Wiring Africa's Newest Country
Email in Eritrea…
September 1998
In the beginning…
Email in Eritrea? Should be easy, right?
The carcasses of previous attempts
Phone line weather and marathon modem test sessions
Reliable but slow (and expensive)
The first expansion: UUPC Internetworking with 286's.
Stepping out with a 501 (3)c, Linux, smail and uucbsmtp
Back to Eritrea: The first factor of 5
The second factor of 5: replacing that little resister.
The problems of success
Showdown at the University, and who is Craig Harmer anyway?
Branching out: "Hey Mr Email Man!"
Government Agencies, Embassies, and NGO's
Going commercial with Eritrea Online and Tfanus
The Il-legality of Email.
Lightning Strikes! and other 'Challenges'
ER: the Top Level Domain
Indigenous Nerds
Internet the Eritrean way: The shape of things to come

This is the story of how a friend (Craig Harmer) and I (Robert Van Buskirk) set up the email system in a new African country almost by accident. Technically, it should have been easy; it wasn't.

We found that out on the electronic frontier, it isn't the laws of mere mortals that govern but instead, it is Murphy's Maxim that rules supreme.  But in spite of both supreme and human law, in the end we managed to put something together that not only works, but works well.

And like any piece of real history, the story of how it happened is filled with success, failure, collective culture and individual personality.

To my Eritrean Friends and Colleagues: This article is written from the perspective of a foreigner for a general audience. As such it perhaps makes light of many infrastructural problems, regulatory misunderstandings, and organizational problems. I know that there is a lot of historical and social context to these problems which I do not present. But I also think it is healthy to take the problems one encounters with a sense of humor and challenge.

I know that at times there can be a fine line between a sense of humor and an attitude of disrespect. I hope that for all concerned, I have stayed on the correct side of that line.

In the Beginning…

The story begins about three years ago, in February 1995. Imagine yourself in a comfortable and quaint University, half way across the world. Everyone speaks at least a little English, people are so friendly they never let you pay for lunch when you go out. Students are excited and enthusiastic. The weather is perfect (70 degrees and sunny all year round). There is an espresso bar on every block and the coffee is good.

This describes the University of Asmara in Eritrea--Africa's newest country--where I was working about four years ago. I am a physicist, and a politically minded person.  I was having a great time working with my Eritrean colleagues to help rebuild their educational system after the destruction of their gruelling 30-year war for independence.

There was just one problem though: my companion/life-partner/significant-other, Angela, was 11,000 miles away in California. And to make it worse, phone calls were $3/minute, letters took 2-4 weeks to arrive, and I was being paid a (high) local salary $400/month. Not only would two hours per month of phone calls be less-than-satisfactory, they would leave nothing for food.

Because email is fast, cheap and reliable, it was the obvious solution for communicating with the home-front. But until 1995, there was no reliable E-mail, no Internet, and no good international modem access in Eritrea.

For us netizens in the richer countries Internet access has become a psychological, if not physical necessity. It has become a component of our basic infrastructure…like electricity, lighting, running water, and flushing toilets. And with recent investments in telecommunications throughout the world, this infrastructure for information hyper-access exists in almost all of the world's nearly 200 countries--even in countries where the vast majority of the population do not have those other more physical necessities like running water and systematic sewage disposal …

Email in Eritrea? Should be easy, right?

But Eritrea is not like other countries. In 1995 it had no email, and I needed it. Fortunately, Craig is both a Unix kernel hacker and a good friend, and when I complained to him about the lack of email in Eritrea his response was: "Email? It should be easy to set up…" And though he was dead wrong on the 'easy' part of it, his sense of personal and technical pride did inspire him to help set up some email that actually worked under Eritrea's peculiar circumstances. And even though lightning struck our modems, email was outlawed, and our international modem connections were nearly impossible, it is now the fastest and most efficient email system in Eritrea today. Every day, the system is shoving nearly 1000 pages of text through a regular pair of phone wires in Craig's living room at cost of 1-2 cents per page in spite of the $1.60/min international phone rate. It now probably ranks on a par with the BBC and CNN as one of the countries more important modes of international information transfer.

