No Road, Drive: The ICDL Goes to the Mongolian Countryside

In June 2006, I brought the International Children’s Digital Library (ICDL) to Mongolia by installing a server in the capital that offers, a Mongolian version of the ICDL. This time, in November 2007, I traveled to Mongolia with graduate student Sheri Massey to bring the ICDL to the Mongolian countryside. To understand why we would do such a strange sounding thing, we must first take a quick peek at Mongolian education and children’s books.

Mongolia was a socialist country for about 70 years with financial support from the Soviet Union, including the propping up of the children’s book publishing industry. In 1989, when Mongolia became independent, that financial support disappeared, and the children’s book publishing industry collapsed. Without new books, and given the very rural population of much of Mongolia, very little culture of reading for enjoyment has developed. This hurts general literacy, and the country’s aims to develop and modernize their economy. Those in rural areas had no access to children’s books unless they went to Ulaan Baatar (the capital). This resulted in a growing gap in achievement between rural and urban areas.

Fast forward to 2005 when the World Bank decided to fund a project to tackle some of these issues head on. This new Rural Education And Development (READ) project sponsored a competition to spur the publishing industry to create 200 new children’s books for classroom libraries in grades 1-5. The books were published, and copies were distributed to multiple teachers in every rural school in the country. In addition, a broad set of teacher training and educational testing activities were launched. We were pleased that just about every publisher agreed to allow these books to also be a part of the ICDL. In working together between the World Bank, the Government of Mongolia and ICDL, we found what I think was a very good balance between the needs of publishers and broad access to the books, including free online access in ICDL, while respecting the copyright of the publishers.

Interestingly, the World Bank and the Mongolian Ministry of Education, Culture and Science (MECS) who is administering READ, decided that they would also include a digital component of the project. The reasons are a little complex, but the essential motivations were to increase technological capacity within the country, to investigate whether technology can be used to increase educational motivation and quality, and because while it is not yet true, the economics of digital content delivery is likely to become less expensive than book distribution in the future – especially if the computers are already there. There was also the goal of increasing urban access to these books whose physical copies would be distributed only to rural schools. So the World Bank contacted us at the ICDL, and we became the digital component provider for the READ project.

Getting ready for this trip took a lot of preparation – ably led by Anne Rose, the ICDL programmer and database maven. We had to work with MECS to get the digital books, and to get their metadata in our format in both English and Mongolian, and ingest them into the ICDL. But even more challenging, we had to modify our technology to run entirely on Windows. Before this, our servers ran on Linux. Despite using open source software that is available on Windows, there still was quite a bit of configuration effort to make things actually work. And we also had to simplify the deployment. I would be setting up the first three pilot schools. And in doing so, I would be testing and understanding the process of installing and deploying the ICDL in this isolated and very rural setting. Afterwards, MECS will take on the responsibility of putting ICDL themselves in a number of other rural schools – so it is important that we figure out how to install things easily and keep it running reliably.

Given the problems I had setting up the server on my last trip, we worked hard to think about every contingency. I brought multiple backups of the library installation on many different kinds of media – a laptop, external hard drives, DVDs, and thumb drives. Not knowing what to expect, I brought Windows installation disks, power strips, adapters, and even an American keyboard. Of course, it is the nature of not knowing what to expect that you can’t plan for what you don’t know – and that truism came to haunt me in the days ahead.

Ulaan Baatar

Getting to Mongolia is relatively easy – but it takes a long time. We got there from Maryland in just two flights, stopping in Beijing. After about 27 hours, we were very happy to check in to the Bayangol Hotel, where we stayed for two nights at the beginning and end of the trip in Ulaan Baatar. About half of the country’s population of 2.5 million live in “U.B.” (or pronounced in full with the second word sounding like “butter” but with the first vowel being like ‘o’ in “top”.) There are no international chain hotels, but the Bayangol is the height of living in Mongolia – complete with fast internet connections in each room (at least for PC users, as it required some special configuration which might be difficult on a Mac – which, by the way, I saw none of in the entire country). There was a hot shower, reliable electricity, a range of food during breakfast, and flush toilets, which we did not see much during our rural travels.

The first perception of U.B. we had was the smell of burning wood. A lot of heating comes from burning wood or coal, and the air shows it. There are also a lot of cars. In the downtown area, there is one main road that we went back and forth on a lot while getting anywhere.

But U.B. was just a jumping off point for this trip, not the destination this time. Travelling with Sheri and me were Solongo, a recent university graduate acting as our translator, Altangerel, a teacher trainer who would be learning from us and running the workshops with future teachers after we left, and two drivers – one for each car.

Ulaan Baatar


The road trip started in luxury. MECS has a very nice, new Toyota Land Cruiser with huge 17” wheels, and they rented an equally nice second car – each with very experienced drivers. The first school was about 250 km from U.B., most of which was on a “dirt” road. Actually, dirt would have been nice – but really it was dirt caked with embedded sharp rocks. It wouldn’t have bothered me, but I don’t think we went 15 minutes without seeing a car on the side of the road with someone changing a tire. And given that we often went 5 minutes or more without seeing a car, I calculated that the odds were against us. Our car only had one spare tire, but at least we had a spare car!

