
  7. Villagers cannot afford to use the network even if the backbone transport and connection are free.
  8. Developing nations cannot afford high–speed connectivity — low–cost store–and–forward technology is more appropriate technology for a poor, developing nation.
  9. We should focus on cities where there is already demand, not rural areas.

We discuss each of these, and conclude with a brief outline of next steps.



In a previous article [1], we suggested that it is now time to undertake a Grand Challenge project: providing Internet connectivity for every village in every developing nation. The argument for this undertaking ran something like this:

Approximately twenty–eight percent of the world population lives in rural areas of low–income nations, and the figure approaches 50 percent if we include lower–middle income nations (see Table 1).


Table 1: World population, 2002. [2]

Income group Rural
Low 1,731 2,495
Lower middle 1,218 2,408
Upper middle 81 329
High 215 966
World 3,246 6,199


For well over a decade, we have hypothesized that the Internet can improve the quality of rural life in developing nations and reduce the pressure to migrate to crowded cities. This hypothesis has generated a lot of activity — training workshops, ICT readiness assessments, local and global conferences and workshops and pilot studies in villages and cities [3]. We have gathered considerable evidence in support of our hypothesis, and made every multinational organization and government aware of the importance of the Internet and ICT.

In spite of this activity and progress, Internet diffusion in developing nations remains miniscule. A person in a low–income nation is 100 times less likely to be an Internet subscriber than a person in a high–income nation, and the vast majority of those who have accounts are in urban areas (Table 2) [4]. That sounds bad, and the actual situation is even worse because a subscriber in a developing nation almost certainly uses an analog modem to connect to an ISP with a very low–speed link to domestic and global backbones. Interactive applications, like Web surfing, are often impossibly slow in a developing nation — the Internet experience is qualitatively different than in a developed nation.


Table 2: Internet subscription rates, 2002. [2]

Income group Population
Subscription rate
Low 2,413 5,424 0.22
Lower middle 2,393 69,762 2.92
Upper middle 331 12,150 3.68
High 961 216,069 22.48
World 6,097 303,405 4.98


There are many reasons for our slow progress in the face of evidence that the Internet could improve quality of life for the half of the people on Earth, but the most obvious is poverty. The income levels in developing nations are insufficient to attract the capital to justify Internet diffusion. The cost of twenty hours of (inferior) Internet access in a low–income nation is 2.5 times monthly per capita gross national income. Twenty hours access in a high–income nation costs 1.6 percent of the average monthly income (Table 3) [5].


Table 3: Cost of twenty hours of Internet access as a percent of gross monthly income per capita.

Income range Percent of GMI/capita
Low 246.4
Lower middle 24.9
Upper middle 8.6
High 1.6
World 88.7


If user demand in a developing nation is insufficient to attract the requisite capital, we should consider a publicly funded infrastructure project, an ambitious grand challenge. One such grand challenge would be to provide backbone connectivity to every village in every developing nation, to build a global rural network, GRNet. GRNet would follow the leveraged approach that was used in developing the U.S. National Science Foundation network (NSFNet) during the 1990s. NSFNet provided a backbone, access links and routers to university and research networks. Similarly, GRNet would provide a backbone network, access link, and point of presence equipment (POP) for every village. The rest would be up to local people.

This would require research, development and procurement and cost billions of dollars. That is daunting, but there are many examples of such audacious projects. It could be done.

I presented the GRNet grand challenge at a recent workshop [6], and, while there was considerable agreement, there were also objections from participants. This is a complex issue, and the objections all have merit and counterarguments. Let us look at several common objections to public investment in Internet infrastructure such as GRNet.

1. Internet connectivity would be nice, but it is not a high priority. Power, clean water, nutrition, employment, and literacy are more important and should be addressed first.

Internet infrastructure can lead to improvement in each of these areas. As Kofi Annan has stated:

"Affordable technologies, in the hands of local communities, can be effective engines of change, both social and material. Access to information and technological know-how is essential if the world is to defeat hunger, protect the environment and achieve the other Millennium Development Goals agreed by Heads of State and Government at the United Nations Millennium Summit in 2000." [7]

This is not to say the Internet is a "silver bullet" that will turn a village into a developed suburb with a chicken in every pot and a car in every garage, but I believe it can make a marginal difference in many areas. Many projects have demonstrated improved economic productivity, health care, education, democracy and human rights, e–government, and news and entertainment via the Internet.

