One Laptop Per Child Talks

In June 2006, Michail Bletsas, Chief Connectivity Officer, One Laptop per Child, spoke at the PARC Forum on The $100 laptop: why it can and should be done.

Michail Bletsas' presentation audio and video.

The transcript of Michail Bletsas' speech:

---

Michail: Thank you very much for the invitation. It is a great honor to be speaking here. I would like to start with some history. It will explain a lot of things of how this project came into fruition. OLPC is a nonprofit association that’s been a part of the MIT Media Lab where most of the people who work at OLPC worked for the past several years; eleven in my case.

The Media Lab has a long history of using computers in education and exploring the potential of computers in education. This is a picture from Senegal in 1992; this is one of the early experiments that Seymour Papert did. Here is Seymour with an Apple 2 and a student. In the late 90’s we went through a phase where we looked at deploying computer laboratories, [unintelligible] computer laboratories in the developing world.

This is a picture from Costa Rica where we built these self-contained laboratories in discarded shipping containers. We employed the Department of Architecture to build this very nice canopy on top of them. It was kind of fitting, of course, because Costa Rica is relatively well developed and also is an easy place to connect to the Internet. It has a mountain spine in the middle and there was a packet data wireless network on top of that so it was easy to get ISDN speed wireless connectivity but you could get an Internet to these things relatively easily.

This experiment reinforces our views about shared personal computers; we don’t like them. We don’t think they are very effective. But also had a lot of positive aspects and the community really; the communities where these things were deployed really embraced them and made, we think, very good use of them. About the same time something started happening, WiFi, namely; it wasn‘t called WiFi back then. But the availability of inexpensive wireless networking gear started making an impact.

This is a very nice picture to me because that’s the jeep of an Indian general in the Kashmir territory. That guy on top of the jeep is working with a school at the other side of the border; the Pakistani side of the border. So, this is a symbolic picture because it shows that communications; in the same way that Moore’s law made computing personal in the beginning of the eighties, its making communications more personal right now and we don‘t have to rely on carriers to provide all of the communications capabilities that we use everyday.

This is the experiment that actually culminated all of these experiences in the mind of Nicholas Negroponte. One day he came to us and said, “You know, I’m going to be building a $100 computer, do you want to help?” A good friend of ours, named Bernie Krisher, who used to be Newsweek’s correspondent in Japan, and who is now exclusively working on philanthropy in the Asia Pacific region started this program in Cambodia where, with about $15,000 we can have a school built.

Cambodia is a very special case because the Khmer Rouge pretty much killed all the teachers that were there before they started killing other people. So, its educational system was left completely in disarray, so there is a great need for help in the educational system there. Well, Nicholas actually gave money to Bernie and he built three schools in a region about seven hours away from Phnom Penh in a place where there is no electricity.

So, the first thing that he did is, “Ok, we need to put computers in these schools and we need to put Internet connectivity.” To somebody who hasn’t dealt with these issues before, putting in Internet connectivity in the middle of the jungle in a place where there is no electricity sounds like a hard problem. Turns out that its not. It cost a few hundred dollars per month, you get a bidirectional satellite link, you set a dish on the back there…

Actually, today it costs a little bit less than $300 per month for a 512 Kbits link down and a 256 Kbits link up from a company called IPStar. And, we actually put WiFi links between the two schools so they can sell the connectivity. We got a generator and then we said, “Ok, we need computers but because there is no electricity we’ll have to give laptops to the kids.”

We really wanted to do a one laptop per child kind of deployment and that’s where even someone as wealthy as Nicholas has to throw his hands up because these things are pretty expensive and we are talking about 100 kids so that’s where the [unintelligible] went off and he said, “You know, the problem is really the device and not connectivity.” And that is what started, pretty much, this whole thing a couple of years ago.

So, being at the Media Lab, we went to our corporate constituency; we got three sponsors. As of today, together, to fund this project, we started a nonprofit association called One Laptop Per Child that had mostly to do with the fact that the [unintelligible] told us that we are really not in the business of building computers so you really have to make a different structure, a corporate structure, that will take care of the financial [unintelligible].

And, we started investigating how we can do that. If you look at a rough breakup of the cost of a laptop in 2005, you see that roughly 50% of it has to do with commercial piece of it; distribution, sales, marketing. And if you look at the bill of materials, half of it is the display and half of it is everything else. Now, immediately, having dealt with computers for all of my adult and half of my childhood, it was obvious to me that we can do a lot of things for this 25% by dealing with the obesity that exists in software today.