Our email endeavor started simply enough. One night in February 1995, Craig came over to my place with a used modem (a telebit T-1000 to be exact) and a share-ware diskette. It was the night before I was heading back to Eritrea, and I had just bought a used 286 laptop. Craig had set up his Unix-box at home to do UUCP (Unix-to-Unix-Copy) and had spent a late night learning the DOS implementation (called UUPC) and configuring his home computer as a mail server. We installed the software, hooked up the modem and it worked without much trouble. Craig seemed right: Email was easy.


Once I arrived in Eritrea, the used Telebit modem was detained by customs along with some donated computers and a printer that I had also brought along. This was as expected. With a per capita income of $150 per year, there isn't much income on which to collect income tax, so every piece of electronics that makes its way into the airport has to pay its share, whether or not it is a donation. But with the cooperation of the University property manager we were able to get all of the equipment out in about a week…which was record time actually.

When I had arrived, I found that the director of the Eritrean computer center had been sent off to Sweden for study and replaced by a jovial, but frustrated, Georgian, Vladilin. He had been working in Africa for the last 15 years, and just arrived with several former Soviet academics on a United Nations Development Program grant of $1,500/month. Not bad pay compared to academic positions in the former Soviet Union.

Vlad found his appointed task of teaching Word Perfect 5.0 to University students about as mentally stimulating as a shot of novacaine in the brain. He welcomed the idea of an e-mail project as something that could liven things up a little bit around the computer center. He loaned me a copy of the computer center key, and gave me free reign to get some sort of computer communications going.

The carcasses of previous attempts

Lying around the University of Asmara (UoA) computer center were the carcasses of previous attempts of trying to establish a National e-mail system. The virtual community of Eritreans abroad--called Dehai ( http://www.primenet.com/~ephrem)--had donated a Sun server and a Telebit 2000 modem in an effort to set up an e-mail node at the UoA. But in spite of having the correct theoretical design and configuration, running the system in Eritrea was just too difficult. Nobody at the University of Asmara knew Unix, the hard disk kept crashing, and the 110 volt modem had fried after being plugged into 220 volt power.

In addition to the Dehai effort, there had also been visits to the University from PADIS, the Pan African Development Information System, based in Ethiopia. They had installed some FIDONet software, a PC-based store and forward email system common in Africa. But again the effort failed because of a lack of timely technical support, and the difficulties of calling and operating an email system over poor phone lines to Ethiopia.

A third e-mail effort in Asmara was made by a commercial provider, Bashir Computec. It succeeded in running a sometimes-available email link called AdalNet through PADIS. But because of poor phone lines, and not being able to charge customers (because PADIS was non-profit), Adalnet was unable to provide reliable email access for e-mail users in Asmara.

Would the effort of Craig and I be another in a long list of Eritrean e-mail failures? The two distinct factors we had going for us, were (1) Craig's technical skill, and (2) My intense motivation to stay in touch with Angela via e-mail.

Phone line weather and marathon modem sessions.

When all the equipment was in order and I had access to the computer center, I sent a fax to Craig and gave him a schedule by which he should call Eritrea. Our plan was for Craig's computer to call the computer center a few times a week from the U.S. and exchange any queued up letters. We'd set a regular schedule and then set up a few user accounts, and viola' email would be operative.

The first call or two went without a hitch, but after that it seemed the modems would struggle making a connection. And even when they did connect, half the time they would lose it and have to start over. Listening to the modems it was easy to understand why. The calls would cut in and out, and there was a lot of noise on the line (you could hear the local radio station for example). Outside the University the phone line was strung as a long piece of parallel strand power cord that whipped around in the afternoon winds. If it was a windy day, you knew there would be no e-mail. Something had to be done.

After several weeks of bad connections Craig and I decided we had to get systematic about the problem. So we scheduled a marathon connection experimentation session. I didn't know much about modems or modem settings, so Craig had to walk me through the first round of technical experimentation. I sat up late night at the computer center with the bats and swallows flying outside the fourth-floor window. Craig would give me a voice call with a batch of modem settings to try. I had one minute to edit the configuration file and make the settings. Craig's computer would then call three times and try connecting different ways. Then Craig would make a voice call again with another batch of setting for me to try. Several hours and 30 attempts later we had the best that we could come up with. A group of settings that worked marginally.