The following examples are from a recent presentation by Raghu Rao of N–Logue [8].

In March 2003, after three hundred chickens died in a southern Indian village, the following e–mail was sent to the department of animal husbandry in Chennai:

"In the village of Pudupatti, all hens are dying one by one. The symptoms for these hens are: first, their weight reduces and finally their necks shrink. At this stage they die immediately. Until now, 300 hens have died in this way. How do we stop this? Please give us a solution immediately."

A doctor from the department then visited the village and vaccinated the remaining chickens for the disease before it could spread further [9].

Mr. Rao also gave several examples of health care applications, including the system illustrated in Figure 1.

Figure 1: A low–cost Web cam and simple video conferencing software are used in medical diagnosis.

Mr. Rao described other applications, including diagnosis of a chicken pox epidemic in one village and the diagnosis and cure of a disease of the local okra crop in another. There are many other examples of the Internet contributing to the solution of the basic problems that those who object to ICT expenditures would address first (see footnote 3).

2. Internet service has been offered in rural areas of developing nations, and there is little demand for it.

For example, the Bangladesh Telephone and Telegraph Board [10] (BTTB) offers Internet connectivity in the 64 district headquarters of the nation, and they hope to connect the 410 local government seats by end of next year. Still, they report weak demand for the service [11].

I do not think their poor sales reflect a lack of demand because BTTB, like their counterparts in many developing nations, offers outdated service at a high price with little support.

Consider customer service. Only BTTB telephone customers are eligible for a dial–up Internet account. To sign up one must submit two copies of a two–page hard–copy form, three passport photos and three choices for user name in case the one you want is already taken. It appears that one copy of the form must be submitted to one office and the second copy, with a fee of 1,000 Taka (about US$16), to another. According to the Web site "the total process requires some days."

Usage charges are approximately US$.5 per hour between 8 AM and 11 PM and $US .3 per hour at night. If, say, one used ten hours during the day and ten more at night during a month, the charge would be approximately US$8. The setup fee of US$16 and perhaps $US8 per month for service must be considered in the context of the annual gross national income per capita of US$380 [12]. DSL service is available in limited areas, but the charge for a BTTB–supplied modem is US$1,000 [13].

Cost, support, and availability surely affect demand, but there is a more fundamental factor: the current BTTB offering is not the service I have in mind. Dial–up Internet over a slow connection is qualitatively different than high–speed Internet over an "always on" connection.

Low connection speed makes interactive applications like Web browsing frustrating or impossible. Even character–oriented sessions may time out. BTTB equipment is overloaded. International link capacity is currently only 4.5 mbps download and 2.5 mbps upload and inter–city links are generally microwave. Bangladesh is typical of developing nations.

For the sake of discussion, I am suggesting a point of presence with at least a two-megabit link to every village. This would indeed be a Grand Challenge, but, if achieved, it would enable a very different experience than what BTTB and their counterparts in other developing nations offer today. It would enable a technologically modern Internet center, many home users, audio and video data, etc.

3. There are no applications of interest or value to illiterate rural people who do not speak English.

Earlier, I mentioned some example applications, and the Web sites I cited lead to many others. These often involve a literate intermediary who sends e–mail or retrieves information on behalf of the ultimate user. These examples were encouraging, but, granted, anecdotal.

However, the network I am suggesting would enable other applications. It would be qualitatively different than the current network. Not only would it be faster than the network used for today’s applications, it would a ubiquitous, end–to–end network.

Greater speed obviously enables different applications. For example, downloading movies and other forms of entertainment would be practical. One could imagine a small digital village theater providing entertainment and news. What if instead of the e–mail messages, photos, and low–speed video chat illustrated in Figure 1, we could have high–resolution video from more specialized cameras, hi–fidelity sound from a stethoscope, EKG data from a Holter cardiac monitor, and other diagnostic input? Could the village theater operator and the village nurse earn a profit? Could they have sustainable businesses?

The nature of applications would also change with ubiquity. A network with one user is clearly worthless. Add a second user, and the first user has someone to communicate with. A third user doubles the possibilities for our first node. Every new user is a "content provider." If everyone in the village and region has access to the network, everyone is a potential conversation partner and information contributor.