If you see what’s going on right now , I mean, you have, I think its pretty fair to say that the last really major advance that we have in user interfaces in computers have been in 1984, I‘m sorry, I mean PARC happened with Alto around here and really haven’t moved that far away.

We see that Microsoft is spending a lot of money trying to release Vista who’s main contribution to computing is that it tries to mimic OS10, whose main contribution to computing is that it took something that was running very, very fast back in the beginning of the nineties in Pentium class hardware and turn it into the mush that is [unintelligible] now with nicely sliding windows and transparent thingy and 3D widgets and all of these things that don’t really add too much to our productivity, so, very big opportunity there.

So, we went, we talked to all of the vendors. Obviously, all of them were very interested in contributing. Steve Jobs offered OS10; well, OS10 needs ten times more resources than we had in mind about using. Microsoft offered Windows CE which we turned down for a variety of reasons. One of them being not having derivative uses license for it. They were willing to give us a shared source license, but they were not willing to give us a license for derivative uses. So, if somebody wanted to do something with that software, they had to go back to Microsoft and get a license for that.

So we are left with Linux. Linux works very well in our case because we actually can modify and we can bring it down to the footprint of the machine that we have. So, that’s the actual board of the machine that we use nowadays for developing work. We use a ZX2500; an AMD ZX2500 x86 processor. Again, a lot of marketing fads here because it is not actually the clock speed, this is what AMD tells you that it is equivalent to an Intel processor. Its actually a 400 Mhz processor but they say its equivalent to a PIII running at 500 Mhz, go figure.

We figured out that the minimum RAM that we would be able to use is about 128 MB. No swap, half a gig of flash. Hopefully, Steve Jobs is helping us a lot there with the iPod Nano driving prices down constantly so that might go up. A WiFi chip that is capable of autonomous operation; there is another nine core inside the chip itself and RAM so it can operate independently of the CPU. And, a system controller, a sensor that takes care of power management and initializing the [unintelligible]. So, that’s pretty much it.

USB ports are the only peripheral connectors including two audio jacks. The microphone one actually can be used as an input, as an analog input. It doesn’t have [unintelligible] bias so you can use it as a measurement in input for experiments. The one chip that wasn’t present yet on the current developer board is the [unintelligible] chip, as we call it, the display controller. Please make a note, I will get back to that later in the talk.

For software, we started with Fedora Core 5 for obvious reasons. Red Hat is one of our funding sources and they have plenty of talent to put into this project. Pretty much standard components; state of the art components. The current installation footprint, as of today, is about 250 megabytes. It has gone down. JFS2 is the file system that we use for the flash memory. It actually has built in compression, so, that should install in about 120-130 megabytes of flash; leaving the rest of the 500 for user installed applications and data.
One piece that is missing here is that one of the main programming languages on the machine, including the language that you have to use to customize user interface components, is Python. We made that decision; not [unintelligible], Python. Although, we can load whatever you want; its a standard x86 machine.

That had to do with the 25% of the cost that had to do with the electronics. We can’t do anything unless you do something for the display. We looked at the electronics in the display. We say that we invented it; you say that you invented it here; we can discuss that later [chuckles]. But the truth of the matter is that ink displays are not ready yet; they are not fast enough and they are not [unintelligible] enough. They have some other very compelling properties for laptops but we can’t really use it yet.

And, if you look at the standard TFT displays… This is a slide that shows the cost breakup per component. The generations down here, the 45- 46having to do with the mother glass, the size of the mother glass on top of which you build the displays, and you see a pretty consistent pattern that the two most expensive components are the color filters and the backlight. So, if you want to know the cost of a TFT LCD display you have to deal with these two components.

So, the basic idea, and I’m not going to go into many details; this is not my area of expertise. Mary Lou Jepsen, who is doing this work is giving a talk in a couple of days at the SID meeting in San Francisco, is, you use LEDs for the backlight; this is something that became possible in the past couple of years and there is some secrets on how you turn this into a very uniform light source. And you replace the color filters with a diffraction layer. It turns out that is easier to manufacture a diffraction layer than it is to manufacture color filters. Those are the two basic ideas.

Besides lowering the cost, we are looking at six different display designs. We are going to have a beauty contest pretty soon and finalize which its going to be; the final design. A lot of it also has to do with the comfort level of the manufacturer of the display. Display manufacturers are a very conservative crowd; and every time you tell them to do something different from what they are used to, they get very upset.

