Cover Image
close this bookBridge Builders: African Experiences with Information & Communication (BOSTID, 1996, 304 p.)
View the document(introduction...)
View the documentPanel on planning for scientific and technological information (STI) SYSTEMS IN SUB-Saharan Africa
View the documentPreface
View the documentIntroduction
close this folderCase studies on the introduction of cd-rom to university libraries
View the document(introduction...)
View the documentThe CD-ROM Service for the University of Dar es Salaam
View the documentCD-ROM for Health Information in Zimbabwe
View the documentBackground Summary: The African Index Medicus (AIM) Project
View the documentCommunications for Better Health Project in Zambia
close this folderCase studies on desktop publishing
View the document(introduction...)
View the documentDesktop Publishing at the University of Zimbabwe
View the documentDesktop Publishing at the Tanzania Commission for Science and Technology
View the documentEnvironmental Publishing Network - ENVIRONET at ICIPE Science Press
View the documentInnovations in Desktop Publishing at the African Academy of Sciences, 1989-1992
close this folderCase studies on electronic networking
View the document(introduction...)
View the documentElectronic Networking for the Research Community in Ethiopia
View the documentNetworking in West Africa
View the documentBackground Summary: SatelLife and HealthNet
View the documentMUKLA: Evolution of a Homegrown Network in Uganda
View the documentThe UNIMA Fidonet Network: Computer Networking for Communications in Malawi
View the documentBackground Summary: A Global Computer Network for Change
View the documentBringing the Internet to Zambia
close this folderCase studies on the collection, management, and dissemination of local information resources
View the document(introduction...)
View the documentNAPRECA and Its Role in the Dissemination of Information on Natural Products Research in Africa
View the documentA Socioeconomic Development Information System for Botswana
View the documentKEMRI's Management Information System
View the documentThe Ghana National Scientific and Technological Information Network (GHASTINET) Project
View the documentConclusion
close this folderAppendixes
View the documentAppendix A List of authors
View the documentAppendix B List of Acronyms and Abbreviations
View the documentAppendix C Glossary

Electronic Networking for the Research Community in Ethiopia

by Lishan Adam

Lishan Adam has a B. Sc. in Electrical Engineering and a M.Sc. in Computer Information Systems. He is working towards his Ph.D at the University of Sheffield, United Kingdom. He has worked in the information field for the last six years, focusing mainly on technology for accessing information, which includes electronic communication networks. Currently, he is the coordinator for the Capacity Building for Electronic Communications in Africa (CABECA) project. His research interests include electronic communications for grassroots institutions, networking technologies, and techniques for building qualitative information systems and networks.

BACKGROUND AND CONTEXT OF THE PROJECT

This case study describes the challenges faced in setting up a research communication network in Ethiopia. It covers the activities I undertook and the constraints I faced while expanding electronic connectivity to a research community. It also discusses the lessons I learned in the process.

Ethiopia is located in the horn of Africa and shares borders with Sudan, Kenya, Somalia, Eritrea, and Djibouti. Addis Ababa, the capital city, is often called "a city for Africa" because it hosts two major organizations: the United Nations Economic Commission for Africa (UNECA) and the Organization of African Unity (OAU). UNECA is the African regional arm of the United Nations Economic and Social Council. UNECA has established and sponsors a number of African institutions to promote all aspects of socioeconomic development of the region.

The Pan African Development Information System (PADIS), one such regional institution under UNECA, was created to build both information and networking capability in African countries. PADIS was instrumental in setting up a national electronic network in Ethiopia and in connecting members of the research community to each other locally and with colleagues worldwide.

There are only a few advanced research institutions and two major universities in Ethiopia. Addis Ababa University, the country's largest university, administers most of the colleges. The Agriculture University of Alemaya, in Harar Province 500 kilometers east of Addis Ababa, is the next largest research center. Health research is undertaken by the medical faculty of Addis Ababa University and at two other research colleges in Gondor and Jimma. The country has several teacher training schools and research centers.

The level of scientific and technological information (STI) and infrastructure in these research centers varies. Some colleges lack the most essential resources, such as paper and ink. Others use computers and connect to the local electronic networking host. The need for STI and communication is great, especially by colleges outside of Addis Ababa.