It was dry. I could tell by the huge bubbling dust clouds that each car would make in passing. And when a truck came the other way, we were immediately in a white out that, fortunately, usually lasted only 10 or 15 seconds – and to give our driver credit, he did slow down. But passing cars, which we did regularly, required a combination of nerves, experience, and The Force. As we would start to catch up with the car in front of us, all visibility would disappear. It didn’t completely matter since even though we couldn’t actually see the car in front, we could sense its trajectory – if we went to either side, the dust cloud would get a little lighter. So, we just drove into the dust cloud until we saw a bit of the car in front peeking through. Then the driver would apparently use The Force. He would start moving toward the left until the wind gave us just a bit of visibility, and when he could see through and gain confidence, he would speed up and pass. And mind you, this was all happening at about 80 kph on very loud, bumpy roads.


It was on the third trip (I’m afraid) on this same road that our driver started to gain confidence. Actually, he was a very good driver. He didn’t rely on The Force too much, and really anticipated the curves, bumps, and dips in the road, and would always drive to, but not past, the limits of the vehicle. Then, I felt a strange feeling and he pulled to the side. Our front left tire was flat with a 3 cm gash in the side wall and missing the valve stem. He wasn’t upset though, so neither was I. We put on the spare, and went on our way. And after that, he drove much more slowly. Fortunately, it seems that he didn’t want to run out of tire any more than I did. And miraculously, after our next stop at the school for a few hours, the tire was repaired. I have absolutely no idea how he did it, but it was as good as new.

And the countryside during this driving was amazing – huge and vast steppe. Somewhere between rolling hills and mountains, and completely barren. Except for a few planted trees next to a school, I did not see a single thing growing more than a foot tall for several days. And with the crystal clear atmosphere (outside the city), and huge vistas, that is saying a lot. Despite the cold evenings (1 degree F last night), there wasn’t much snow – just spots here and there. The sun was delicious and bright. And the land was spotted with continuous herds of cows, horses, sheep, goats and one group of camels. Often there was a herder nearby, sometimes on foot, and often on horseback. The herders, and it turns out, many countryside men were dressed in the most amazing outfits, which from my outside perspective, was hard not to think of as a costume. Going back, seemingly, hundreds of years, they wore a full body coat with long sleeves that covered their hands (no gloves), a bright yellow or orange sash that wrapped around their waist several times, and boots with curved up toes, and occasional hat to match the era.


But it was on the road to the second school – another 250 km or so – that I came to appreciate the value of roads. The roads were always bumpy and full of rocks, but I would see tracks on the side of the road that looked so smooth and interesting, winding, joining each other, and weaving like the braided rivers I’ve seen in Alaska. I was wishing we could just drive on them – it seemed so quiet and peaceful without the rocks. Or so I thought. We got our chance when we came upon several large truck loads of dirt blocking the road. So, we pulled off and started driving on those “dreamy” paths. For a long time. Apparently the road was being rehabilitated and was full of piles of dirt for about an hour’s worth of driving. So, weave, wind and join we did. And bump. It felt like we were driving bumper cars, without bumping into each other of course, but crazy turning this way and that to find the smoothest track – of which there were none, it turns out. The smoothness had been a deceit, and instead it was up and down, left and right, and only by great fortune did I avoid getting sick.

Then, back on the road, I fell asleep for some time, and woke up with a jolt. A bump had jostled me awake, and now in the dark, I saw that vastness of the steppe all around me. Or rather in front of me in the headlights, but I could sense it everywhere else as we turned and twisted, trying to stay on the path, which was visually just the slightest bit different than nothing. The road had apparently disappeared some time back, and we were left with land between us and the Soum (small rural town) whose lights we could see in the distance. And then water – the other car kindly went first. Maybe 10 meters across, up to the base of the car. At least the ground was hard and frozen enough that we didn’t have to worry about sinking further in mud. But thank goodness for the drivers’ ability to harness The Force. There was absolutely no way to see how deep the water was, and I was very glad not to be the one driving. I have no idea how the drivers navigated. Sheri said they did stop to ask directions a few times while I slept. But I think they just know the landscape. The traditional landmarks, like road signs, structures, or even roads just weren’t there. But what’s a road among friends when in a good car?

It was on the road to the third school that I started to think a little more philosophically about what a road is. Where I come from, there are different kinds of roads – some concrete, some paved, and some dirt – but all are clearly roads. On this trip, we regularly switched from driving on what are unambiguously roads to places that clearly were not roads to something in between. Those things in between are what confused me. There were tracks in the ground where clearly many cars had passed before, but there were often many tracks to choose from – braided in twisting and weaving choice, which the driver had to pick among continuously. They formed naturally, much like animal paths – but animals almost always create just one path. Humans, in the search for the perfection, would wander off the path whenever it wasn’t satisfactory, creating many variations. This, sadly, had the effect over time of deeply scarring the land. And like the tundra in northern Alaska, the land is very fragile. With very little water, it takes many years, probably decades, to recover. But of course, each individual driver’s decision was inconsequential. It is only the accumulation of years of many vehicles that the tracks build up and the land becomes scarred.

