In the past three decades the personal computer has made an unprecedented impact in improving education in the classroom. What started as a large machine in a back-office to help a school organize things, followed by far smaller machines to do more personal tasks such as word processing and playing the classic MECC edutainment games, has now evolved into a device that is inseparable from education. In the most developed nations, if a child isn't using a computer regularly for schoolwork some time in primary school, they most certainly are in secondary school.

Although this evolution came at what amounts to much expense and many growing pains, few can argue the benefits of computers in education when used correctly. Word processors allow for a far more streamlined ability to write, correct, and revise writing; printing of said writing has created a common platform of legibility replacing (and perhaps harming) the myriad of different handwriting styles. Optical discs have compressed bookshelves of text down to a 12cm piece of plastic that contains the basic knowledge of the human race. And the Internet has altered the face of communication and research forever, connecting the most unlikely of people and giving them access to the rest of the world's knowledge that can't be fit on a disc.

All of these education benefits, as modern education theory goes, have in turn driven great improvements in these developed societies. The modern information society builds a population that has access to the answer for any question at their fingertips, citizens who have the ability to be better informed than at any time in the past, and educational enrichment has driven a new period of invention and understanding. If education is the cornerstone of stability, growth, and prosperity, then the computer is a mighty tool that can help a society better reach these goals.

In spite of the raw power of computers, they cannot overcome all obstacles. There are over 6 billion people on Earth, fewer than 1 out of every 5 have access to an internet connected computer, and the benefits of the information society have been concentrated into the hands of those nations with the resources to undertake the required investments. In such a situation computers are in essence a problem, not a solution; it's people that are the solution.

There have been many charitable efforts over the past 100 years to help with education in underdeveloped nations; few can claim to be on the scale of the modern push to get computers into the hands of more school children for their educational benefit. Governments, schools, and companies have come together to solve this problem. Each has their own reason, but the goal is the same: bring the benefits and knowledge of computers and the internet to those children who do not currently enjoy such access.

Today we'll be taking a look at the One Laptop Per Child (OLPC) program, a program designed to build and distribute low-cost laptops to schoolchildren. Sticking with the focus of AnandTech we'll be looking at things primarily from the technical side, though we'll deal with a bit of the politics too as such is inseparable from a program of this magnitude. As we will see, much is being done to improve access to computers, the Internet, and information; and the children receiving the laptops and eventually their home countries stand to gain a great deal out of the program.

What Is the OLPC Program?

Fundamentally, the One Laptop Per Child program is focused on the design and distribution of rugged, cheap to build/purchase, and cheap to operate laptop computers, to be distributed to schoolchildren. The program was pioneered by the Massachusetts Institute of Technology Media Lab, an MIT lab that focuses on researching how computers and other digital technology are used for communication and expression, and the invention of new technologies to further these needs. At its core, OLPC is an egalitarian-centered effort to bring to other parts of the world the kind of educational benefits that the developed world has enjoyed due to digital technology - or in other words, to close the digital divide.

The OLPC trade organization, responsible for overseeing the program, views the effort as an educational program and not a laptop project. The organization believes in constructivist education theories, which is to say they believe children learn best by making things and expressing ideas. The laptops that will be distributed are the means of helping children discover and extend their learning, rather than the laptops simply being the end goal. The project and the goal are massive, but after three years of heavy preparation, the OLPC organization is ready to see if it can be done and if their educational theories are valid on a wide scale deployment.

In forming the OLPC trade organization, the MIT Media Lab has been joined by over half a dozen companies offering financial and technical support for the project. The current members are AMD, Brightstar, Chi Lin, eBay, Google, Intel, Marvell, NewsCorp, Nortel, Quanta, Red Hat, and SES Astra. It goes without saying that while the companies joining the organization are doing so for altruistic reasons, many also stand to profit from the project in one way or another, even if the OLPC organization itself is non-profit.