The Current STI Environment in Ethiopia

Ethiopia is the second poorest country in Africa, having been devastated by long-lasting war, drought, and mismanagement. The national STI environment is generally weak, reflecting the poor economic situation in the country. The transition' economy is too weak to support the population, which is growing at an annual rate of 3.1 percent, with the fertility rate mounting to 7.5 percent. The STI environment has suffered from the poor economic performance during the last twenty years.

All STI centers, including university libraries, have been unable to expand and keep up with demands. There were hardly any books coming to the country during the previous socialist government. According to the head of the Addis Ababa University libraries "most collections are those of the 1960s." Those were the "golden years" of library collection development - when resources were made available to support new acquisitions.

Most colleges are not equipped with computers and other information technology tools. The more remote the college from the capital city, the less it is exposed to computing technology. (See Box 1.) Under-use of existing equipment due to lack of training and low quality of maintenance is also common. Lack of knowledge about different hardware contributes to the under-use of computer accessories.

Computer equipment is generally expensive. A non-error correcting, 2400 bps modem that costs $40.00 in the United States costs the equivalent of $200 in Ethiopia. The local price of accessories is between 400 percent to 1500 percent more than the original cost. A bureaucracy for clearing equipment through customs exasperates most possibilities of getting it through mail order or from friends residing in foreign countries.

One additional problem, unique to the country, is related to the use of Ethiopian script. Ethiopia is one of the oldest countries using its own script. The shortage of good, easy-to-use software that is adaptable to local script is a major problem. Generic native-language software interfaces that allow easier storage and retrieval of textual information in local languages and scripts are not well developed. Modern tools, such as Windows [Byte, 1994], that have features for processing data in all languages simultaneously are not readily available in Ethiopia.

A lack of good training schools and colleges in computing technology has created a chronic shortage of trained personnel in advanced networking, although Addis Ababa University now offers courses in computer applications and system design. Due to the lack of incentives and of an enabling environment, one seldom finds dedicated computer "gurus." The culture for competition and independent effort to solve complex computing problems is not well developed.

Several small computer companies were set up during the transition period but they have not become a source of objective advice in computing technology. Most companies focused on generating quick income through computer applications training. Qualitative and advanced services in networking and local area network (LAN) management are not available at these companies. Due to the lack of a good professional background, techniques such as application development, system integration, and networking are not practiced.

The telecommunication sector is another underdeveloped area in the country. National telecommunication service in Ethiopia is owned by the government. Analog telephone2, digital leased lines, and radio links to rural areas are major services of the national operator. There is no modern communication technology. Rural telecommunications is still operator assisted. Packet switching and ISDN (Integrated Services Digital Network) are under consideration.

A three to five year queue is usually the case for individual households that wish to install a telephone. Business and international organizations may get telephone lines within one to three months when lines are available. Some institutions have to wait for months to get direct lines.

The above national STI environment dictated the type of network technologies we chose to use, as well as the pace at which we could install and apply them. Policy, regulatory, management, and cultural problems were by far the most challenging - even when compared to the technical bottlenecks faced in setting up electronic mail links to the research community. All the same, electronic networking in Ethiopia was started during one of the most rigid governments in Africa.³

BOX 1 Under-use of Equipment

A recent survey conducted prior to linking colleges outside of the capital city with the network showed that one college received a few modems in 1993 and did not know what to do with them. The modems were locked up in cupboards for two years!

PROJECT DESCRIPTION

Networking activity in the country began under a project funded by International Development Research Centre (IDRC) entitled "Computer Networking in Africa." The aim of the project was to assess the viability of networking between African institutions. The specific objectives of the project in Ethiopia were to:

· improve the exchange of scientific information within Ethiopia by establishing a working, efficient, and reliable electronic network that brings contact with other networks locally, regionally, and internationally;· develop human resources in electronic networking through training, transfer of skills. and university teaching;· develop corps of skilled users in the country through training, troubleshooting, and ongoing technical support;· establish a national scientific network that supports all levels of technologies (Fidonet, UUCP, and TCP/IP, which stands for transmission control protocol/Internet protocol and connotes a full, interactive Internet connection) for various categories of colleges and users under different situations; and· set up an Internet link through development of the user base that justifies the cost of a TCP/IP connection.