In the end, I decided what makes a road a road is human design and engineering. If someone explicitly decided the location of the route and then did some form of construction to support driving, then it was a road. And when the road is good enough so that everyone drives on the same road, the “roadness” is clearer. If on the other hand, there was a mark on the road by happenstance of people driving there without planning or construction, then it was merely a path and not a road. These are the thoughts one has when driving day after day through the Mongolian country side. Hours of silence and pondering about the meaning of roads.


There were times, though, when it was unambiguously clear we were not on a road. In the morning, the drivers were looking for a shortcut through a range of low mountains – apparently to avoid the long road around the mountains. There were a few paths that we followed going up a pass, but the path petered out in a forest. And we tried again, and again. After at least an hour, I thought surely it was time to go the long way around, but they tried a fourth time and magically found our way to the top – up and over, down the other side. And it was easy to see why the “path” was so hard to find. I’ve gone on many hiking trips where this path would have represented a good hike. It was forested, narrow, and full of rocks – and based on every other vehicle I’ve been in, not a place that I would have considered trying to drive through. But we managed, and the drivers didn’t seem to be anywhere near as challenged by the path as I felt.

The third school was far away. We stayed the night at a “hotel”, two ger’s for rent – about $10 / night per ger, holding up to 5 people each. It was a cold night, but we were promised they would come in every 3 hours and stoke the wood stove. However, at 6am, I woke up and it was very, very cold. The fire was completely out, so we went to find the owner. It took a bit of searching, but eventually we got the fire going again and were on our way for a long, long drive. Sadly though, the fire was started with pages torn out of a book – which caused a real conflict in my head given the goals of this project and the fact that I had been listening to Ray Bradbury’s Fahrenheit 451.

This day, we drove about 300 km in 9 hours – about 20 mph average. Looking at a map I did eventually find, it looked like driving in the U.S. The map showed big roads and little roads. But looking at the land out the window, I saw paths and little bits of roads. We just drove through mountains and valleys, over passes, through forests, and even through one canyon. But the biggest road we drove on in that 9 hours was smaller and less maintained than just about any road I’ve driven on anywhere in the U.S., including those I found on many pretty remote outdoors trips. But we had no problems, and arrived at school #3 in time for dinner and, happily, our first shower since leaving U.B.

Finally, after visiting the last school, we had about 600 km to get back. This trip was on slightly better roads for the first half, and an actual paved road for the second half as it was along a straight shot between an Aimag center (one of the 13 provincial centers) and U.B. But still, there was no way to make this trip in one day, so we left the school at about 4:30 pm to make the second day a bit shorter. But if driving during the day was difficult, the drivers really showed their skills at night. The roads still needed quite a bit of decoding, and if I were driving, I think I could have easily ended up off the main track and in some mountain. But we avoided that fate, and except for the fact that both Solongo and I were both pretty carsick, managed just fine. We did have to assist someone else, though, who wasn’t so lucky. We came upon a smallish white truck which had broken through the ice on a “puddle crossing” – a frozen long skinny bit of water that had collected in a large depression. The truck was leaning precariously to the side, up to the driver’s door in shards of thick broken ice and water. We had crossed dozens of these without incident, so it was quite bad luck for these two guys who we found standing around in the dark making absolutely no progress in moving their truck. But we crossed the ice without incident, and with a strong web rope designed for this purpose that one of our drivers carried, we readily pulled them out and we were all on our way. It did emphasize the hazards of driving at night – not only because it increases the risk of something going wrong, but if it does, you could well spend the night by yourself in the bitter cold which would not be at all pleasant.


We drove, and drove, and drove – time passing much more slowly at night when we didn’t feel as well. It was bumpy and twisty with mountains looming around us with absolutely nothing else. We passed one Soum at about 7pm, but the driver clearly had no intention of stopping yet. Then at about 11pm, the driver slowed down and as I opened my eyes in astonishment, I saw a big modern hotel at the edge of another Soum (well, big in comparison to what we had seen recently). The Jiguur Hotel was like nothing I had seen anywhere in all of Mongolia – new and stylish with marble tile and clearly built to higher standards. It turns out that this had been the driver’s destination all along – while we had envisioned spending the night in the car, or perhaps in a one-room shop, all sharing a large bed, as we had seen along the road earlier on the trip. So with this oasis in the steppe, we gladly took three rooms, and settled into bed after the treat of a hot shower and indoor plumbing.

The Requirements of Living

I’m a pretty well-seasoned traveler, so I was not too surprised to find access only to a very limited diet, personal cleaning options and “restrooms”. But the details are worth commenting on.

First off, vegetarians or people allergic to wheat would find it exceedingly difficult to travel in rural Mongolia. Except for “milk tea” and fermented horse mare milk, the only things we were fed for breakfast, lunch and dinner, were variations of meat and wheat with an occasional morsel of onion or potato. Mostly we had meat soup with noodles or dumplings and wonderful fresh bread. Occasionally we had meat pockets. The meat was usually off the bone in small pieces with various amounts of fat and gristle ranging from a lot to more. Sometimes it had the deeper flavor of richer kinds of meat of perhaps organs, but it was impossible to tell. I believe the meat was usually from a cow, but it could have occasionally been from a yak, goat, sheep or horse.