The laptop the OLPC program will be distributing, the XO-1, is a moderately powerful system designed for low power usage and survivability in harsh environments. Although its planned low price tag of $100 may be temping to richer purchasers, the OLPC organization has made it clear that they intend to only sell the units en mass to governments for distribution to schoolchildren. However, the OEM responsible for manufacturing the XO-1, Quanta, has stated that they intend to build a variant of it for the mass market; the similarity and prices remains to be seen.

Argentina, Brazil, Cambodia, Costa Rica, Dominican Republic, Egypt, Greece, Libya, Nigeria, Pakistan, Peru, Rwanda, Tunisia, Uruguay, and even the states of Maine and Massachusetts have all publicized their intentions of purchasing XO-1 laptops for their students. As the laptops have not entered production yet - test versions are still being produced with final production beginning within a couple of months - these aren't strong commitments and the countries that actually end up placing their orders will most likely change.

The XO-1 Hardware

Given the conditions that the XO-1 was built to endure, virtually all parts of the XO-1 deviate from traditional laptop designs. Externally, the case needs to be able to withstand harsh environments, including heat and water, requiring that the entire thing be built to standards similar to those of mass-market high-durability laptops (e.g. the Panasonic Toughbook) which strike a balance between usability, cost, and protection. The plastic casing as at least 2mm thick in all covered locations, and the laptop is lightly sealed when packed up for travel. As an added level of durability, the keyboard is completely sealed, preventing the most common destination of spills from getting inside the machine.

Besides the physical requirements of such a laptop, the XO-1 is also built to deal with the logistical problems of where the laptops are intended to go, mainly the power problem. The laptop is already a sub-notebook in size for several reasons, but even most sub-notebooks have a relatively high power requirement, which is a problem when it comes to generating heat, running off of a small battery, and working with an acceptable recharge source. Much of the hardware as a result has been selected for power usage, if not already for physical constraints.

Specifications:

• AMD Geode [email protected], 433mhz
• Chipset: AMD CS5536 Southbridge
• Graphics: Integrated
• Memory: 256MB DDR-333
• Mass storage: 1024MB SLC NAND flash
• Display: 7.5" Dual-mode TFT display, 1200 x 900

At the core of the system is AMD's Geode LX processor, a near system-on-chip design that incorporates an x86 processor core along with Northbridge functions and the video controller. As a distant relative of the Cyrix MediaGX processor, performance is in no way notable with the Geode LX, but with the integration of features and a low clock speed of 433MHz it has a very attractive power profile for an x86 chip. Maximum TDP/power consumption is only 3.1W, with an average power consumption measured in milliwatts.

A hard drive is out of the question due to the high power usage, the high costs in manufacturing, and the lack of durability. Instead the system uses 1GB of flash memory as its primary storage, far smaller than a hard drive but also far more suitable for the design. Optical drives are also not included for the same reasons, so external storage is handled via SD cards and USB devices.

Due to the small amount of primary storage, wireless communication has a high precedence on the device and is the main method for accessing data. To that extent the laptop features a Marvell 802.11b/g controller and uses a pair of antenna (the so-called "ears") for best reception. The WiFi controller like the rest of the hardware is also optimized for power as WiFi is another major power drain. We're not entirely sure how it has been optimized but we believe that this is via a reduced transmission rate.

The WiFi module also serves a secondary purpose on the unit, which is the formation of a full-time wireless mesh network. When the laptop is powered down, the WiFi module can stay operational and route wireless traffic so that other laptops in the area can be linked together as part of one large mesh. By having an active mesh children are not limited in who they can wirelessly communicate with as long as they can see at least one other node of the mesh, and it allows all the laptops in the mesh to share internet access in a wider area if it's available. It's one of the more novel elements of the XO-1 hardware, and the OLPC organization estimates a XO-1 can run for an impressive 40 hours in its router mode.

Because the XO-1 is intended to pull double-duty as a laptop and an eBook reader, it has a very unusual screen that is both backlit and reflective. When reflecting light the screen operates as a 1200x900 6-bit grayscale screen, as the reflective layer sits above the color components. This mode uses a minimal amount of power because there is no need to drive a backlight, ideal for use as an eBook. When the backlight is activated the unit can produce 6bit color (assuming that the majority of the light is coming from the backlight), although the design of the color filter is such that the unit loses some of its effective resolution compared to grayscale mode.