PROJECT EXPERIENCE AND IMPLEMENTATION

The first activity of the project in Ethiopia was initiated in September 1991, when the Ethiopian Science and Technology Commission, which coordinates the national STI networks, participated in the first African workshop on low cost networking. The workshop provided training on low cost communications technology and charted connectivity strategies among research institutions in Africa and those in Ethiopia. Subsequently other academic institutions (universities, colleges and faculties) began to link to the main national host.4

To date the main national host serves over 1,000 users drawn from NGOs, research institutions' government, business, and others. Researchers evolved from network-"hesitant" to power users. For example, a connection was made to the chemistry department of Addis Ababa University at the beginning of 1993. At that time, a professor said openly "this is a waste of time and of computer resources." In 1995, the department became the top user of the national hub and a center for network support to Addis Ababa University users.

Network usage and development activity at the university intensified in October 1993, immediately after a workshop was organized for participants drawn from various departments of the Addis Ababa University. The workshop created conditions for setting up a steering committee to promote networking within the university. The steering committee evolved from a few actors to a full fledged university-wide networking group. It established a networking committee to promote networking in the country and develop connectivity to the Internet.

The research community became the second largest user of the national network (See Table 1.) It constitutes 14 percent of PADIS installed sites and 26.2 percent of total local users.

TABLE 1 Composition of PADISnet Users (April5 1995)

User Category

Number of installed sites

Percent of total sites

Average number of users

Total number of users

Percent

Research and academic

39

14

7

273

26.2

NGO

91

32.7

4

364

35

Government

11

4

4

44

4.2

Business

16

5.75

3

48

4.6

Individuals

48

17.25

2

96

9.1

International

72

25.9

3

216

20.7

Media

1

0.4

2

2

0.2

TOTAL

278

100

-

1043

100

Source: PADISnet registration forms.

In addition to developing a sound user base that would eventually justify the upgrade to a full Internet connection, one of the major activities of the project at the beginning involved experimenting with a mix of technologies to select the hardware that was most appropriate for working with poor telephone lines. We conducted the experiments using a mix of technologies to arrive at appropriate solutions. We evaluated the following four connection techniques during the process:

· long distance dial-in to UNIX hosts in North America;· UUCP link through a leased line by running a gateway software;· packet-radio connection to health institutions; and· Fidonet based mail connection to Internet using GnFido gateway in London.

Our first attempt to link researchers in the country used long distance dial-in to UNIX hosts in Canada and the United States. Access to large academic networks such as BITNET was the main interest at the time. A direct dial-in to a UNIX server at Carleton University in Canada was made every day to collect and send messages to the research community. Rudimentary terminal access programs such as Procomm were used to link to the Carleton server. This turned out to be one of the most frustrating means of making connections as the lines broke at almost every trial. Successful connections were so rare that the connection was discontinued after a few months. The cost of telecommunications mounted every month. The cost was made up of:

· long distance telephone charges;· low speed connections (successful connections were at 300 bps); and· additional procedures required as a result of long distance dialing and line breaks (i.e., transferring files several times after line breaks; login procedures to packet switching network in the U.S.).

Paying enormous bills every month for long distance connections forced PADISnet to switch to a more cost effective site. While exploring sites, a link using the United Nations Alternate Voice and Data (AVD) leased line to the Institute of Global Communications (IGC) became cheaper and more attractive. The IGC link enabled us to convert local Fidonet messages into WCP format before leaving the PADIS host via a gateway software. This created conditions for developing local expertise in running gateway software and in experimenting with WCP packets. However, since the line was devoted to UN communications (voice and fax)? it became slow and inadequate for data communications. PADISnet was given only a 15 minute window to send and receive WCP messages. This was an inconvenience and resulted in the final suspension of the connection after ten months of operation.