The Internet was conceived as an end–to–end network in which the only function of the network is to route packets of information as efficiently as possible from one computer to another [14]. The network is not aware of the content or data type of those packets. The network routers do not know if the packets they forward contain healthcare information or movies. They do not know if the packets contain text or images or the designations of player positions in a chess game. The computers that run the applications — the clients and servers — are at the edge of the network. The application intelligence, and hence the innovation, takes place at the edges of the network. Every user is a potential application inventor and developer.

People often find unanticipated applications for technology. The telephone was invented to broadcast music and sound recording for correspondence. Phillips expected their cassette tapes would be used in dictating machines. The sub–head of the New York Times article covering an experimental television transmission in 1927 read "Commercial Use in Doubt," [15] and The Times covered the first public demonstration of the transistor in four short paragraphs at the end of the "News Of Radio" column on page 46 [16]. The IEEE 802.11 network standard (WiFi) was developed for local area networks in offices, but it has been used to build community networks, point–to–point links and commercial hot spots. (As Iqbal Quadir of Grameen Telephone says, 802.11 will also be used for hut spots).

Rural applications will be determined and funded locally, and they too will surprise us. Farmers in developing nations use roads for drying and threshing grain as well as for transportation. When HiTech City was being built in Hyderabad, I saw modern construction cranes and women carrying sand and cement on their heads working side by side. Necessity truly is the mother of invention, and rural people in developing nations will develop network applications to solve their own problems using their resources and knowledge of those problems. For example, Chile was able to export systems for the logging and banking industries because they had developed and refined them for internal use [17].

Once applications are developed, the network will provide a platform for deployment and testing. It will be a resource for the global development community.

4. There is no sustainable business model.

Many developing nations have experienced the frustration of seeing study reports gather dust and pilot projects disappear when external funding ran out. In such cases, the beneficiaries of the investments were the people working on the reports and projects, not the nation.

There are many paths to sustainability. Let us consider three examples: one subsidized by a donor, one financed by a government, and a market–based business (see Figure 2).

Figure 2: Indian and Cuban Internet Centers. The Indian center is run at a profit and the Cuban center is run by the state.

The government finances the Cuban Youth Computer Clubs (YCC) [18]. Since 1989, 350 YCCs have been built, and they are used for Internet access, training, game–playing, etc. They are geographically dispersed throughout the island, and are successful but insulated from market forces. Fidel Castro personally ordered seed funding for the YCCs, and photos of his visit when the Havana Club opened hang on the walls along with his autograph (Figure 3). The YCCs will be sustained as long as the government values them.

Figure 3: Fidel Castro’s autograph (reading "I envy you") at the Havana YCC.

The United Nations Development Programme’s Sustainable Development Network Programme (SDNP) began in 1992 and ran through 2000 [19]. The SDNP sponsored projects in 44 countries, and established a Small Island Developing States Network [20]. SDNP projects focused on both connectivity and content. They often established the first small ISP in a nation, trained technicians and users, built Web sites and portals, lobbied governments on the importance of the Internet, established public access points, etc. SDNP expected their projects to become economically self–sufficient after a subsidized startup period, and a recent independent evaluation of the program found that:

"In short, SDNP has a record of some considerable success, and some failure, in relation to sustainability — precisely whether the balance tips more in favour of one or the other may become clearer in a few years. In some cases too, its impact is sustained, though institutionally, its task completed, it no longer exists. ... A dozen to fifteen SDNP organizations are well established (though not to say secure) with significant capabilities in several dimensions ... A further group numbering approximately less than ten continues with a more limited range of activities and in a more uncertain environment ... A final group comprising about nine countries has evolved in directions related to, but different than SDNP." [21].

The SDNP centers have had varying degrees of success, and some, as in any sort of business venture, have failed all together.

A franchise–like market model has also been used in several nations. We saw sample applications from N–Logue’s rural Kiosks earlier. N–Logue centers offer computer training, digital photography, desktop publishing, e–mail/voice and video mail, telephony, and access to government, medical, veterinary, and agricultural experts and information. The center operator invests approximately US$1,000 for a PC with a Web camera, printer, power backup, and local language software equipment, and breaks even with revenue of about US$75 per month [22].