But we still weren't satisfied. We knew that Telebit modems had to work better than this, even if the phone lines were crappy. Craig went to the manufacturer, and he found that there were in fact a set of registers that controlled the details of the data modulation. Telebit technical support even called Eritrea to test different settings. And with in a few months, and after testing almost every modem setting in the book (and even some not in the book) we finally had a working modem configuration, using Telebit's proprietary modulation protocols. This solution seemed to work reliably (if slowly) over our crappiest phone lines. Victory was ours--almost.

Reliable but slow (and therefore expensive)

Out of the kindness of his heart, Craig was footing the bill for all this e-mail experimentation. With a few calls per week, and five to ten minutes per call, this was $50 to $150 per month. I was staying in touch with Angela, which was my goal, but it was still costing $0.30 to $1.00 per message. Better than fax ($3) but still expensive.

So even though the email was now working by June '95, it was obvious that several things still had to be done: First, it had to get more efficient because $1.00 per message was still too expensive. Second, if Craig was going to be donating money, he should at least get a tax break. Third, we needed some way of providing more general access to others. Fourth, we needed some scheme for recovering our expenses. And last but not least, we had to set things up so that the whole system didn't collapse when I left Eritrea.

The irony is, while it took us about a minute to list the things that needed to be improved, it would take us nearly two years to implement them.

The first expansion: UUPC Internetworking with 286-s.

While we were getting things to work, Craig and I were having a little debate. Craig insisted that running PC servers with DOS was for the birds, and that if we upgraded to a variety of UNIX we would have the flexibility to make things work better. I insisted that with dust, lightening, bad electricity, and WordPerfect 5.0 counting as computer experience, there was no way I could sustainably keep a UNIX server running.

Neither of us convinced the other over the short term, which was good. This means I set my mind to using the UUPC to build a little store and forward email network, and Craig set his mind to designing a Linux server that would be custom-configured for Eritrea work.

The UUPC email 'server' (and I use the term loosely here), would be set up to be in a dedicated answer-the-phone mode during certain times of day. And the users would have UUPC set up on their computer and would type a batch file command called 'getmail' to fetch their mail from the server. This way we actually had a little email network in the University with half a dozen nodes: Physics, Math, Marine Biology, Agriculture, and Engineering. On each machine people could set up as many user accounts as needed so we could have a dozen or so users utilizing each of these machines.

All of this time, we also set up a node at the US Information Agency office. I first came to Eritrea as a Fulbright Scholar, and head of USIA was helpful. He had a Telebit 2000 at his office, and better phone lines than the University. This allowed us to use his office as a back-up email route for the less reliable University.

With the various University departments and the USIA, the nucleus of an email system was growing.

Stepping out with a 501(3)c, Linux, smail and uucbsmtp

After a semester of struggle (both email, academic, and bureaucratic), I returned to California for a break in July 95. I wasn't returning until October, so Craig and I had a few months together to make some changes.

First we set up a little non-profit project to administer donations and finances. We got help from my friend and mentor Bob Lange, a physics professor at Brandeis University. I met Bob through Nicaragua solidarity organizing that we did in the mid to late 80's through an organization called Science for the People. Bob had gone on to organize innovative science teaching camps in Zanzibar, Tanzania. He had set up a non-profit of his own to administer the finances of his work: the International Collaborative for Science Education and the Environment (ICSEE). The name is a mouthful but its descriptive.

Well email development in Eritrea is international, collaborative, scientific, and educational. So since we have four out of five and since we are doing good work, he agreed to take our project on under his non-profit wing. We became the Eritrea Technical Exchange (ETE) of the ICSEE.

That was the easy part.

Next Craig forced me to learn minimal Linux system administration skills. He got the Linux Bible. We went computer shopping. And we installed and set up a Linux server. He wrote a customized transport for SMAIL which compressed and batched email before sending it via Taylor UUCP. He called it Unix to Unix Compressed Batched Simplified Mail Transport Protocol or UUCBSMTP… another mouthful.