The power system of the XO-1 is also worth mentioning, again an example of the rugged design. The initial designs for the XO-1 floated several ways to recharge the unit assuming a regular electrical grid was not available, including an integrated hand crank. The OLPC organization has since then opted to not build in any such charger; instead the unit has a DC input that can handle highly variable voltages (-40V to 40V) so that the unit can safely be hooked up to any number of power sources, including solar or pedals. While the initial units will be shipping with a traditional (albeit well built) NiMH battery, future units are planned to ship with a higher density exotic LiFePO4 battery, once mass production of those batteries begins.

Unfortunately, the OLPC organization missed its initial price target on the XO-1 by a good amount. The goal was $100, but the actual price looks to be closer to $175. While it's only a $75 difference, that amounts to the laptop coming in at 75% more than was intended. Until costs further come down, the XO-1 will likely remain unaffordable to some of the target nations.

Software

At the end of the day, it's not the hardware that will make or break the XO-1, it's the software that the hardware will be running. Without good software, the device will simply be a cheap but ineffective laptop.

In an unusual twist, it's the software where much of the charity in the OLPC program is found. Every last bit of software is free and open source; no part of the cost of developing the software is being included in the price tag of the XO-1, and the OLPC organization has spent the last two years recruiting developers out of the open source community to design software for the unit. Most of the volunteer man-hours in the program have been spent creating the software.

At the center of the XO-1 software package is a customized version of Red Hat's Fedora Linux distribution. Besides fulfilling the open requirement, Linux is fairly simple to scale down to the tight storage, memory, and processing restrictions imposed by the hardware. Sitting on top of Linux is the custom Sugar GUI, a non-traditional uni-tasking GUI that is intentionally extremely simple, a fitting design for the target audience.

Moving to the applications that will come with the XO-1, we have to admit we're not sure what to think at this point, other than that we're worried about what we're seeing so far. The OLPC organization has floated dozens of ideas for software to include, but the current builds which we've used on the QEMU emulator have a fraction of these programs. With only a scant couple of months left before final production, we hope that there will be more software than what we're seeing today.

In keeping with the focus on education while minding the limited storage space on the XO-1, the centerpiece application of the device is a web browser for accessing internet educational resources. A stripped down version of the full Gecko rendering engine used in Firefox runs the browser, offering high compatibility with modern web pages; notably Flash support is not included because it does not meet the open requirements. As per the current plans the browser will also be the main vehicle for email, handled via webmail. As far as browsers go, they don't come much simpler than this, with only an address bar and buttons to go forwards and backwards.

For offline data manipulation, the XO-1 comes with separate programs for reading and writing documents. Writing is handled by the competent and compact Abiword word processor, one of the many programs on the device pared down and ported over from larger Unix distributions. The OLPC organization has said they're not concerned with teaching office productivity with the unit since it's a learning machine, so Abiword is not joined with any sort of spreadsheet or presentation software. For reading documents (mainly PDFs) an unnamed document reader built from the evince document reader is used.

The second pillar of software, designed to reinforce the "learning to learn" concept, is the suite of edutainment software included with the unit. In what is a very obvious reflection of the interests of the volunteers creating the software, the XO-1 includes several programming environments/languages, such as JavaScript support through the browser, Python support through the operating system, and variations on the LOGO and Smalltalk programming languages. Smalltalk is represented by the eToys application, and LOGO by the Turtle Art application, a variation of the traditional turtle project for LOGO.

Rounding out the edutainment category are the less "edu" and more "tainment" applications. A clone of Tetris is included, as is a painting application. A Connect Four clone is also included, this one utilizing the WiFi abilities of the device. Last but not least is the interesting TamTam application, a music sequencer we're not sure what to compare it to.

The final batch of software for the third pillar is the communication software. As the XO-1 features a camera and microphone, a couple of applications are included to utilize these. This includes recording applications, and a media player to play back recorded media. An instant messaging client is also included; we can't test it on our simulator but we're assuming that it's a local-only client utilizing Jabber.