Of course we may see mixed sources of capital and operating funds. For example, based on their experience with a pioneering Village Connectivity project in southern India [23], the MSSRF has suggested a government–industry–donor coalition in pursuit of the goal of establishing a knowledge centre in every Indian village by the year 2007 [24].

While the record is anecdotal and mixed, we see that examples of sustainable business models for Internet access do exist. But, this is only part of the sustainability evidence. The business case will improve as technology improves. Some centers are sustaining themselves in spite of their outdated technology. What might they do with modern technology?

Based on the track record and the need for vast scale, it would seem that market–based projects like the N–Logue kiosks are a better bet than the state run or subsidized centers. As we have said, applications should be invented, installed and financed locally — in the villages at the edge of the network.

However, this is not an argument against centrally funded infrastructure projects. There are many examples of excellent tangible and intangible returns on government infrastructure investments. For example, in the U.S. we have had regulated monopoly for telephones, rural electrification, many municipal utilities, the Global Positioning System, the Interstate Highway System, etc. Perhaps the first government funded telecommunication infrastructure was Samuel Morse’s telegraph line from Baltimore to Washington (Figure 4).

Figure 4: A US$30,000 congressional grant seeded the American telegraph industry. [25]

The NSFNet is another networking example. A research and procurement investment of a less than US$200 million paved the way for the Internet [26]. The NSF had no sustainable business model for their Internet backbone. In fact, its phase–out was planned from the start. While it lost money, it improved the cost–benefit equation for every network that connected to the backbone. A backbone connecting every rural village would play a similar role. It would be a common investment that made the network possible and improved the business model of every application in every village.

Governments encourage infrastructure creation in many ways — by funding research, establishing regulated monopolies, procuring infrastructure, operating the infrastructure, or a combination of these. The projects also change over time. In the case of the Internet, the government financed the research primarily through grants to universities, procured the NSFNet, and contracted for its operation. Once it was well–established, they phased out of operations over four years. NSF procurement was restricted to backbone infrastructure and a connecting link for each university. The university networks were responsible for their own funding, staffing, application development, etc., and, in the aggregate, they invested much more than the NSF.

Note that even in a government–organized project, much of the work is done by private industry. Companies bid on construction of roads, satellite systems, dams, power distribution grids, etc. In the case of the NSFNet, AT&T, Sprint, Merit (a university consortium), and others won contracts for the actual work. The same would be true of developing nation backbones. Existing fiber owners might contract for the fiber backbone; the network operations center might be contracted to a university; the military or a private contractor might be responsible for the physical security of towers and cables (with pay as a function of uptime); etc.

5. Developing nations lack the people and resources to do research.

One way to answer this objection is to point to the leading edge research being carried out today in developing nations. The technology employed by N–Logue came from the Telecommunications and Computer Networks Group [27] at the Indian Institute of Technology in Chennai, India and Technology Review recently called Microsoft’s Beijing research lab [28] the "world’s hottest computer lab" [29].

But a rural backbone network would be a global project designed by a global research team. The history of networking in the U.S. might be instructive. The ARPANet and the experiments preceding it were underwritten by a government agency, the Information Processing Techniques Office (IPTO) of the Advanced Research Projects Agency (ARPA) of the Department of Defense. Anyone with experience in developing nations is justifiably skeptical of government projects; however, IPTO decisions were not made by government bureaucrats, but by visionary faculty members from universities who were brought in for temporary assignments. Similarly, outstanding people, often on temporary assignment, designed and oversaw the contracting for and operation of the NSFNet.

The ARPA and NSF networks were research projects, not bureaucratic government programs. At the time, people were still debating the relative merits of packet and circuit switching and TCP/IP versus OSI. No one had ever built a nationwide internetwork. The design and deployment of GRNet would also be a research project.

In the U.S., we often bring university and other research people into agencies to guide research and complex development programs. In Singapore government technical positions are prestigious and pay well, so good people take permanent positions. Regardless, a relatively small group of highly qualified people should be responsible for the vision, experimentation and awarding of contracts for GRNet.

What sorts of research skills would we need? They would include:

They should be the best people in their areas, brought together for this project in much the same way as ARPA–IPTO did when they funded the networking and interactive computing research to which we are so indebted today [31]. Some of these people would be from developing nations, others from developed nations. The ARPA–IPTO projects focused for the most part on top researchers at a few universities and laboratories in the U.S. Today we would draw from the world at large, and enjoy the luxury of communicating via the Internet.