The rationale for the new transport was that a lot of time of the UUCP transfers was consumed setting up the transfer of the individual files. In UUCP, for each message, the routing and transport information, and the message are sent as separate files. So each message is two files. In addition there is a second or two of latency going through slow modems and 30,000 miles of satellite link, so it took 5-10 seconds just to set up the file transfers. This means 20-40 seconds of every minute was wasted with the computers saying to each other "OK I am going to send another file are you ready" "Yes I am ready send that next file now…"

BSMTP allows you to put all the messages in one file…it just uses key words or symbols to separate the different parts of the message…so this in itself saves 30% to 50% of the time needed to transfer the data.

The compression adds another factor of three if you use software compression (gzip in this case). The lines are so bad, that the modems often don't negotiate good modem compression. Also since modem compression is run-time compression, it is not as efficient.

In combination, UUCBSMTP improves throughput by a factor of 5 or so. With that, the price per message drops to about $0.10 per message: Much more reasonable.

Back to Eritrea and The first factor of 5

When I returned to Eritrea in October 1995, I was now working at the Research and Training Division of the Department of Energy rather than the University. This was a good thing. For one thing, the president of the University felt a need to micromanage the details of University activities (a tendency of many University presidents I am sure). This meant that even minor things had to be approved through his office. And with only one person to approve things for over 100 academic staff, it gets hard to get things done. He had valid reasons for such rigidity I am sure, but it just wasn't a good environment for innovation. And innovation was definitely necessary on the email front.

At the 'Energy Center' where the Research and Training Division was housed, the situation was the opposite. They had gotten funding to build an innovative research and training center emphasizing renewable energy development. Now they needed people to populate the facilities and translate the equipment and infrastructure into concrete results. When I arrived, my jovial, slightly frustrated, and supportive 'boss' Mnat put his hand on my shoulder and said: "Your job is to do what you think is right for the Energy Center." I couldn't have asked for a better job description.

Of course, one of the things right for the Energy Center was facilitating international information exchange through email. So the Department of Energy paid the import taxes on the server, sped it though customs in two days, and allocated one of its only two phone lines to email. This was crucial support for the email effort at a critical time.

And when we did set up the Linux server….it worked just as predicted. About a factor of 5 faster than the old PC-based UUPC system.

The second factor of 5…replacing that little resister

Luckily, our first factor of 5 improvement in efficiency was not the last. Craig noticed in the back of the Telebit manual was a description of how one amplifies the modem output signal by replacing a plug-in resister on the modems circuit board. Essentially we could make the modem 'shout' on the phone line. This was great because that was another noticeable feature of Eritrea's phone lines.  The other end seemed somehow distant and hard to hear.  Not to mention the local radio station that was often heard playing on the phone line.

By making our modems 'shout' we were then able to connect at 14400 baud. This was about another factor of 5 above what we were getting without amplification. We were now down to a cost of only a few cents per message. Finally the email worked like it should. We would later get another factor of 2-3 by converting to 36 kilobaud modems but that was icing on the cake.

The problems of success

So by November 1995, it seemed that we were stylin' with our incipient email system. It was efficient and low cost. We had a dedicated server, a dedicated phone line, and more and more people were getting interested in getting connected.

There are two problems associated with such success. The first is that demand may increase faster than one's capacity to satisfy it. The second is that as a new communication medium grows in popularity and notoriety different people and institutions my feel obliged to assert control over the new, unregulated medium.

Showdown at the University: and who is Craig Harmer anyway?

The first struggle for email control occurred at the University shortly after the new server was installed at the Energy Center. The jovial, but now disillusioned, Georgian had been eased out of his position at the computer center and replaced by the returning Eritrean head.  [To protect the not-so-innocent, lets say that the name of the computer center head is Tesfai--the Eritrean equivalent of Jim]. Tesfai didn't like his academic hosts in Sweden, and they didn't like him. So his Ph.D. study was cut short after six months, and he was sent back to Eritrea.

I didn't particularly like Tesfai either. He had a reputation of yelling at his students when they did something wrong, he would talk about certain people as if they were stupid, and seemed a little conspiratorial, controlling, and paranoid. But hey, he had a very frustrating job, and it takes all types to make the world. So I was willing to work with him on an objective basis to keep the email working in the University.