The soups were almost always delicious, spicy, and belly warming – with enough fat to keep you going for a long time out in the cold. It was actually perfect food for this climate and the kind of work most people do here. But since we spent almost all of our time indoors or in a car, its richness was overkill – to say the least. And even I who am pretty good at eating this sort of thing often had a hard time getting past much of the fatty meat. But the broth and noodles were uniformly delicious.

The “milk tea” also takes some getting used to as it is salty – sometimes just a bit, and sometimes so salty as to burn your lips. The tea is a very mild kind of green tea, and so it tastes more like hot breakfast cereal without the grain than tea. I resisted it at first, but came to find it warm and almost comforting by the end of the trip. The fermented horse mare milk, on the other hand, is clearly an acquired taste. It is very thin, and has the sourness of yoghurt – but much stronger. It also apparently has been fermented enough to have a touch of alcohol. I was able to drink most of the bowl I was given, but did not volunteer for more.


The other side of eating was also limited. While there was centralized heat in the schools and larger buildings, there was no indoor plumbing of any kind in any rural building we saw (except for that one shower). Sleeping rooms often had a cleaning place which consisted of a small metal holder for water with a valve that you could wash under with a collecting sink underneath, usually emptying into a bowl behind a curtain. We had brought a fair amount of bottled water with us, so we were able to wash our hands and face this way. But there were no kind of indoor bathrooms, instead relying on variations of outhouses. Most had just slats of wood on the floor with some space between the slats – but one tourist hotel where we stayed in a ger had a seat on top of the floor which was a greatly welcomed.

Most men just went outside “to see the horses” wherever and whenever it was convenient – and we saw school children all run outside behind the school for a break. Even the girls just squatted in the field behind the school. I don’t think anyone went into the outhouses themselves more than necessary.

Personally, the part that bothered me the most was when you needed to use the facilities at night – which meant completely waking yourself up to put all your clothes on, and navigate the circumstances in the dark in the bitter, bitter cold.

The Best of Plans

The power of the internet and communications becomes particularly apparent when you don’t have it. Despite what I thought was tremendous preparation with so many backups, and multiple tests on different systems, it turns out that I was fighting the last war. On my last trip, I was set back by a disk failure. I wasn’t going to let that bother me again. But it turns out that new problems presented themselves.

School 1 – Rashaant, Bulgan Aimag

At the first school, I was eager to set up ICDL for the first time. The computers were all new, having just been installed the previous week, and all looked good. It was a little discomforting with a number of people in the room hovering around me – the drivers playing solitaire, our translator, the teacher trainer, the lab teacher, and some other people. But I set to work, copying the 10GB of page scans to the server, installing the web server software, database, configuring everything, and installing Java on the student computers, etc. I ran into a few small problems, but they were just details that I quickly worked around. Then when I finally went to test it, things went south fast.

The 200 new Mongolian books that are at the heart of this project wouldn’t load. With a quickening heartbeat, I took off my sweater and tried to think. Everything else was working but those books, which because of the publisher’s security requirements, were stored encrypted on disk. They could only be viewed in our Java book readers which decrypted them on the fly. But they worked every time back home. Now, the images just wouldn’t show up. The logs didn’t show anything. It just didn’t work, and I was flummoxed. It is hard to describe the sense of isolation and despair that I started to feel. Back at the lab, this would have been easy to track down – with access to the internet for help, my number one programmer, Anne Rose, who can solve any ICDL problem, and all kinds of technical tools that help me diagnose things. But here, all I had was my laptop and a system that didn’t work. So, what was different than our setup back at the lab? Of course we weren’t on the internet, but we had tested that – or so I thought. It turns out that because of our rush to get things ready for this trip, we tested without internet only *before* we encrypted the books which was done at the very end. Sure enough, I tried the ICDL on my laptop, and the Mongolian books didn’t come up there either – even though they had back home. Clearly, the internet was needed for something, but what? Digging deeper, I found the ICDL book reader log files that I had missed earlier, and sure enough, they reported a complaint that the book readers couldn’t access our main server at when decrypting the books.

School 1

But that was never supposed to happen – we wrote the code so that server names were configurable, and of course, we reconfigured them for this deployment to use the local server that I was installing. But we had missed one, and our incomplete testing had missed it. The only way to display those books was to access a 35 byte key, and it was in Maryland. Alternating between panic and amusement at how incredibly and outrageously stupid I was to have gotten myself into this predicament, I started to think through the ways I might be able to get those measly 35 bytes. First I checked my laptop, hoping that somehow I had stored it there somewhere – but no dice. Then we tried the satellite phone – I would call Anne Rose. It was Sunday evening at this point, but early Sunday morning back in Maryland. But we couldn’t get it to dial internationally (although later we figured it out). I confirmed that there was no internet access or land lines of any kind. And while amazingly, this area did have cell phone service, it often went down in the evenings, and in fact, was not working at this point.