Another technology used for STI connectivity in Ethiopia was a HealthNet link. PADIS helped to establish this link and to secure its ground station license. The HealthNet ground station, which links to a low earth-orbiting (LEO) satellite operated by SatelLife, was licensed in April 1994 - after 18 months of negotiation with the national telecommunications operator. The ground station was installed at a teaching hospital, one of the largest in the country. This connection was not used as effectively as hoped due to the following difficulties:

· Unreliable power supply. Fluctuation in electric current at the site of the ground station led the system to crash from time to time.· Very complex software. A home-written Fidonet software made configuration difficult. Lack of documentation and troubleshooting tools heightened the problem.· Synchronization of systems. The HealthNet system is composed of a receiver, transmitter, modem, antenna, microcomputer, satellite, and software. Failure in one or more of these systems resulted in a complete crash of the node from time to time.· Lack of an overall plan for user support. The national HealthNet node did not come up with a plan for introducing the technology, connectivity, or regular user support.· Low bandwidth. The uplink and downlink speed of the modem and satellite pass time was limited. With growing user demand for band width intensive documents, such as reports with graphics or images, the system became inadequate.· Marketing. The failure to develop an expansion plan and lack of marketing of the services limited its use in the country.

A direct Fidonet connection to the Association for Progressive Communications (see description on page 189) in London using public telephone lines remained the most attractive of all the above techniques. This connection is currently our main link to the global networks and provides all store-and-forward Internet services. Due to a lack of knowledge about the potential of store-and-forward technology, researchers mainly use the email and bulletin board services of the linkage. Some students and a few researchers are exploring other potential services such as:

· searching the PADIS database using email query;· accessing the Hornet bulletin board system (BBS);· batch Internet services such as ftp, gopher, and archie through mail; and· conferences and Fidonet echo mail.

The Hornet BBS is perhaps the most unique application of low cost technology in Ethiopia for information exchange. Hornet offers an introduction to the potentials of electronic information systems end provides a reservoir of background information and discussion on the Horn of Africa. [Parker, 1994] The bulletin board offers:

· Fidonet conferences (or echoes) to as wide an audience as possible, with no on-site software set-up necessary;· a mechanism to exchange information and skills among users of the PADIS system; and· a central store of information, files, help, useful Internet addresses, and so on.

Simplicity and minimum hardware requirements on the users' side make the BBS more popular. Prompted by this offering, a number of research institutions, including the Addis Ababa University and the Ethiopian Science and Technology Commission, are setting up a scientific bulletin board system. Local bulletin board systems connected to information reservoirs, such as databases and tools on CDROM, are found to be the most useful tools for STI networking under the local infrastructure.

During the last two years, increasing user demands for real time connections, the need for joining the Global Information Infrastructure (GII), and congestion of the host telephone line forced the national node to look for a better technology. At the time of this writing, the node is exploring full Internet connectivity and/or other intermediate solutions. The effort to get a full Internet (TCP/IP) connectivity is hampered by three major problems:

· high cost of leased lines;· institutional and national constraints; and · financial resources to cover implementation and running costs.

Leased Line Costs

The national telecom operator provides leased lines at a very high tariff.
Table 2 shows approximate monthly charges for international leased lines to the United States and South Africa.8

TABLE 2 Cost of Leased Line from Ethiopia to Selected Countries

Type of circuit

Country

Cost/month

19.2 kbits analog

USA

$4,700

19.2 kbits analog

South Africa

$6,200

64 kbits digital

USA

$11,700

64 kbits digital

South Africa

$15,500

Source: Action plan document of a committee for Bringing Internet To Ethiopia (BITE).

National Infrastructure and Political Constraints

The low level of infrastructure as well as policy-specific problems continue to hamper the progress of setting up full connectivity to global networks in Ethiopia. Changing the attitude of the government and of the PT&T (the national telecommunication provider) remains one of the most challenging tasks. Some of the national infrastructure and political constraints we face include:

· competition and unwillingness of the national telecom operator to endorse operation of national cooperative or private nodes,· low participation of government institutions in overall network implementation;· lack of knowledge of electronic communication by government officials that results in delays to approve research network (see Box 2);· education structure and low value accorded to information management, transmission, exchange and use, and· lack of a critical mass of trained UNIX experts or "gurus."

Despite the above problems, the project continued to work towards improving network access to researchers. The bottom-up approach for building a network (starting from users, soliciting institutional support, and then approaching the government) helped us bypass an often difficult bureaucratic structure in the country.