Now as head of the computer center in the University, he had this email system running, but he really wasn't in control of it. Craig's machine was calling from California during particular times of day to a computer in the Center, which was programmed to answer the phone for both the U.S. server and any academic departments who wanted their transferred mail. It seemed weird that some stranger in the U.S. would pay a few hundred per month of his own money so that the University would have email. The other problem was that the new server was not at the University, but at the Department of Energy. And the Department of Energy doesn't have a 'mandate' to be running the country's email. People--at the University at least--were used to things being run from the top-down, not from the bottom-up.

So one morning, Tesfai called me in for questioning. He apparently was not happy.

    `Who's paying for the email?' he asked
    `We have it organized as a non-profit, but basically a very generous friend of mine who is a computer expert and very interested in this project.' I gave him the address of the project, and noted Craig's picture hanging above the mail server.
    `What is this new server?'
    `Its a Linux system running on a 75 megahertz pentium box that is at the Department of Energy. With the improvements we have implemented it does email transfers about 5-20 times as fast as the transfers at the Computer Center'
    `The server should be centered here.'
    `I am at the Department of Energy…So it is at the Department of Energy--which provides great support-- and I was the one trained on maintaining it and that is where I work, so there it is. Again I am happy to work with you to install it here, we could start next week.'  I was trying to be concilliatory.
    `Well, I will have to talk to the authorities.' he warned
    `Fine, in the meantime are you going to let the computer answer the phone?' (the phone has been ringing about 10 times during this conversation as the U.S. server tried to call the computer center server. Tesfai just picked up the receiver and put it down, costing Craig $1 every time he did.)
    `Not until the issue is resolved.' he responded.
    `You mean you are just going to unilaterally shutdown the University email?'
    `Until things are discussed, yes.' he replied.
    `So I can tell everyone that Tesfai shut down the email.'
    `OK.' I threw up my hands and walked out.

A little later in the day I ran across the 'gang from the UNC,' a small group of folks who worked at the University through a cooperation program with the University of North Carolina sponsored by the Fulbright program. They were concentrated at the Law department which had recently gotten an email connection through the system. One of them, Rich Rosen is a nice guy who was also a gratefully rejuvenated email addict. We agreed that as a temporary measure Rich would act as University email courier. Every night he would use his home phone to fetch mail from the Department of Energy server. Then every day he would bring the laptop into the University, and allow the departments to pick up their email over the internal University extension lines. We had the reconfiguration done in about a day, and email was flowing again to the destined departments within 48 hours--in spite of the shutdown at the computer center.  Rich was the University email angel...and dutifully carried email back and forth to the University on his laptop for months.

Eventually Tesfai came to his senses after several weeks.  He got over his paranoia, and we configured the University as a regular Linux-based server. The old 486 with a 200 megabyte hardrive that we used in 1996 only just recently was upgraded to a more powerful machine.

Branching out…Hey Mr. Email Man!

As soon as word got out that we had a real, working email system in Eritrea, all of the  information-deprived expatriot net-addicts started coming out of the closet.  Strangers would literally walk up to me and ask: "How do I get connected to email?  I've GOT to get email."  And as an email junkie myself, I could understand.

In one of the more memorable scenes about this time I was riding my bike through the dusty streets near the Energy Center.  And I noticed a white pickup truck turn around and start following me with the driver staring at me intently. I decide to stop and see what the guy wanted.   The truck stops, a man leans over, rolls down the passenger side window and asks in a Southern accent that seemed out-of-place in Africa: "Hey! Are you the guy who does the eee-mayl?"
    "Well yes" I answered a little hesitantly.
    "I'm Thad, I'm sooo glad to meet yah.  I have heard so much about yah... and just yesterday I was giving witness..."  And he explained how he had heard about the email from Richard (the guy shuttling the email into the University with his laptop) and he wanted to know if there was any possibility of him getting hooked up.
    I assured him there was no problem as long as he reimbursed the non-profit in the States for the expenses, scheduled a convenient time for me to come over, and fed me lunch.