I did have the source code though, so I thought maybe I could find some way to avoid this problem – but no, we had designed the security properly, and it was just plain impossible to access those images without the secret key. As I was beginning to feel pretty confident that my only solution was to drive back to U.B., the cell phone service started working again. So, we called Khishi (the project manager) in U.B., she called Anne Rose in Maryland who emailed her the key. But it turns out that the key was in a binary format – not a simple text file, and there was just no way to transmit the 35 bytes by voice from Khishi to me. Perhaps if I could have talked to Anne Rose directly, we could have done something a bit trickier to get it to me, but we couldn’t call internationally, and it was getting late. And even with the key, the software had to be modified to remove the hard-coded dependency on our Maryland server. I could do that, but it turns out that even that was difficult because Java code has to be “signed” to run within a web browser, and I hadn’t brought those signing certificates with me either – who would have thought I would be modifying the source code to the ICDL on the road? I did figure out how to create a “self-cert” signature, but without the secret key to decrypt the images, it didn’t matter anyway. And even that relatively simple task took me a long time to figure out because I had no Java documentation, so it was trial and error with command line tools. At 10pm, I decided I had to go back to U.B. Sheri was running the teacher training workshop at 9am the next morning, so I left her and everyone else there – and one of the drivers and I took the rock-encrusted dirt road back. I did manage to sleep some in the car, which was good because I needed energy for when I arrived.

School 1

Getting back to the Bayangol Hotel at 1am that I had left that morning, I told the driver to come back at 10am, and got to work. The fast wired internet connection in my room was like a fresh water to me. Anne had gone to the university on Sunday to help me. She hadn’t told me that to do so, she was missing her good friend’s baby shower for which she had stayed up many late nights crocheting a baby blanket. At the end, when I found that out, my respect for Anne and her devotion to helping children in rural Mongolia increased even further. And I continued to battle my own feelings of stupidity and distress that was now hurting not only me, but those around me. But anger and frustration only gets in the way, so I did my best to focus on the problem at hand, and half-laughing at the absurdness of it all, got to work.

During my drive, Anne had modified the code, and emailed me the updated signed java code and the secret key. I could have just taken them, turned around and driven back – but we stayed up for the next several hours, using Google Chat to make sure that we had not missed anything else. We tested multiple times with no internet connection, setting the PC date forward to make sure the code would continue to run after the code signing certificate expired. We did load testing on the PC server to try to duplicate some flakiness I had observed. But eventually we convinced ourselves that things really were working this time. So I went to bed for a few hours, and then at 10am, met the driver, and drove one more time on that rock-dirt road.

I arrived about 1pm, and the teacher training workshop was going great. Sheri was doing a fantastic job, and along with the translator and teacher trainer who was learning all about the ICDL, was working with a clearly very engaged group of 10 teachers. I copied the new files onto the server, and presto – the 200 new Mongolian books worked. Thank goodness. And just in time for the workshop activity that depended on them.

I relaxed for about two minutes, and then started to look into why our server was occasionally stopping. Everything seemed fine, but after a while, it just wouldn’t serve any ICDL pages. Looking at the logs, I saw that “apache”, our web server had died. I restarted it, and all was well – but after another 15 minutes or so, it died again. We had never seen this back at the lab, so what was going wrong? Sadly, I was not able to figure it out – it seems to be a network issue, and not something wrong with our software, so I had to leave the teacher with instructions for restarting it, and hopefully I’ll be able to replicate the problem elsewhere and come up with a solution. But I couldn’t do any more work on the server because of, believe it or not, viruses…

In my entire life of computing – which is now about 30 years, I have never once, not one single time, had a computer I was working on be infected by a virus. Naturally, I am extremely vigilant, but still, I’ve always considered myself lucky in this regard – until this moment. When installing Java on the student machines the day before, I had noticed that two of the machines were acting very strangely, and I couldn’t install Java – but I was so busy with everything else, I just ignored that and went on. Then, as I was looking at the server logs, the server started to act in the same strange way. The key things that were real giveaways were that all administrative tools were disabled. I couldn’t use the Task Manager, run a new program on the command line, or even configure Folder Options in order to see hidden files. At first, I was very confused, and then I had a flash of understanding. A virus was on the server, and the web browser crashed whenever I opened it. And a flash it was – my temperature shot up 10 degrees, my sweater came off, and I again tried to balance panic, amusement, and focus.

What had happened? How could a virus have materialized from nowhere? These machines aren’t even on the internet. And then I figured it out. I was using my thumb drive to copy the Java installation files from one student machine to the next. At the second to last computer, I noticed a few strange new files on my thumb drive, but at the time, didn’t think anything of it. On the server, I noticed “New Folder” on the thumb drive, which hadn’t been there before. I double-clicked on it to make sure it really was empty before deleting it, but double-clicking did nothing. And I couldn’t delete it. And it was immediately after that things went screwy. Of course, it wasn’t an empty folder – it was a virus application, disguised as an empty folder, and I had just infected the server. The virus on one of the student machines had no doubt copied itself to my thumb drive when I had used it, counting on my actions to try and clean up the “New Folder” that had appeared on it, which I unfortunately did.

Again, I was helpless – no antivirus software, no internet, no tools of any kind to diagnose or fix the problem. There was absolutely nothing I could do. Amazingly, the ICDL server continued to run well though, so in frustration, I closed up my laptop. I left this new classroom hobbled, 3 machines infected with a flaky server, but at least the ICDL was running, and the teachers were excited. The IT company that had installed the computers had a service contract, and they would come back soon to clean things up – and fix the flaky network which hopefully is responsible for the ICDL flakiness.