Eventually, the increase in the availability of local "hand holders" and troubleshooters encouraged expansion of the network. The local troubleshooters created institutional champions and then the institutional champions helped users get around configuration problems. In time, we motivated others to get connected.

Reliability of the network has also contributed towards the diffusion of networking technology in Ethiopia. Network reliability is a function of good systems operators (sysops) running the node, advanced computing technology tools, and reliable telecommunication infrastructure. PADIS' effort to make the system reliable by hiring additional system operators9 and acquiring the latest technologies favored the situation.

Other factors that contribute to further network expansion include:

· the availability of dedicated local champions and· the availability of simple network learning tools, such as the Hornet Bulletin Board System.

PADIS management played a major role in leading and supporting networking in the country. The availability of information10 at PADIS provided incentive for several users to seek connection to the network. PADIS made its resources available via the network to satisfy users who wanted to go beyond electronic mail. Individual willingness - especially from top government officials - to embrace existing technology and their demand for better services are perhaps the major motivating factors for expansion to a full TCP/IP. Influenced by their former contacts (usually at universities or NGOs in developed countries ) researchers who are not connected demand quicker and better connectivity. Contact requirements and realization of benefits of networking compelled several users to demand email services even under difficult circumstances. What are the major benefits that prompted researchers to look for electronic networking?

BOX 2 HealthNet in Ethiopia

Often government policies work against the introduction of new technologies. Lack of understanding can load to delays and other frustrations. For example, it was a painful process to secure a HealthNet ground station license for Ethiopia. The process took over 18 months. Simply clearing the ground station equipment from customs took over a month. An engineer from SatelLife, who could have been used for training and set-up, ended up running between the national telecom operator's office and the Ministry of Health. The license was approved with a traffic compensation fee of $350 per month. The ground station, which was donated as national asset to aid in the transmission of health information, was instead treated as a CNN hub. The word ´'ground station" was found to be misleading for technicians who did not bother to know what a packet radio connection to low earth orbit satellite was. The traffic compensation fee was waived six months later.

RESULTS, IMPACT,AND BENEFITS OF THE PROJECT

Reduced Communication Costs

The most outstanding benefit of electronic networking to the research community in Ethiopia was that of bringing a new tool for reducing high communications costs. Ethiopia is a country with one of the lowest pay scales in the world. The following vital figures shows the situation.

Professor or researcher salary per month$ 162.00

Cost of:

1 page fax to UK

7.20


5 page fax to UK

16.10


3 minute telephone call to UK

8.20


5 minute telephone call to UK

13.70

Electronic networking brought a considerable amount of communications cost savings compared to fax, telex, telephone, and courier service. Though email costs are complex and difficult to quantify, we do know that networking offers a far cheaper option to all existing transmission media. (See Table 3.) The following tables and computations show a considerable comparative advantage of email over a fax - without involving costs of other elements such as hardware, or training.

Using compression tools, such as PKZIP, an average document can be compressed 2.5 times. The amount of bytes per minute after compression would be:

Compression factor 2.5

Total bytes per minute 2.5 x 103342 = 258357

High speed modems, with good error correction, synchronized to the receiving modem, can also increase the amount of bytes transmitted per minute. Using 28.8 Kbps modem the amount of bytes per minute would be: bytes per minute of 28.8 kbps modem (1.5 x 285357= 387535)

Using the above figures, we can calculate the cost per kilobyte of different means of communication.

TABLE 3 Raw data used in computation

Date and Time

Number of bytes transferred

Time taken

Bytes/minute

10 July at 6:00

565,583

5:25

104,737

11 July at 6:00

728,410

6:53

105,873

11 July at 1:00

353,496

3:30

100,998

11 July at 5:24

562,256

5:24

104,121

Source: Log files of the PADISnet host July 1995.