6. Even if the world community can justify sponsoring the research leading to a concrete backbone plan, developing nations cannot afford to implement it.

That is true. We are talking about a "Grand Challenge" to the world community, not to the developing nations. How much are we talking about?

A multidisciplinary feasibility study involving the people with the skills suggested above would be needed before we could make even an estimate of the cost, but we are talking about a large problem. China has approximately 930,000 villages with a population of 900 million [32]. India and Bangladesh together have a rural population of 854 million, and 638,365 and 86,000 villages respectively [33]. These three nations average approximately 1,060 people per village. There are 2,949 million rural people in low and lower–middle income nations. If they lived in villages of 1,060 people, GRNet would have to reach nearly three million villages. This would indeed be a grand challenge.

Using the "back of an envelope," we can envision the following costs:


Item Comment
POP equipment and installation Router, server, peripherals, and software
POP power Various power alternatives
POP building Best left to local people
POP radio, tower and antenna Should radios be homogenous?
Backbone radio, tower, antenna Should radios be homogenous?
Backbone maintenance and security Private firm?
Optical backbone upgrade After assessment
Network operation center Key to maintenance


The above can be divided into backbone and POP expenses. If we wish to connect two million villages, the cost of village facilities will swamp the cost of the backbone and network operation center.

The POP equipment in the village includes a router, server and peripherals. This can be standardized, and should be designed with performance, upgrade flexibility, maintenance, and power requirements in mind.

Manisa Pipattanasomporn and Saifur Rahman estimate the power requirement for an Internet kiosk as between 213 and 237 watts [34]. They assessed the cost of supplying that using solar power, a fuel cell or an internal combustion generator, and concluded that solar would be the most expensive. Solar may be taken as a worst case, but since the solution should be tailored to the village/national circumstance, various alternatives should be designed. As with the POP equipment, we should strive for a few uniform designs.

The POP building should be left outside the scope of the common infrastructure. Availability of facilities will vary from village to village, and construction skills are locally available. Building specifications can be defined.

The radios, towers and antennae really cannot be considered until the backbone architecture is designed. One possibility is to view the fiber backbone as a bus with wireless feeds, implying three classes of radio/antenna: those interfacing directly with the fiber bus, those in the villages, and intermediate radios (Figure 5).

Figure 5: Village POP radios connect to the fiber backbone access radios through an intermediate mesh.

The intermediate radios (and perhaps the others) should form a mesh for performance and reliability. The backbone interface radios should be able to communicate with their neighbors as well as intermediate radios and nearby village radios. Village radios must communicate with intermediate and backbone interface radios and perhaps each other. We must determine the extent to which the three classes of radio can be identical and where should they be differentiated.

Heat, cold, humidity, rain, insects, etc. create a hostile environment for telecommunication equipment [35]. An isolated backbone radio tower is also a tempting target for thieves or sabotage [36]. Provision must be made for maintenance and security. A private contractor or a government agency might do this, and financial incentives could be provided for uptime. There must also be plans for the logistics of spare equipment, training, and a state of the art network operations center to monitor connectivity and performance.

The cost for all of this would be great compared to the scope of the pilot projects we have seen to date, but the G8, ITU, UN, World Bank, and others have talked extensively about the human development returns to Internet connectivity investments. Seen from that perspective, the cost would be appropriate for a Grand Challenge.

7. Villagers cannot afford to use the network even if the backbone transport and connection are free.

Villagers have formed viable businesses around cell phones, cows, and sewing machines. These are often funded using a very small, micro–credit loan. Grameen (village) Bank invented and perfected micro–loans [37]. Muhammad Yunus, a Bangladeshi economics professor, founded Grameen in 1976, and it has thrived by making very small business loans. Ninety six percent of their clients are women, and nearly all are landless. During the 12 months from April 2003 to March 2004 Grameen Bank disbursed US$377.19 million, and they made a profit of US$1 million and paid US$5.9 million in pensions to retirees during 2002. Grameen has evolved procedures that yield a loan recovery rate of 98.69 percent, and they employ 11,988 people [38].