School 2 – Battsengel, Arkhangai Aimag

When I got to the second school, I was feeling a lot more confident because this time, I knew that the ICDL was under control, and I knew to look out for viruses, and how they might limit me. But technology being what it is, I again was thrown a curve ball.

I methodically and carefully examined all the computers when I first arrived, and sure enough, found the same virus on two of the computers – including the server, which I knew was going to give me trouble. They actually did have some virus checking software. Although it was out of date, we ran it on all the machines, and a few infected files were found and removed – but the administrative tools were still disabled. I didn’t need them, though, to install the ICDL, and without much choice, I carried on.

The ICDL installed smoothly, and with great confidence, I opened a browser to test it for the first time, and was heartbroken to see a completely new broken behavior. A number of the images wouldn’t load completely, and the Java book readers wouldn’t launch. It was so strange because the website was working perfectly overall except for these two specific things. When a full scan of a book page was displayed, it would often display just the first half or two-thirds of the image, and the bottom portion would be sort of gray. But it was very inconsistent, happening with some images, but not others. And the way that the book reader failed is that it complained that the very file Anne Rose had sent me previously was corrupted. Naturally, I was sure that the file really was corrupted, so I recopied it from a different source, but it still failed. Then I copied the “corrupted” file back to my laptop (which thankfully still worked), and compared it with a working version, and they were identical, so I knew it wasn’t really corrupted.

I did a number of experiments and convinced myself that the Apache web server was occasionally truncating files as it served them. Now, Apache is rock solid, and is used all around the world to serve a huge number of the world’s web pages – so I knew it just couldn’t be Apache being flaky. Yet, that is exactly what appeared to be going on. This Soum, it turns out, did have very slow dial-up internet access. So, I was thankfully able to painfully use to try and figure this out. But I could not find a single report anywhere of Apache truncating files that it served. Once I convinced myself that every single thing was right and there was no good explanation for why things were screwy, the only thing left to explain it was the virus that I thought was on that computer.

With no ability to remove the virus, the only thing left was to really start over. I had told the computer teacher earlier in the day that it might come down to reinstalling Windows. He clearly did not want me to do that, so I had spent a huge amount of time getting to this point where there really was no alternative. Finally, he agreed, gave me his windows installation disk, and I started. Now this is a process that I have done many, many times in my life – and it is pretty much failsafe. So, imagine my surprise when after formatting the hard disk and being long past the point of no return, the windows installation died with an unexplained I/O error.

This was a case for psychic debugging, the kind of problem solving where the only approach is to close your eyes and think. What could cause this? It looked like it might be a hardware error. But even despite the no name brand kind of computers, I didn’t think that was the problem. And even though the CD had booted fine and seemed to be working properly, it did have a few scratches. So, maybe it was the CD. Thankfully I had brought a Windows CD with me, and I started the installation procedure again, formatting the hard disk again for safe keeping – and with delight, everything went smoothly – until it was time for installing drivers.

Drivers in Windows are the bits of software that make specific kinds of hardware work. Each display, network card, and sound system needs its own drivers. And because there are so many third party hardware vendors, Microsoft does not distribute the drivers for all hardware. You need the drivers from the specific hardware vendor you are using. And because these computers were put together from various components, and not from a major manufacturer like Dell, the only way to get the drivers is from the vendor directly – or from the person that sold you the computer. Fortunately, the IT teacher did have two CDs with drivers for these computers that the IT company gave him when the computers were installed.

However, one of the CDs had dozens of drivers for many different kinds of hardware – clearly generic disks for all the computers that they sell, and the other CD was unreadable. The problem was that I couldn’t figure out what kind of network hardware was in the computer, and thus couldn’t figure out which driver to use. Again at a blocking point, it finally occurred to me that the student computers probably used the same network cards. I was able to figure out which kind of hardware they were using by looking at the driver configuration – and thought I was done. But it turns out that there were no drivers for that vendor on the CD that worked, and the other CD just would not work. But I eventually tried the CD on another computer and that worked, I found the driver I needed, and I was able to complete the setup of Windows.

Then, with a fresh installation of Windows, installing ICDL went very smoothly and to my great delight, it worked perfectly the first time. All the problems with truncated images and corrupted files disappeared, and I was able to leave this school with a perfectly functioning lab and ICDL. My temperature went down, and I felt a huge weight lift off my shoulders.

Looking back on these two days, I realize that the kinds of problems I faced are what I as a computer technologist deal with every day. But the difference was the lack of site-specific knowledge, the internet and resources to make problem solving easier. Furthermore, I felt great internal pressure. The pressure was all my own, but after the huge effort it took to create the ICDL, get to Mongolia, and then to these remote schools, I really, really did not want to leave without having installed ICDL successfully. And then there was the practical matter that we had a schedule that allowed exactly one day per school. And I couldn’t stay any longer in Mongolia because my family was greatly missing me – so it was a sense of acute deadline that drove me to finish.