In addition to the reduction of communications costs, email made the following contributions to the research community in Ethiopia [Adam, 1994]:

· It facilitated the organization of international seminars, joint authorship, and the execution of joint projects that otherwise would have been expensive and time consuming. According to the Director of the Institute of Ethiopian Studies, Dr. Bahru Zewde, "Global conferences in Ethiopia and those on Ethiopia in 1994 were cost effective and successful because of email connections."· It inspired the introduction of standard computing operating systems such as UNIX. Recently the School of Information Science for Africa (SISA) purchased a UNIX server as a networking backbone. The Mathematics Department of Addis Ababa University received a UNIX computer in connection with its networking effort from Ethiopians residing in North America. Deans and department chairs were forced to plan to "wire" their offices and create local area networks (LANs) to cope with an increasing demand from researchers to have independent accounts on work stations.· It facilitated a switch from the hierarchical model of organizational procedures to the informal interaction among researchers in the country and those worldwide. User feedback indicates that delays in signing regular correspondence were eliminated upon arrival of email. (See Table 4.)

Email also brought about the reunification of the research communities in Ethiopia with those elsewhere. It improved family and informal ties between Ethiopians residing overseas and those at home."

Improved Training Options

Electronic networking facilitated the availability of more national and institutional trainers. It created the expansion of knowledge on the use and benefits of networking and TCP/IP connectivity. The methodology used in training contributed to improved skills in the country. The national host uses the following techniques to train users.

TABLE 4 Cost Comparison Between Fax Message and Electronic Mail Between Addis Ababa and London

Medium

Average kilobytes/per minute

Cost/Kb

Cost ratio to a fax

Fax12

40

0.18

1:1

Email using 14.4 modem

237

0.0093

1:19

Email using 28.8 modem

355

0.0062

1:29

Introduction

Users are introduced to key features of networking. This covers basics of local networking? what it involves, cost, and types of services.

Site configuration and training

Site training is meant to connect researchers to the network. It involves configuring the hardware, setting up telephone lines, providing hands-on training on system usage, and training in basics of troubleshooting.

Ongoing assistance

The host provides continuous online assistance to improve users skills in dealing with different utilities and to work on improved and cost efficient techniques.

Workshop

This is used both as marketing tool for the national node and for advanced training. Experience indicated that initial introductory workshops overload users with new concepts. Successful workshops could only be held after usage of the networks. Advanced workshops were found to be useful in promoting connectivity.

Improved Skills

Another considerable impact of the network was an increase and improvement of the skills at the main national node. We developed skills in managing networks, dealing with administrative problems, developing techniques to respond to users' problems, managing system malfunctions, developing tools and guidelines, and gathering data. The experience we gained was immense.

Improved Access from Rural Areas

The spread of the network to the rural area is one of the most significant achievements of the project. Despite technical problems, such as unreliable connections and more focus on the capital city, a number of users were able to link to the PADISnet from outside of Addis Ababa. NGOs with field offices outside of the capital found networking a convenient tool for sending logistical information and data on field situations.

Doctors with no telephone connections were able to use networking. Medical doctors from Medicins Sans Frontiers, headquartered in Belgium, who were working 800 kilometers from Addis Ababa were able to send messages and data to their correspondents all over the world. Lacking telephones at their field site, they send messages via small airplane to Addis Ababa for uploading to their host to resend to PADIS. All international messages received are sent back by the same plane the following day. Other mobile field researchers from the same location were able to link their modems to the nearest city where phones are available.

A number of colleges outside of Addis Ababa are discovering the value of networking. They face double isolation - from both the developed world and the country's capital where most of the activities and decisions take place. Everything in the country is hierarchical and centralized. Decisions at higher levels (at the capital) take a long time to reach institutions in remote areas. Networking can reduce the isolation and make communication more efficient.

Connectivity to the rural areas facilitated the coordination of resources and the diffusion of technology. Connectivity at the Mekelle College (710 kilometers from Addis) recently stimulated the establishment of a LAN for connecting individual professors to the global network. A project run by the Ethiopian Science and Technology Commission to link over thirty-six colleges outside of the capital stimulated further international cooperation between the colleges and other universities worldwide. The Commission intends to "wire" all colleges and high schools in the country while working on a full TCP/IP connection at the capital.

ANALYSIS OF LESSONS LEARNED

Lesson 1

The resources required for achieving access to knowledge and wider national coverage are a fraction of the benefits gained by building networks. Coordinating activities at a national level assists in realizing cost effectiveness and efficiency. Networks can subsidize themselves. Lack of an infrastructure should not be seen as a major bottleneck to networking.