We decided very early on that it is important to have a sunlight readable display. And we also decided very early on that its very important for the laptop to be used in an eBook mode; as a textbook display. And, for that matter, black and white, we can have a dual mode display where, in black and white, we can use all the pixels in grayscale mode. And, that’s our base resolution; its 1200x900, 200dpi in sunlight reflective mode. And, then by turning the backlight on and using three pixels as one color pixel, we can have the color mode.

This is less than 800x600; its somewhere between 1.2 times VGA to SVGA depending on which mode will win because you can do either special color or fill sequential color or something in between. I know for a fact that its not going to be completely filled sequential; it doesn’t look very good and stresses your eyes but a hybrid approach might actually work and that’s what we are looking at, currently.

This is actually my laptop; very bright display under normal circumstances; under the sunlight. And this is a Sony Libre, they are using the electronic ink display and you can see the difference; one is perfectly redoubled and the other, you can’t see anything. This is the special color idea which has been the original idea; we might deviate a little from it. And those are the specs of the display. Notice that we can do slightly better than one watt with the backlight on.

More difficult than doing the 100 dollar laptop is doing the 3 watt laptop and we can’t really expect to deploy lots of them in places like Nigeria where ministries in the capital don‘t have electricity for the whole day. We can’t expect to make a laptop that is appropriate for the kids all around the world when half the kids around the world don‘t have electricity at home; that’s 500 million kids. So, you really have to make something that very, very low powered so that it can be human powered or powered from alternative sources.

So, this is [unintelligible] working only mode when the laptop only acts as a router for other laptops; I will get back to this. eBook mode, black and white mode, eBook with a color mode with the backlight on. Here, the cpu starts to do some work; same work with the backlight on; normal workload and this is full cpu load with disk drive powered off the USB port.

So, the way we try to address the power consumption problem is by very aggressive power management. So, we integrated a frame buffer in our timing controller chip, which is a chip that all TFTs have, we call it the display controller, and this allows us to shut completely down the cpu and the graphics card and still keep the display running. Now, in an eBook application that is really crucial. If we had used an ink display, obviously, we’d get that for free but with a normal TFT you have to refresh it. You actually have to refresh it at a certain rate; if you refresh it at the lower rate, you can destroy the display.

In order to do that, of course, we have to create a power management layer, which is not really present in Linux, what is present in Linux is very rudimentary, that has to do suspend and resume in less than 200 milliseconds. That’s one of our main software contribution.

Again, you have to find some way to power these devices. Human power is the obvious solution. These are the various solutions that we’re looking at. This is the most promising one in the sense that it only requires two points to generate power. Some, like me, could do up to 30 watts with this thing and we can generate ten watts for about a good ten minutes.

Now, if you have a laptop that takes three watts in an average use scenario, and you generate 10 watts, you have a ratio of 1:3, 1:4 between cranking and operation. If you are talking about eBook mode, it becomes a lot better; it becomes 10:1. So, its almost good; not good enough, but its feasible right now.

These are off board solutions. In the original design, including the one you have out there in the poster, the crank was on the laptop itself. We had to remove it because the torque that you exert is just too high and you will have to make the laptop weigh heavier than it needs to be. Also, the form factor for the motor is not very good for generating larger amounts of electricity without spinning the motor incredibly fast. It turns our that its much more efficient to have, instead of a long, thin motor, to have a flat and wide one like a pancake.

So, this is the original design that was unveiled in [unintelligible]. Certain, important pieces of the design that haven’t changed are the four modes of operation; laptop mode, tv mode, eBook mode, and game console mode. Another important quality attribute is that the display and the electronics are in the same unit unlike most conventional laptops and that’s why, in laptop mode, we have to have the pedestal in the back so it doesn‘t tip over.

This was important for a variety of reasons. First of all, it eliminates complex wiring going from the keyboard unit or the base unit to the display, which is one of the most common modes of failures for laptops these days. It also eliminates the need for differential expensive driving electronics for the display. We can get away by using standard [unintelligible] to drive the display which lowers our cost substantially. Just because that [unintelligible] is extremely short.

The original design was done under the premise that hinges are very expensive, in general, and fragile. Which, if you had one of the original Titanium Powerbooks, for example, you are very well aware of it. So, we did this very simple, 360 rotation hinge display; hinge design, so the keyboard rotates to go all the way in the back of the display unit.