Micro–credit is now available in developing nations worldwide [39]. It is well suited to financing of enterprises at the edge of an end–to–end network. N–Logue estimates the break–even point for an Internet kiosk at US$75 per month. If an average village has 1,060 people, that works out to about seven cents per person and less than one U.S. dollar per family. This level of expenditure is within the reach of a developing nation villager, particularly if savings in transportation cost, say substituting a voice or e–mail for a bus trip, or an increase in income, perhaps finding a better price for a product, offsets it.

Note again, that these estimates are based upon today’s applications using low–speed connections. Better technology will enable new and more valuable applications like digital news and entertainment, and, the existence of a GRNet will lower the breakeven point. Applications that are successful in one area (domestic or foreign) will quickly spread. This is the level at which markets will operate.

Still, one should not underestimate the task faced in the village. A village should be required to demonstrate some level of readiness before receiving equipment and being connected to the network. The costs of operator training, and equipment installation and handover of the network connection must be covered. In developed nations we speak of "truck rolls" when a technician must visit a home to install a DSL or cable modem. The cost of "oxcart rolls" will be substantial and should be anticipated.

With so many villages, one should invest considerable effort in the design of POP equipment and software, training, and remote monitoring from the network information center in order to simplify installation and maintenance. The network itself should be used for update distribution and maintenance related communication. Every dollar saved by clever, comprehensive design will save millions.

8. Developing nations cannot afford high–speed connectivity. Low–cost store–and–forward technology is more appropriate technology for a poor, developing nation.

Asynchronous news and e–mail preceded the Internet. As the Internet spread, typical users gained access by logging into a shell account on a machine that was directly connected. Those shell accounts gave way to IP connectivity over analog telephone lines. Today many users connect via television cable or DSL.

Should poor, developing nations stick with low–cost asynchronous access? Satellife may have been the first organization to deliver asynchronous connectivity in developing nations [40]. They used a low–earth orbit satellite that allowed those on the ground an access window every few hours. More recently, DAKNET places radios on busses or other vehicles to periodically connect and do batch transfers with villages they pass [41].

While very cheap, asynchronous connectivity limits the range of possible applications. It is surely better than no connectivity, but it should be seen only as an interim step. To do otherwise would guarantee developing nations remaining permanently at the back of the "digital bus."

Some applications are impossible without a fast, symmetrical connection. Even if an application is feasible using outdated equipment, delays are both confusing and frustrating. Fast, persistently connected applications are more powerful and easier to use than those using older architectures. Developing nations should use leading edge technology for appropriate applications.

9. We should focus on cities where there is already demand, not rural areas.

One of the hopes for networking in developing nations is that by bringing education, health care, news, entertainment, contact with the outside, employment, etc. to rural areas, the quality of life may rise to a point where migration to crowded cities will be diminished. That, of course, demands focus on rural areas.

There is also the question of ethics, of a just society providing the greatest good for the greatest number of people. Gandhi spoke of the concept of antyodaya in which we focus attention first on the poorest, most needy people. They live in the rural areas of developing nations.

We should note in passing that the leveraged infrastructure approach advocated in this article could also be applied to a municipal network. Figure 5 depicts a national network, but the same general architecture may also work for a metropolitan area network.

Mesh networking standards are under development for both IEEE 802.11 [42] and 802.16 [43] wireless networks. Within a few years, we may see commodity–priced mesh nodes. One could then imagine low–cost "appliances" incorporating two–smart radios (one for user access and one for mesh connectivity) linking to a municipal backbone [44]. These could be packaged with antennae and other accessories to create mesh "kits." To scale effectively, the kits would have to configure and adjust themselves automatically. Interference issues might require different kits for areas of varying density: villages, urban houses, apartment buildings, and favelas.

Next steps

If out goal is to have GRNet operational within, say, ten years, how might we proceed? One approach would be to begin with a research team designing the network for a pilot nation. Once design and architecture was complete, the team would award contracts for implementation. After that experience, we would be ready for a global rollout.