Goverment Agencies, Embassies, and NGO's

So during this period, email spread via word of mouth, and people started dropping by and calling at the Energy Center 'needing' email.  Our basic priority was Eritreans first.  They had the hardest time getting used to e-mail.  And the government was providing the basic infrastructure.  So any government agency that wanted to try email got a free connection just so long as they reimbursed us $20 for the 2400 baud modem we gave them.  These connections included ministers, and assorted government departments totalling a couple dozen within the first few months. In addition, the different foreigners (including the U.S. ambassador at one point) and foreign NGO's that wanted connections got them if they could reimburse the non-profit for some of the expense.  This allowed us to decrease the negative cash flow so Craig didn't have to dump so much of his own money into the venture.

The problem with this popularity was that being an email provider was not my real job.  And desperate foreigners dropping by the Energy Center with technical email problems were trying the patience of my institutional hosts.  Forturnately, I had a brilliant second year physics student--Daniel--working with me at the Energy Center.  He is the type who in spite of having only about a year of computer experience could take a software manual and start programming batch files and macros within a day.

Daniel helped deal with the routine email installations and problems, and we made it clear to the 'customers' that email problems were to be dealt with outside of working hours and at their place not ours.  You can give us a call, but no visits please--unless you are government.

Going commercial with Eritrea Online and Tfanus

In spite of our popularity, there actually was no way that email in Eritrea was going to survive if it depended on Daniel and me running around doing installations and servicing everyone demanding access.  We were working by the good graces of the Energy Center, and we were volunteers.  The only way email was going to survive in the long term was if it went commercial one way or another.

The problem with commercial email in Eritrea was that it was just too hard for 90% of computer businesses. Almost no one knew Linux, and you needed phone lines, an extra computer or two, and an ability to reimburse our U.S. non-profit for the phone bills.

Fortunately about this time, two Eritreans, Menghis and Tewelde, independently returned from the U.S. and set up their own separate computer businesses: EWAN and Tfanus respectively.   They both decided to set up email services.

The first one was Menghis Samuel with his 'Eritrea Online.'   Menghis about two decades earlier was fighting with one of the two factions of the Eritrean liberation forces (the Eritrean Liberation Front).  His faction had lost in a civil war that occured behind the frontlines of the independence war in the early 80's.  As a war refugee, he later arrived  in the U.S., got a degree in Engineering and joined ATT--eventually working in the digital signal processing section. During his stay in the U.S. he rallied behind the faction that eventually won Eritreas independence in 1991 (the Eritrean People's Liberation Front....the party of the current government).  And now he was leaving his lucrative ATT job and returning to Eritrea to open a computer business with significant financial backing from patriotic Eritrean investors.

Menghis is also one of the founders of Dehai, the virtual community of Eritreans on the Internet. As such he also helped organize one of the first attempts to start email and Internet at the University. He knew UNIX, had phone lines, and some capital.  His business was a natural for commercial email services.

Tewelde didn't have the financial backing or the phone lines that Menghis had. But his previous position was at Hewlett-Packard (HP).  Also his assitant, Tedros, is obessessed with computers and the Internet. Tewelde's experience, and Tedros' obsession and skill have both been definite assets in Tfanus' email service development.

Both businesses quickly installed Linux servers, learned the system, and started providing services.  By July 1996, commercial Eritrean email was in place.  Sustainability now seemed assured, in spite of a few 'minor' problems.

The Il-legality of Email

One minor problem with commercial email was that some people in the government (e.g. the head of telecommunications for one) considered it an illegal infringement on exclusive rights activities of the national Telecommunication Services of Eritrea.

The reasoning went like this: The government was planning on opening telecom markets in a controlled and regulated fashion.  But so far no regulations had been promulgated (actually about 80% of regulators had no experience with email).  So since there were no regulations allowing email, it was therefore illegal.

This is a classic legal dilemma: Is something legal only if it is explicitly allowed?  Or is it il-legal only if it is explicity dis-allowed? The answer divides the disciplinarians from the libertarians.

One of the email providers even received a letter from the head of telecom, asserting that while email would be legal in the near future, it was currently illegal and all such email services should be stopped immediately.  The irony of this declaration was that ministers, the ruling party, and even the computer section of telecommunications itself were using the email services.

All email providers tacitly agreed on the appropriate response to this declaration.  If email was going to be shut down by the government we were not going to do it for them.  We would  insist on first having our day in court.  And fortunately, that day never came.  The letter was not enforced, and the promised regulations were finally promulgated about  two years later in March 1998.  Three years after it began, email services finally had a legal framework in which they could operate.