And, our partner, at that point in time we signed up Quanta, which is the world’s largest computer manufacturer to make this, and Quanta looked at it and came up with a rotating hinge design which wasn’t as expensive and, when put to the mechanical test, was able to survive. So, we abandoned the original design for a more tablet-like design that has the display rotating 180 degrees to do the eBook or game console mode.

So, the design advanced. This is with a hinge with the rotation axis on the bottom, this is with the rotation axis on the top. This is another concept, and this is where we are right now. We have actually a working model of this thing. There is only one and it travels with Nicholas Negroponte usually. Unfortunately he is traveling these days and I couldn’t bring it with me.

Some important pieces; everything is sealed, even the ports are covered and sealed by the antennas; those are the WiFi antennas, which was another big discussion with the designers given that this laptop has to operate in completely different modes than normal laptops that usually assume that there is a WiFi access Point, infrastructure access point available.

And, with that, I will get to my piece which is connectivity. From the beginning, one of the important, or fundamental tenets for this project is that learning is happening by many ways, a small subset of which is teaching. Another subset, that is very important, is communicating with your peers, so we wanted these laptops to have very good communication capabilities; to have these capabilities in places where there is no infrastructure.

What has become increasing obvious in thirty some years, is that if you don’t have an infrastructure, the first kind of infrastructure that you are going to see in that place is usually wireless. There are very, very few examples to the contrary. It just happens to be that this wireless infrastructure is often times build it yourself. It doesn’t require a license.

In the Cambodian school they are installing the access point antenna that is used to hook up the other two schools with each other. Another important attribute, also, is to recognize the fact that bandwidth is a perishable good. When you have a link of a certain capacity and you are not using it, or you are using it for only 20% of the time, you are throwing away a lot of bandwidth.

A big, important piece of my work is to persuade telecommunications companies to make that bandwidth available for this kind of work and I am having relative success; especially with satellite companies that have lots of empty transponders and also with cellular phone companies. MTM in Africa is going to do a pilot with us selling excess or unused capacity at very low cost.

The Internet is a strange place. When the Internet started, there was something called the M2M Principle that governs most applications, most communication over the Internet. This pretty much went away in the recent years and nowadays, all applications use intermediate servers to connect to each other. That has been partly by design, partly by an accident.

Most of our computers at home are living behind, [unintelligible] routers these days so they have private IP addresses so they can’t communicate with each other directly. You have to have an intermediary somewhere out with a public IP address that they can go through. Obviously that situation is a big problem in the developing world.

You deploy 50 laptops and you want the kids to start using instant messaging and they cannot use instant messaging unless they have a connection to the instant messaging application servers somewhere in the United States, usually. That’s an issue, so we would like to break away from that. We need to get the Internet back to its M2M days, or its peer to peer days as its called today.

It also became very obvious that in order to do things without infrastructure we have to use some kind of self-organizing network technology on the WiFi chip on the laptop. The industry has come out recently with the 802.11s standards first draft. We’re following that work. We had some technology ready made at [unintelligible] that we were planning to use. We sort of changed our plans and we’re going to follow the 802.11s protocol as it emerges.

The idea is very simple. You use intermediate nodes as routers for nodes that want to communicate. The architecture, to the upper layers, looks exactly like an Ethernet segment. There are mesh portals that act as bridges between the various logical Ethernet segments. There are ad hoc protocols, routing protocols operating inside those nodes that operate with layer two addresses. So, its kind of confusing in a first look but it is relatively simple conceptually from the application point of view. It just looks like an Ethernet and you don’t have to do anything.

And so, with our partner, Marvel, whose making our wireless chip, we are working on a first implementation of [unintelligible] stack. We’re actually, probably going to release something inside the summer. Again, based on the 802.11s draft the routing algorithm inside is called the hybrid wireless mesh protocol. That’s the standards terminology. For those of you familiar with ad hoc routing networks its essentially ODV with some static, preconfigured routes to the gateways to the portals.

We are doing some small deviation from the standard in the area of autonomous mesh operation. By that, I mean that you can suspend the laptop, the main cpu, and the stack will keep operating on the WiFi chip by itself only taking about 350 milliwatts of power. So, you can leave your laptop to provide connectivity for other laptops in the vicinity.

And, we are adding an energy aware path selection, best path selection through the cloud. It takes into consideration how much battery is in the intermediate nodes so you should always use the nodes that have more battery.