There are two major goals of electronic networking of the research community:

· wider coverage of users by reaching many scientists, researchers, students etc.; and· improving access to knowledge by bringing global information resources closer to the scientific community.

Attaining these two goals needed institutional and infrastructure readiness from the outset. When these two goals were first proposed to managers and policy makers their initial reaction was that "there are no resources" and "we can not do it under our telecommunications situation"!

Lesson 2

Training is the main element in infrastructure development. Do not start training on DOS based networking at universities, start with UNIX' which I believe to be the multi-user system of choice for networking universities.

Wider national coverage requires local infrastructure development. Local infrastructure means not only the physical set-up of machines and software, but also building a sense of awareness and increasing the participation of users. Training should cover a wide range of groups at different levels and should introduce all standard technologies such as UNIX to the academic community from the very beginning. Starting training in the DOS environment limits universities' capacity to move quickly on to sophisticated techniques. Using DOS for networking universities instead of UNIX is like using typewriters instead of word processing. Good documentation saves time.

Lesson 3

Build as many local area networks as possible. A LAN is a building block for Internet connectivity.

Capacity building in universities requires setting up of a university-wide network for inter-connecting many LANs, serving individual departments, libraries, and administrative offices, and offering the connections to the Internet through gateways. LANs are building blocks for good research networks.

Lesson 4

Good network management practices should begin when the number of users is very small. Good network administration can prevent user and donor frustration.

The professional management and administration of a network should begin as soon as you have two users. A well-run network needs a good billing structure, well developed methods for gathering data on the type and nature of users, and mechanisms to backup log files, document problems, and track progress. Put these systems in place before the number of users grows to an unmanageable size.

Lesson 5

Networks should be interactive on both an inbound and outbound basis.

Interactive sysops are the best assets of good networks. Inbound interaction with local users and outbound links to other networkers worldwide assist in sharing knowledge, resources, and time. Documenting tips and passing them on to users helps networks to expand.

Lesson 6

Using software with simpler user interface is important.

One of the major components in infrastructure building is the provision of simple tools to users. Users should get uncomplicated tools. Using software with good user interfaces reduces frustration and improves users' interests in networks. The community of Macintosh users should not be neglected.

Lesson 7

Do not focus only on global information sources. Assist in building local capacity in STI exchange.

Improved knowledge access can be achieved by creating a collaborative atmosphere within a country and by connecting to global networks. National STI sources should be strengthened to enable local exchange of information and collaboration. Networking should be built with maximum participation of local STI providers and users.

Lesson 8

Do not promote the concept of subsidy. Promote the concept of "pay a small amount to keep your network healthy."

Networks should sustain themselves. Income-generating should start at the beginning. Users should be told to pay for services. Subsidizing researchers is important; researchers should, however, know the costs and participate in paying for services from the beginning. The payment could come from university budgets or projects.

CONCLUSION AND RECOMMENDATIONS

Scientific and technological information in Africa needs to be strengthened. African input to global information resources emerges from the local STI institutions. Databases and low cost networks will ultimately serve African information needs. This is especially true for local loops where telecommunications and other infrastructure continue to be problematic. The international link itself is also inadequate. Except in South Africa, the other eleven African countries with Internet connectivity have a maximum speed of (usually) 19.2 Kbps. Reaching the global community and providing African STI to global users need efforts in strengthening the local capabilities both in networking and in information generation, management and dissemination, while upgrading the bandwidth for international connectivity.

Quality, reliability, and sustainability of African STI and networks is becoming another critical focus. Surveys and experience indicate that databases on STI lack good quality and are mostly unreliable. Database and information systems development in the region need to build quality tools and strategies for sustainability of African STI databases. The quality and sustainability of low cost electronic networks serving STI in Africa should also be improved.

Electronic communication will become a major tool for distributing global information to Africa. This does not mean African researchers need and can afford full TCP/IP connectivity. The key questions become: Is the African STI environment ready for full Internet connectivity? Will communication be two-way (from Africa to developed world and vice versa)? What will be the overall implications of Internet for a few in the capital cities versus those who are isolated due to a poor infrastructure and who need to break the isolation?