Lightening Strikes! and other 'Challenges'
 Another thing that took the 'easy' out of Eritrean email was the fact that the environmental conditions were at constant war with our equipment.

One minor example: While the laws of probability may say that it is highly unlikely for lightening to strike the same place twice, the laws of Murphy say otherwise, especially if you are unprepared for it. In fact it seems the infrastructure in Asmara is specifically designed to distribute lightening surges to email users.  Power poles are made of metal pipe and set in the ground, so they make good lightening rods.  And 15 Amp power cable which is good at transporting the surge current, is often used between the neighborhood distribution pole and the customer premises.  So after the lighten strikes the lightening-rod a.k.a. utility-pole, its surge current is efficiently distributed to dozens of neighborhood telephones.  In our first experience with this, we had turned off all computer equipment because of an electrical storm, but the lightening strike came in over the phone lines and took out a Telebit modem, a 2400 baud modem, and even erased the bios on the CMOS of one of our computers. That rainy season 15% of email modems in Eritrea were taken out by lightening. A year later after installing surge-protection on every piece of metal in the Energy Center computer room, we survived a second lightening strike without damage, though the email computer at a technical school down the street was fried.
Then add to lightening:  randomly fluctuating power supply voltages, no grounding, dust, heat, telphone lines severed by road construction, changing phone number assignments, identical 110 volt and 220 volt power sockets, intermittent power supply, hard disk crashes, and a two to three month turn-around on parts orders; and you have a 'challenging' computer maintenance experience.

ER: the Top Level Domain

For the first couple of years, we were running the whole country's email through Craig's little private domain name: punchdown.org.  Punchdown is the name of Craig's private non-commercial wine operation where he and friends make premium wines out of the basement of his flat in the Haight in San Francisco. The name comes from the process of 'punching down' the floating grape skins into the wine during the early stages of the fermentation process.  And it was sort of cool that a minister  in Eritrea would have an email addresses named after  fermenting grapes.  But names like phys.asmara.punchdown.org were getting a bit long, and it seemed that it was only appropriate for the country to start using its proper domain name.

So we applied to administer the ER top level domain (TLD).

When we did, we found that someone from the 'Eritrea Information Services Agency'  (EISA) was applying at about the same time. And they had applied via FAX.   We were a bit perplexed.  We knew we had the major  functional email system in the country, and that Internet was still a ways away.  But EISA is the official government computerization agency, and it is directly under the office of the President, so some diplomacy was required.

Our main concern was that our users should have more rational email domain names, and that the technical administration went well.  So after a couple of meetings we agreed on a compromise: EISA pre-approved the major domains of concern to the email users (eol.com.er, and gemel.com.er for the commercial users, uoa.edu.er for the University, etc.), our non-profit would do the technical administration, and EISA would have the authority on all of the top and secondary domain names.

Presently ER is not a very active domain name space, contrary to the desires of the occasional domain name profiteer.  Like the Mr "Burg" from Northern Europe who wanted to register Burg.er for purely personal use, of course!

Indigenous Nerds

In both Eritrea and the U.S., nerds are the human glue that keep Internetworking together. I know that there are some 'computer professionals' who will argue otherwise.  But to keep complex systems working based on a petulant and problem-prone technology, one has to have a core of computer-obessessed individuals who are struggling with it day in and day out.  I know there are nicer names, but these people are nerds, and I am one of them.  The problem was that when I arrived in Eritrea there were few if any real computer nerds.  And creating a community of such beings was almost as important to sustainability as economically commercializing the service. Fortunately, the Physics Department of the University of Asmara was a fertile recruiting grounds for the new Eritrean nerd generation.

Even before I arrived with the UUPC-laden laptop, I had already organized a group of about four second-year physics students to process meteorological data for some research projects.  Of these I recruited three to work with me at the Energy Center later.  One of these, Daniel, is the first indigenous Eritrean email nerd.