The problem with mesh network implementations, right now the biggest problem is that most people concentrated on the routing protocol and not the link state information fed into the protocol itself so you run into the garbage in garbage out kind of situation. The original, you remember academic AODV implementations used bigger frames to test for nearest neighbor connectivity where bigger frames are relatively short and they are always transmitted at the basic rate, 2 megabits, and their reception could tell you very few things about to probability to receive a bigger data frame package.

So, we have used existing work from the open source community to add radio link information into our 802.11 frames. We already have implemented the frame forwarding engine on the radios themselves. The 802.11s standard pretty much defined what the routing algorithm would be. Thanks to Marvel for providing really good [unintelligible] that are really pushing this forward.

And, we have also built a set of good field debugging tools because debugging wireless mesh protocols can be a very hair pulling experience. Our timeline; on November 17th, we announced the prototype on the World Symposium of the Information Society in Tunisia. December 12th, we signed up the manufacturer. We just had a meeting of the selected [unintelligible] task forces in Cambridge, Massachusetts.

A lot of countries are pushing us to sign contracts now because they have elections coming up. [laughter] And they cannot sign and we are trying to push back and saying that you shouldn’t sign until you see the final prototype. So, there is a lot of tension there. Its kind of a funny tension. Brazil want to sign the contract now and we say to them, “Look we haven’t even selected what the display technology will be yet.” But they have to do that because they have rules that say that they cannot sign contracts six months and three months before an election.

We already have started distributing developer boards and we think that we will be able to start distributing fully assembled prototypes sometime in the fall, early in the fall. And, hopefully we will be on target to do a large number of rollouts by the end of 2007.

And I will close this talk and take some questions with why we are doing this. We are doing this as an education project. We think that the greatest resource a country has is its kids and education is an essential component of the solution to all of the big problems that we have in the world these days. Thank you. [applause]

Yes?

Audience member 1: What’s your volume plans for next year?

Michail: I think the minimum will be, the minimum we want to do with the countries we have selected is about five million units. Now, it is about a million per country and I’m putting Thailand, Egypt, Nigeria, Brazil, and Argentina in there. [unintelligible] and India are big question marks. I think that if you want to have a global mission you have to do something about this country, but, again its an optimistic number from the production point of view.

The world’s total laptop production is 45 million units so, trying to do five from one factory in the first year is kind of optimistic. We’d love to. If you ask Nicholas, who is a lot more optimistic than I am, he says five to ten million units.

Yes?

Audience member 2: What about lifetime? Have you given much thought to the lifetime of these units and what the replacement rate is going to have to be to maintain a growing deployment and, your know, there are lots of service issues here?

Michail: Well, to begin with, we tried to minimize the failure mode of the units so these units have no moving parts. They are low power so the probability the electronics will fail is lower. They are sealed because dust is a big problem in the developing world. The plastic, thickness of the walls, is about 70% more than a normal laptop. The display is mounted in a plastic bumper, a very soft bumper, and we are actually investigating putting bumper around them.

And, we are using nickel hydrate batteries as opposed to lithium ion; not that much because of cost reasons but because they are much easier to charge and they can be made to have a much longer lifecycle. We’re talking about 2000 cycles. They are also very easy to dispose; you don’t have to worry too much about disposing them.

Now, our projected lifecycle is five years and how we came up with those five years is, first of all, we tend to keep our laptops ourselves for about four, believe it or not. But also, $20 is the average yearly cost for textbook distribution in those pilot countries. So, we figured out five years by $20 makes $100 so that pretty much settles it.

Audience member 2: [inaudible]…computers across the world, one of the problems that we had was that, if you didn’t do environmental testing of the units; various humidities and temperatures, I’m sure that these are going to go through even greater extremes in the way that they will be used, you have [unintelligible] failures of the devices that show up. We would have corrosion of aluminum parts and things like that. And, so, that was more of where I was heading in terms of survivability.

Michail: Ok, yes, we are going past those through very, very, rigorous testing. All of the tests are going to be to break it. We are building 500 machines to break them completely; to figure out exactly where the limits are. But, obviously, we will have some failures in there. We designed the machines from the beginning so that, from what we know, we should have a lot less than normal laptops. We have deployed laptops in difficult conditions.

What is very important to understand is that’s one of the main reasons that we insist on our one laptop per child principle. We want to give the laptop to the kid. If we give the laptops to the kids, the kids are going to take care of it. It it’s a loaner machine, you don’t care that much so, a big piece of our support strategy is to use the kids themselves.