School 3 – Burentogtokh, Khuvsgul Aimag

The third try is the charm. I knew everything that could go wrong and I was prepared, the eternal optimist in me thought. First I checked all computers for viruses, and there were none. It looked like the computers hadn’t been used since they were installed last week. I carefully installed the server software and completed with no troubles. While the files were copying, I started installed Java on the student computers and looked them over. One of the loose power connectors got jostled and a row of computers went off while installing Java. But no problem, I just uninstalled and reinstalled to be safe. The server was finished, I was on the second to last student computer feeling like I finally did something right when the lights dimmed and all the computers went off. Then on for a minute and then off. With the lights low and twinkling, it was clear that the voltage was low and unstable – which is not good for computers. There was no surge protector or power stabilization system. When I asked if this was common, I was told no, but that they were doing electrical work tonight. Before I had a chance to do anything, two of the computers started to come up with BIOS errors. I sensed things were going to get worse quickly, and I immediately unplugged everything and called it a night with a sinking feeling in my stomach. One more lesson learned – start with a stable power supply.

The next morning, I had an hour before the workshop started, and turning on the computers, I found that two would not boot, and one’s network card would not function. But the remaining seven computers started all right, and ICDL worked perfectly. And then the power went out. I guess it wasn’t just evening electrical work. We unplugged the computers as the power went up and down for an hour. People were pooling outside the lab door and we were told to wait while the school director went to talk with the electricity provider to see if it could be stabilized during the day.

Concluding Success

As more and more people started to congregate outside the door, we realized that an opening ceremony was being prepared. Dozens of children, their parents and many other people had organized their positions. We were given seats of honor, and listened to speeches by the school director and Soum “mayor”. We were given drinks of ceremonial warm milk, cut a ribbon, and then watched an amazing performance by a series of children singing and dancing modern and traditional pieces. Then after 30 minutes or so of greatly uplifting ceremony, all the adults were welcomed into the lab to see the great ICDL!

Now, at this point, three of the computers were broken, the ICDL had not been running for more than about 60 seconds at a time in this lab, and all the computers were off. As about 50 people piled into this tiny lab, we quickly plugged in the computers and powered them on. I stood in a corner and lifted up my laptop (which did have ICDL running), and gave a silent demo to rapt attention. After a couple of minutes, most of the computers had come up, and ICDL ran perfectly. The adults started playing with it – and clearly enjoyed it! I saw people searching, reading, trying the book readers, and experimenting with the different language options. I saw one elderly woman in traditional dress having trouble clicking on things. Without talking, I put my hand on hers and showed her how to use the mouse – showing her the left mouse button, and lifting her finger to move the mouse, and clicking when over a button. And not a minute later, she was navigating the ICDL herself. We were told that several adults asked if they could come back to read books themselves.

And I knew that technological hurdles aside, we had created something important and valuable – and that these adventurous people were going to figure out how to take the best of what we had to offer and put it to good use.

Teacher Workshops

The second point of our visiting each school was to perform a teacher training workshop. The workshop aimed to help the teachers answer three main questions: 1) What are digital libraries?; 2) What is the ICDL?; and 3) How can the ICDL be used to support teaching and learning in schools?

Sheri led the workshops which included 10, 18, and 8 teachers, respectively, in the three schools. Solongo translated, Altangerel paid close attention during all three workshops since he would be leading future ones in the rest of the READ/ICDL schools after we left. I usually spent my time fixing the remaining computer problems.

Teacher Workshops

The workshops lasted about four hours and consisted of a short presentation at the beginning by Sheri introducing digital libraries and the ICDL. But most of the workshop was participatory, with the teachers involved in a number of activities. The main activities which aimed to both educate the teachers about the ICDL and how to teach with it were:

  1. A scavenger hunt where the teacher learned about using the ICDL by using it to answer a number of questions about Mongolian books in the library.
  2. Creating an alternative ending to a book in the ICDL.
  3. Collaboratively creating one or two activities using a book from the ICDL.

Going into these workshops, Sheri and I were both pretty nervous. We just had no idea how the teachers would react to the ICDL. We didn’t know what their computer background was. It was an entirely new concept and experience for them in so many ways. With this hesitation, our expectations were exceeded in just about every way.

The teachers unabashedly loved the ICDL. Of course I am biased, but this really did seem to be true. They were fully engaged, participated fully in every activity – and we had to pry them away from the keyboards each time to move on to the next steps. We got clear feedback that they understood and could easily use the ICDL. And more importantly, they believed that the children would feel similarly, and that searching for and reading books on the computer would be a strong motivator for children, and would nicely complement the other literacy activities in the curriculum.

In retrospect, the fact that these workshops went so well is really quite amazing – especially given how difficult it was for us to communicate, and how different our experiences were with so much room for cultural misunderstanding. I managed to demonstrate the challenges in this regard in a simple manner. Toward the end of the last workshop, one of the organizers came in and gave me a small hard white thing – about the size of a dumpling. I had no idea what it was or what I was supposed to do with it, but it looked somewhat like a cookie, so I tentatively put the end of it in my mouth. But it was hard like a rock and tasted like plaster, so with all eyes on me, I gently put it on the desk as laughter erupted all around me. I looked at it more closely and saw small finger prints in it – as if it had been squeezed by a child. As I realized it must be a memento to remember the children, I felt extremely silly. Then a few minutes later, I saw the translator with one in her hand – and she took a bite out of it, and she asked me how mine was! Now feeling really stupid, and with all eyes on me again, I had to try to eat it again – or more like gnaw on it. It was exceedingly hard and very difficult to break even a small piece off. But when I did, I tried to recognize the flavor which seemed to combine rancid milk and sugar – and it was another of those acquired tastes, at least to me. It turns out to be a specialty made by drying cow’s milk in the sun for several days – sort of like sun-dried tomatoes, but with a very, very different flavor. Then Sheri, not being able to eat it at all, found a plastic bag and made a big point of bringing it home to share with others. At that point, everyone was very impressed – and when it came time to give us gifts later on, we each found ourselves with a huge bag of this. So, we’ll be sure to distribute this “Mongolian Chocolate” to our friends and family for months to come upon our return.