Daniel is interested in everything technical, and was happy to put in long hours so that he would have the chance to learn.  He helped me with research and email development, and both picked-up on things quickly and experimented with new configurations and programs.  On one of my trips back to Eritrea, I brought him a laptop computer in gratitude for all of the help that he had provided.  Talking to Daniel's mother, she noted that Daniel's used laptop was his little "baby", and that Daniel barely talked to her anymore in the evenings.

Another pioneering nerd is Ezana.  I first started working closely with Ezana when he wanted to do stove efficiency for his senior research. I tried to convince Ezana that pen and paper calculations were enough.  But no, Ezana wanted to write a computer program that simulated where each literal calorie of heat flowed in an Eritrean stove.  Ezana is now one of the main instigators of Internetworking activities at the University of Asmara, in addition to being a pioneering researcher on Eritrean stove efficiency.

To facilitate nerd cultural development I tried to provide an enabling context for technical work and play.  We created spaces at both the University and the Energy Center where the students could come and work/play with the computers at nearly all hours, and during the weekends. And I recruited students to do thesis projects on interesting computer-based projects: remote sensing, meteorological simulation, stove efficiency, and of course computer networking.

Maybe teaching young Eritrean students to obsess on technical work is a type of cultural imperialism. Even so, it resonates with an important part of the modern Eritrean personality.  Eritrea won its independence by obsessing and working far too hard.  During the height of the struggle, they ran factories in caves, at night in Northern desert mountains, supporting themselve with a sense of Spartan sacrifice.  And as a new nation, many Eritreans were attacking the problems of nation-building and development with the same focus and determination. For many Eritrean students, obsessing on technical problems came natural.

These pioneering Eritrean techno-nerds: Daniel, Ezana, Tamrat, Seare, Petros, Robel, Bereket, and many more are now maintaining the current system and developing Internetworking capacity within Eritrea.  As you read this article, they are making web pages, configuring experimental Linux servers, interfacing TCP/IP with existing Novell networks, writing Perl scripts, and training the next generation of neo-nerds.  And they are doing these things WITHOUT real time access to the Internet.

Internet the Eritrean way: the shape of things to come

Eritrea will be one of the last countries in the world to have a live Internet connection.

I used to think that this was a bad thing--but now I am not so sure.

Like email, Internet is 'easy.'  Any Joe Schmoe computer dude can buy a router, a Sun Server and an access server, rent a VSAT line, and start an ISP.   But in Eritrea, this just might not work.  The best case is that one winds up with an expensive and inefficient connection that is used by a few relatively rich foreigners and hardly anyone else.  The worst case is that the business goes bust and one is left with the carcass of a subsidized, isolated system, or just some broken equipment.

Besides just common-sense precautions like back-ups, power conditioning, and surge suppression, an ideal Eritrean Internet will need to be optimized to squeeze the maximum service out of expensive international bandwidth.

This will require many things including: Fine tuning of the communications parameters of the VSAT connection.  Optimizing custom queuing of both service type and source/destination addresses on routers at both sides of the connection; Doing an on-going demand analysis so the right amount of bandwidth is allocated to difference services and institutions; Configuring an efficient transparent proxy service for web browsing;  Mirroring and development of local content to decrease demands on the International connection; And development of simultaneous high-performance/high-cost access, and lower-performance/low-cost  access so that all sections of the Eritrean market can access the Internet according to their needs and means.

Sustainability also demands that Eritrea's future Internet be done in conjunction with local businesses and the loose community of now professional nerds at the University and elsewhere. Training, persistence, and experimentation will all be requisite components of the new system.

In some ways being last is not such a bad thing.  Eritrea is the last country in Africa to win its independence, and the last to join the World Bank and the global economy. Eritrea has used this position to learn from past experience and more carefully and cautiously plan its development.  It now experiences "Asian Tiger" growth rates of 6% to 8% per year in spite of all of its problems and the global market turmoil.

Maybe being last to connect means that Eritrea's Internet will be a bit more Eritrean, rather than just some other international access point for some rich net-addicted expatriots.  And ultimately that would be a good thing for all of us--even if net-addicts like me have to wait a little longer before we can surf the world wide web from the Tigrigna highlands of East Africa.

Contact Info:

     Robert Van Buskirk
     Eritrea Technical Exchange
     110 Clayton Street
     San Francisco, CA 94117