Yes?

Audience member 3: You mentioned the textbook distribution cost; whose building the textbook to put on these machines?

Michail: A variety of places. Hopefully, the governments that will buy the laptops that already have textbooks that they are printing; that they own the rights to. But, the situation differs from country to country. The best country that I know of right now is Mexico. It has a huge electronic library of educational materials that they have full royalties, full rights to and that they are able to distribute. I’m pretty sure that the Argentinean people will be looking at some of that material.

But it is the main responsibility of the government; if they are going to pay that much money to buy these laptops, they should pay so much money to digitize textbooks to distribute with them. Let’s take somebody else and then I’ll get back to you.

Yes?

Audience member 4: I like seeing your pictures from Senegal. I worked for Seymour [[unintelligible]] in [[unintelligible]] lab. One of the things, you mentioned about the microphone, one of the things that we found kids would do is use the, not necessarily the elementary school kids, but they would use the two game paddles as inputs, so I was very interested to hear that and wanted to encourage you to do two channel input, not one channel input, if you can.

Michail: It is two channel input.

Audience member 4: Thank you.

Michail: We are also doing a range of peripherals, not ourselves, but we are working with a lot of other people, one of the most interesting ones is a $50 vital signs monitor so that the EKG, pulse, and blood [[unintelligible]].

Audience member 4: Its just not usual that laptops have stereo input…

Michail: It is stereo and also we took special care so that it doesn’t put DC out that most microphone inputs put in order to power electric microphones that would totally destroy any measurement potential that this thing had.

Yes?

Audience member 5: [inaudible] …external storage… [inaudible] …use it?

Michail: Not really, its going to be a perfectly fine device for most purposes. Now if you want to recompile the kernel, for example, you’d better have an external drive. If you want to store video, you’d better have an external drive. But its not necessary; the machine will be perfectly usable without an external drive. Also, DFS2 has built in compression so for most practical purposes it is 2:1 so it is perfectly fast.

Somebody from up there.

Audience member 6: Are these going to be available for sale in the US or do we have to move to Brazil to buy one?

Michail: In theory, these specific ones shouldn’t be available for sale anywhere but, what we are trying to do is to persuade some manufacturer to come out with a commercial version. Now, there are a few guys around, in the Internet right now that are actually collecting $300 pledges. So, if you buy one for $300, we are going to give you one and send one to some kid in the developing world. And we are looking at all ideas on how to do that.

But, we have to be very careful. We don’t want to create a grey market, so, whatever we sell has to be distinctly different from what we distribute to the kids. If its not, then you open a much bigger potential for a grey market. If somebody can buy that machine for $300 here and a kid, that is very poor, has that machine somewhere in the developing world, you are immediately creating the potential for somebody stealing the machine to sell it for $300 or $250.

Yes?

Audience member 7: A lot of these developing countries are using their own native language card in [[unintelligible]] education now with [[unintelligible]] or US keyboard coding, they won’t be able to take…

Michail: We are going to have multiple layouts. To change that layout, the designers [[unintelligible]] is trivial with this keyboard; just change the membrane on top. But, we are going to support everything that Linux supports these days, and Linux is pretty well localized for most scripts. Obviously, certain Indian scripts present a challenge.

Thailand, which is one of our deployment countries, presents an even bigger challenge because they have huge words. And if you try to translate interfaces written for English, you end up with labels that are like, that big and go over. But, internationalization, these days, is pretty advanced these places and we are going to have localized versions of the machine in every country. That’s the nice thing about that if you are doing millions of those you can afford to do that.

Yes?

Audience member 8: Is there any way that putting a solar panel on would help your power issues?

Michail: Yes, that’s very expensive. it’s the cost right now. We are looking actually, yesterday I was discussing that we want to do solar powered routers, infrastructure routers, and even that is kind of, I mean you have the router that you can make for $60 and you need a $200 solar panel and [[unintelligible]] battery combination to power it so its… Hopefully, it will happen, but right now, unfortunately, human power seems to be the only source that we can really rely on.

Yes?

Audience member 9: I missed what the educational value of this laptop is supposed to be. If you are going to take five years of budget of textbooks, is this intended then to simply be a way of distributing textbooks electronically or is it for learning about computers?