The research team should be relatively small and include very good people with skills, including those listed earlier. The team would be charged with design and feasibility study. They will have to invent and apply network design and GIS tools, answering many questions like: What sorts of low-cost radio will be available at the time of rollout? How should the radios and their antennae be tuned for various environments? What frequencies should be used in various environments? Should developing nations go beyond harmonization with world standards for license–free spectrum allocation? If the developed nations cap power in license–free bands, should developing nations comply with those power constraints or should they adopt a more liberal "Wolfman Jack"W approach since they have little competing transmission in rural areas [45]. The team will have to be oriented toward the future as both technology and spectrum regulations are changing rapidly. Cooperation with industry will be needed to find these answers, but the research team should remain in control of the process since manufacturers and telecommunication providers have conflicts of interest.

Once design alternatives and equipment modules are defined, we would be ready for deployment in a pilot nation. The initial research consortium would retain responsibility for contracting for and evaluation of this trial. Criteria for selecting the trial nation would include:

With the pilot nation lessons under their collective belts, the team would issue requests for proposals for the global rollout. The order of rollout could be determined using criteria similar to those listed above, and some minimum standards would be needed. Again, it is important that the research team, not the many vendors actually building and operating the network, retains control over and responsibility for the project.

It may sound as though GRNet would be a charity project for the rural poor in developing nations, but immense benefits would accrue in both the developed and developing nations. The story of the young mathematician Srinivasa Ramanujan rising to fame after writing Professor G.H. Hardy at Cambridge from his village in Southern India is well known [46]. How many Ramanujans will we find on GRNet? How many tropical drugs will we "discover?" How will we be empowered by access to open–standard, mass–produced medical instruments and data analysis software developed for use in two million villages? Different problems and cultures give rise to different worldviews and lead to the asking of different questions and development of different application. We will all be enriched.

Which nation shall we begin with? End 
of article


About the author

Larry Press is Professor of Information Systems at California State University, Dominguez Hills.



1. Larry Press, 2004. "The Internet in developing nations: Grand challenges," First Monday, volume 9, number 4 (April), at http://firstmonday. org/issues/issue9_4/press/.

2. World Development Indicators database, World Bank, queried June 2004.

3. For descriptions of a few of these activities see:

4. International Telecommunication Union, 2003. World Telecommunication Development Report. Economies are divided according to 2002 GNI per capita, calculated using the World Bank Atlas method. The groups are: low income, US$735 or less; lower middle income, US$736–US$2,935; upper middle income, US$2,936–US$9,075; and high income, US$9,076 or more.

5. World Development Indicators database, queried May 2004.

6. International Workshop on Nationwide Internet Access and Online Applications, Dhaka, Bangladesh (May 2004), at

7. Kofi Annan, 2004. U.N. press release (17 May), at http://ww

8. Raghu Rao, 2004. "Internet Services in the Villages of India," International Workshop on Nationwide Internet Access and Online Applications, Dhaka, Bangladesh (May), at

9. Sherry Morse, 2004. "Virtual Vets Help Animals in India," Animal News Center (14 January), at


11. With 410 points of presence, BTTB would still not reach the rural population. Professor Jamilur R. Choudhury estimates that perhaps 35–40 percent of the people live in these urban and semi–urban areas; e–mail conversation, June 2004.

12. World Bank database, 2002.

13. Saifur Rahman and Manisa Pipattanasomporn, 2004. "Nationwide Internet Access & Online Applications," International Workshop on Nationwide Internet Access and Online Applications, Dhaka, Bangladesh (May), at

14. J.H. Saltzer, D.P. Reed, D.D. and Clark, 1981. "End–to–end Arguments in System Design,&#quot; Second International Conference on Distributed Computing Systems (April), pp. 509–512; published with minor changes in ACM Transactions in Computer Systems, volume 2, number 4 (November 1984), pp. 277–288.

15. "Far–Off Speakers Seen as Well as Heard Here in a Test of Television," New York Times (8 April 1927), and at 8apr27.htm.

16. "The News of Radio," New York Times (26 June 1948), p. 46, and at ess/NYTTransistorcolumn.gif.

17. Larry Press, 1993. "Software Export from Developing Nations," IEEE Computer (December), and at .

18. b/.



21. UNDP, "Sustainable Development Networking Programme, Report of an independent external assessment," February 2004, at Final.pdf.

22. These figures were supplied by N–Logue, and the picture might not be as rosy if an independent assessment were made; however, the value proposition will improve with technological advance and the invention of new applications. A GRNet backbone would further enhance the case for sustainability.