The answers to these questions remain a challenge to STI providers in Africa. African STI institutions need Internet connectivity. Internet connectivity means full participation of local providers and STI institutions in information generation and usage on the global networks. The current trend in most African countries focuses on Internet "proximity": bringing Internet closer to the users with no expansion plans for wider connectivity using low bandwidth links. Encouraging STI institutions to participate in information exchange will become the next challenge after bringing an Internet link to a country.

The situation in Ethiopia, which is often similar to that in other countries in Africa, indicates that careful analysis of the national STI environment should be made to design the most appropriate solution to improve overall research capacity in a country.

Efforts should begin by building the capacity of national science and technologv institutions, such as research institutions, colleges, and small STI service providers. They should be encouraged and trained to manage information and use global information sources. Interim innovative services, such as connections to major databases or Internet services via local Bulletin Board Systems, should be available. Users should be made fully aware of the services available to them even through existing low-cost, store-and-forward technologies.

One of the major tasks facing African STI networks is a lack of knowledge of what is happening in the same region, the same country, or even the same institution. We should develop directories, dissemination tools, and guides on STI networking efforts. We should disseminate via our networks STI meta-information (information on scientific and technology information). Without a solid regional STI base, connectivity alone cannot serve its purpose.

NOTES

1. Ethiopia's longest crisis and war years were between 1974 and 1991. During the transition period (1991-1995) the government attempted to eliminate centralized and rigid policies to make favorable conditions for economic growth.

2. International telephone linkage to major cities in the world is relatively good. Users can connect with up to 28 Kbps to London without difficulty. A 300 to 900 CPS connection can be achieved using error-correcting modems. It is often easier to telephone internationally than across regional cities. The call from Addis to Mekelle is often compared to that of Addis to London. Cost of international communication is also high. The tariff was raised by 75% in May 1994.

3. The Ethiopian socialist government which had fallen apart in 1991 promoted strict policies on communications. Modems were not allowed due to reasons related to national security. Any move towards opening up communication was looked upon as a "war against socialism."

4. The national host was set up at PADIS and the network is called PADISnet. It is open to all categories of users on a fee-for-use basis.

5. Since April 1995, the number of installed sites has increased by an average of 5 per week, thereby nearly doubling since that time.

6. WCP is a UNIX based store-and-forward technology. Developments were fast and based on international cooperation between programmers. It is a suitable start up technology for the academic community.

7. Fidonet is a store-and-forward networking technology by which computers poll each other to exchange email and conference messages. Ability to work over poor telephone lines and software that enables off-line working make Fidonet attractive to NGOs and African institutions with small budgets.

8. Leased line costs to South Africa are almost 1.5 times those of links to the United States. There is double hop to reach South Africa.

9. In June 1995, PADIS used an intern - an Ethiopian PhD student from Columbia University - to look at university-wide networking and to improve the reliability of local systems.

10. The PADIS information system consists of bibliographic databases on African socioeconomic development, databases on African experts, on-going research projects in Africa, research institutions in Africa, and other numerous complementary and specific databases on Africa.

11. The author received numerous requests from Ethiopians studying abroad to link to their families at home. Families separated during the war and those who fled economic problems are beginning to use networking to reunite with their families. Ethiopian researchers abroad who discovered the existence of the linkage are connecting back with their former teachers and institutions.

12. Fax costs vary based on text type. Graphics can be 100K or more. Regular fax message can be 10 to 70K in size. An average size was taken for this calculation.

13. A 28.8 Kbps was taken to be 1.5 times faster than a 14.4 Kbps modem.

BIBLIOGRAPHY

Adam, Lishan. ( 1994) Sustainable Academic Networking in Africa: System Operator's Perspective paper published for Advancement of Science,Accra, Ghana.
Beyond Windows: Globalization of Windows. Byte June 1994, pp. 177-183.
International Association of Universities ( 1991) University Based Critical Mass System for Information Technology.
National Research Council ( 1991) Status of Scientific and Technological Information Systems and Services in Selected African Countries National Academy Press, Washington, D.C.
Parker, Ben. (1994) The Hornet: The Horn of Africa's Electronic Information Exchange. Addis Ababa, Ethiopia