Michail: If you ant to put it like this, then yes. I mean, that’s an easy way for people to realize it. Its also a tool for expression. All of the creative people in the Western world have decided that the laptop is their tool of choice and I don’t understand why it is so hard for some people to realize that this has to happen with kids because if you want to enhance creativity, the laptop is a tool for creativity, it is a tool for innovation.

If, out of the one million kids, ten of them tend to be top class programmers, that will be a gain for the society that they are in. It is a tool for accessing information, it is a tool for having a lab; a physics lab. I went through schooling and I had three physics experiments through all of my years. If I’d had a laptop and some basic physics software in there, or some chemistry software, I could have become a much better student in those subjects. So, to us, I’m sorry if I’m becoming a little bit emotional, but to us its obvious.

And, you can find many ways to do it. Now, we are very good at building computers and thanks to Seymour and his group, we know what are the results and how you can use them effectively. If you go Maine, for example, which is the first state in the US that has a one laptop per child program for sixth and seventh grade, without going into test scores, and test scores is the wrong thing, to measure from the laptop deployments, you see that the writing skills of the kids have improved tremendously after the introduction of the laptops because they have a lot more incentives to write now, even instant messaging each other.

In Cambodia, the kids have a big, big incentive to learn English to communicate with other kids. In Maine, again, the dropout rates have fallen dramatically and both teachers and students report that they are much happier in classes where they work in teams with the laptops. So, to us, its extremely clear what the value is. And, because the laptop, at the end of the day, has to be an innovation machine, the most important use is, by definition, the ones we don’t know about; and hopefully, we will have some of those.

Yes?

Audience member 10: Most educational studies show that there’s limited advantage of technology, if at all, sometimes actually negative. Now, your not talking about having a laptop on top of textbook, which is what happens here in the US generally, if at all, you are talking replacing, and I would be, again, like the previous speaker said, very, very cautious.

Michail: I’m not talking about replacing the textbooks, for example, but having the laptop solves the one textbook problem, for example. That’s a big problem is a lot of educational systems around the world; the single textbook problem. But, the laptop has tons of other uses and, again, a lot of the people involved in this project are doing it because of personal experiences. A computer was the only thing in my life that when I asked for it, I didn’t have to justify it, specifically I didn’t have to fight. And I really appreciated my father a long time afterwards for that, I mean, it changed my life.

Audience member 11: Yeah, I have a question about is Fedora Core 5, or Fedora Foundation, or Red Hat going to support the IP infrastructure in terms of software updates because I can see a big security problem here; you’ve got like a million laptops distributed all over third world countries, there’s no IP infrastructure there, I can, because I’m a sys admin, see this major factor for viruses and rude kids and stuff like that so have you provided…

Michail: We are going to provide the security patches. The standard, I mean the core of the operating system is not different than other things so we don’t have to replicate too much. What we have to replicate is some kind of infrastructure for the distribution of those patches. We already have a $100 server appliance that we will be able to put in all the schools that have electricity.

Audience member 11: So you have some kind of software update server?

Michail: Yes. We are going to have some software update infrastructure; you can’t operate without that. But, also, please have in mind that our laptop is a lot simpler than the number of lines of code, so that, by definition, means that we will have fewer security issues. I can only take one more question, so, I’ll take the gentleman here.

Audience member 12: I’ve been trying to find anyplace where the research from which this [[unintelligible]] can be located, aside from some anecdotes, there seems to be nothing available online; the research from which this develops…

Michail: You mean the educational benefits of the laptop?

Audience member 12: The experiences and the conclusions that result from that.

Michail: There are plenty of sites on the Internet if you want to see about the benefits of laptops for education, for example, if you go to [[unintelligible]] Rockman’s website, somebody here local in San Francisco, you can find a lot of papers and a lot of collections in there. As far as the research, as I said, this is a lot of personal experience, this is a lot of experience from the Media Lab, this is a lot of work from Seymour Poppet and his group over the course of 25 years now. So, this is a lot of work of Alan Key

Audience member 12: But nowhere on the OLPC website is this research even referenced together.

Michail: We are pretty busy building the laptop right now, [laughter] which is, other people have done this work. I am not an expert in the use of educational technologies so, my job is to make communications work for the laptop, May Lou’s job is to make the display work, Mark Foster’s job is to make the system architecture work, Quantas job is to manufacture the laptop. We are twelve people working on that thing. Unfortunately, we have to outsource the website to somebody else. We do believe in this; we are not forcing anybody else to believe in that and, I don’t think we have to justify it.