23. Larry Press, 1999. "A Client–Centered Networking Project in Rural India," OnTheInternet, volume 5, number 2 (January–February), pp. 36–38, and at

24. "Rural Knowledge Centres: Harnessing Local Knowledge via Interactive Media," Policy Makers Workshop (8–9 October 2003), M.S. Swaminathan Research Foundation, Chennai, India,

25. http://memory bin/ampage?collId=llsl&fileName=005/llsl005.db&recNum=655.

26. Larry Press, 1996. "Seeding Networks: The Federal Role," Communications of the ACM, volume 39, number 10 (October), pp. 11–18, reprinted in OnTheInternet, volume 3, number 1 (January–February 1997), pp. 13–22, and at http://som.csud


28. http://w

29. Gregory T. Huang, 2004. "The World’s Hottest Computer Lab," Technology Review (June), pp. 32–42.

30. Glib marketing promises that IEEE 802.16 (WiMAX) standard radios will cover 50 kilometers at 70 mbps without line–of–site are misleading. In reality, there are several 802.16 standards, multiple frequency bands, and many tunable parameters, algorithm options, and amplification and antenna variations. Initial devices will be designed with the U.S. carrier market in mind, and IEEE 802.11 (WiFi) is also evolving rapidly. Considerable experimentation is needed to configure radios and antennae for specific climate and environments. Different configurations may also be used for backbone–interface, intermediate mesh, and village radios.

31. Larry Press, 1993. "Before the Altair — The History of Personal Computing," Communications of the ACM, volume 36, number 9 (September), pp. 27–33, and at http://som.csud

32. Anne F. Thurston, 2002. "Testimony at Roundtable Discussion on Village Democracy in China," Congressional Executive Commission on China (8 July), at http://

33. Numbers of villages:
Bangladesh: Miah M. Adel, 2000. "Arsenification: Searching for an Alternative Theory," The Daily Star, volume 3, number 236 (28 April), at
India: Census of India, http://www.ce
The population figures for India and Bangladesh are from the World Bank’s World Development Indicators database, queried June 2004.

34. Manisa Pipattanasomporn and Saifur Rahman, 2004. "Distributed Generation Solutions for Internet Access in Remote Areas," International Workshop on Nationwide Internet Access and Online Applications, Dhaka, Bangladesh (May), at

35. Marut Lueprasert, 2004. "Experience with TDMA/WLL Implementation in Thailand," International Workshop on Nationwide Internet Access and Online Applications, Dhaka, Bangladesh (May), at

36. Maoist rebels have bombed a number of telecommunication towers in Nepal, for example, see "Rs. 10 million Loss in Telecom Bombing," Kantipur Online (12 June 2004), at http://www.k or, "Nepal rebels kill 29 in telecom tower strike," The Telegraph (3 March 2004), at

37. Larry Press, 1999. "Connecting Villages," OnTheInternet, volume 5, number 4 (July/August), pp. 32–37.


39. See the Virtual Library on Microcredit and Microfinance, at html, and the World Bank, at http://topics.d


41. Alex Pentland, Richard Fletcher, and Amir Hasson, 2004. "DakNet: Rethinking Connectivity in Developing Nations," IEEE Computer (January), at http:/ /

42. While there are currently many 802.11 mesh products, they are not yet standardized. We can look forward to mesh extensions (802.11s) and extensions for dynamic frequency selection and transmit power control (802.11h). Standardization and technical progress will lower prices significantly, and these and/or 802.16 mesh products will be key components of the network described earlier.

43. 802-11-ess-mesh.doc.

44. My university is installing such a network in a dorm complex. We will use 802.11A radios for backhaul to the university backbone and 802.11G for user access within the buildings. This is costly today because we must use separate radio modules and proprietary media access and routing protocols; see http://som.csudh.e du/fac/lpress/471/hout/dorm/.

45. "Wolfman Jack" was a U.S. disk jockey who reached much of the southwestern united states by broadcasting a 250,000 Watt signal from Mexico; see http://www.ra

46. See, for example, http://s

Editorial history

Paper received 30 June 2004; revised 26 July 2004; accepted 29 July 2004.

Contents Index

Copyright ©2004, First Monday

Copyright ©2004, Larry Press

Refuting objections to a Global Rural Network (GRNet) for developing nations by Larry Press
First Monday, volume 9, number 8 (August 2004),