Teacher Workshops

Lessons Learned and Reflections


Getting technology to work well is just so darn hard. It doesn’t seem like it should be so difficult since conceptually, what we are doing is relatively straight forward. But the devil is in the details, and there are so many details. As with any complex project, it is not enough for the idea to be good. It must be executed to completion, with all the details filled in. So, what have I learned in this process? Obviously, the technology must work – but what does that mean?

First off, the infrastructure must support the applications being deployed. The power must be stable and the computers must be virus-free. The local IT teacher must be trained in basic maintenance of computers and networks, including reinstalling and configuring Windows.

The application being deployed must not only be backed up in many ways, but it must be tested, tested, and tested – with the same environment as the deployment target. I knew this in advance, but Anne and I had been working ourselves to the bone. We just didn’t have time to prepare fully. And with the trip date set far in advance, and many other responsibilities, it was just a very difficult problem to avoid. But somehow, next time, we will include weeks of testing in our scheduling.

And the second thing is to come armed with tools of every conceivable time. It was good that I brought what I had, but now I know I need to travel with everything I could ever need to rebuild the entire system, including source code, compilers, build systems, signing certificates, etc. I needed the ability to rebuild operating systems, diagnose and fix viruses, and other various maladies. Again, it is hard to do this for a small operation on a shoestring budget – but to do this kind of operation successfully clearly requires it.

The third thing, fortunately, I knew. Which is to be creative, smart, and maintain a positive attitude, holding back the moments of panic and despair. I am here for a worthy mission, and all I can do is the best I can – and that’s pretty good. So, yes, I am a little disappointed in the problems I encountered, but also proud of what we accomplished.



It is not easy for an adult to travel in the developing world for the first time if they have not experienced it before. Half of what is needed is a positive and non-judgmental attitude. This means you must respect other cultures and what are often very different ways of doing things. Practically speaking, you must have the ability to eat a wide range of foods and the ability to sleep in a wide range of settings including variations in beds, temperature, and who sleeps in the room (or ger) with you. In addition, you must have the ability to use minimal or no facilities for calls of nature, and to go for extended periods of time with minimal opportunity for personal cleaning. It is not that the native people are unclean, but travelling in these environments leaves you limited privacy and know-how that might help you get around the limitations in resources.


We always created the ICDL not only for American children, but with a wide range of users in mind. But of course, not being those users ourselves, it has been hard to know if we were designing it right for the people represented in this trip. Time will tell more fully, but at first take, I am very pleased with the design and features of ICDL, and its learnability and applicability in rural educational settings.

I was pleased, but not surprised to see the level of teachers’ engagement and enthusiasm for trying new ways to teach students. In one day, it was only possible to teach teachers. Despite their enthusiasm, it seemed too much to ask them to involve students without giving them time to process, internalize, and develop their own understanding of the ICDL and how they want to apply it. And so, I am looking forward to following up in the months to come to see how the teachers’ attitudes toward the ICDL change over time, and especially to see how children use it.

Seeing the ICDL in this setting, future possibilities that we have been thinking about now become more important. We have been getting ready for some time to offer translated books, and that will clearly be a fantastic additional resource here. Learning English is an extremely important activity, and when ICDL can support that explicitly alongside of more traditional reading activities, its value will significantly increase.

Finally, it is hard to think about this project without thinking about the incongruity between using advanced and expensive technology in places where much more basic needs are being only marginally met. This is a legitimate question and one that I think is important to be raised whenever technology is used in any application – educational or otherwise, in the developing world or otherwise. Technology is not for every place and every thing. But technology does have amazing potential to support people in doing things that are just not possible in any other way.

So, is it time to deploy ICDL widely throughout the rural developing world? It depends. Many places have computers but no content where the ICDL is needed right now. But for those without computers, it probably isn’t cost effective to deploy computers just for the ICDL. As the computers come to be deployed for other reasons, and electrical and network infrastructure becomes available, applications like the ICDL will be much more cost effective. And is it time to start experimenting and learning about what it would take, and teaching about how to make it work now? And should we continue with this and further experimental deployments of the ICDL now? In my opinion, the answer is an unequivocal yes.

I think the World Bank and Mongolia’s MECS took a very wise and balanced approach with this project – focusing the majority of resources on paper books and support for teachers on good instructional practices. But also pushing forward technological solutions that offer a richer set of possibilities for when the infrastructure and social capacity is in place. And by creating, the books are available more broadly in U.B. and around the world. I am honored to have been able play a part in this exploration, and look forward to continuing in this process in Mongolia and elsewhere.

Go to top