As we stated earlier, much of the planned software is still missing from our test builds. Key to these is the entire reference library, especially the included version of the stripped down Wikipedia article database. We have heard that there are still technical issues with including Wikipedia due to size constraints, in spite of Wikipedia's own efforts to find ways to better compress the article database. We'd imagine the rest of the reference library, such as the Project Gutenberg and audio/visual libraries, is having similar difficulties with storage requirements.

Given how critical the software and libraries are for allowing use of the laptop away from an internet connection (something we'd imagine is common) it's premature to say much about the software side of the XO-1 at this point. If the software that ends up shipping is as planned we believe the OLPC program could do a lot in reaching its goals, but right now we can only sit and wait.

AMD's Stake

As we've alluded to in the title, the OLPC is about charity, prosperity, and capitalism. The software is the charity, the end results of the program are the prosperity, and the companies backing the OLPC like AMD are the capitalism. AMD has made it clear in press and business reports how it intends to leverage the XO-1 and its own technologies for business purposes, so we'd like to touch on exactly what AMD's stake is in everything.

There are two real extremes in turning a profit in the computer chip industry. One is to sell a small number of high profit chips; this is the server industry, where AMD has not had the best of luck lately (though Opteron did pretty well back in Intel's NetBurst days). The other is to sell a large number of low profit chips, an area that is traditionally represented by AMD, though at times they have been victim to these low profits being a little too low.

While not in any way giving up on the high-end, AMD is also looking to capitalize on expanding their stake in the low-end, looking for new markets in that area instead of taking Intel on directly. The focal point of that desire is the AMD 50x15 program, an initiative to bring internet access to 50% of the world's population by 2015. Participating in the OLPC program is one such way to reach that goal, but the OLPC program will never ship enough laptops on its own to reach that goal, never mind that as a non-profit program AMD will not be making any money on the XO-1.

However, much of the first 50% is composed of developing societies that have rapidly growing economies, improving electricity coverage, and an expanding middle class. In other words, there are a lot of people finally entering financial situations where they have the kind of income to afford luxury goods like computers. Over the past two years, and again reiterated in their Q2'07 financial report and even by Hector Ruiz himself on CNBC, AMD has stated their intentions to claim a large piece of the computer market that is forming in these societies as one of their profitable enterprises.

Getting into these markets requires computers priced similar to the XO-1, which is why AMD believes they have an advantage over Intel in this pursuit. The XO-1 is a roadmap for how to build such a computer today, and AMD intends to follow it after their first effort, the Personal Internet Communicator, failed to take off. Thanks to the XO-1 AMD can already build a computer to meet their immediate goals, and we'd imagine they're looking forward to Quanta's XO-1 variant going on sale.

AMD's story doesn't end there, as reaching their 2015 goal will also require new technology. As far as the $100 price target is concerned, the holy grail is a complete x86 system-on-chip as a SoC minimizes the number of parts that need to be manufactured and the electronics needed to connect them. Their current Geode processor is close to this in design, but is heavily lacking in performance due to its decade-old design. AMD has decided not to try to further upgrade the Geode and instead build a new SoC x86 processor, culminating in Fusion.

As we saw last week in AMD's Analyst Day presentation, AMD announced Bobcat, their ultra-low-power ultra-simple x86 processor intended for mobile devices. When put in a Fusion design with a proper low-power graphics core and Northbridge, a Fusion Bobcat will be capable of power usage similar to the current Geode, but with far better performance in all areas. This is critical for entering the mass market in developing nations, as users will want to use more mainstream software than the Geode can handle.

Keep in mind, none of this is to vilify AMD. Capitalism is a very powerful force to drive and spread progress. AMD's 50x15 goal is extremely lofty and not everything is in their hands, but from what we've seen with their technology they're going to be able to create the chips they need to reach their goal. This ultimately results in many students and children benefiting today from the XO-1, many more people tomorrow from Fusion Bobcat, and then yet again more in the future when Fusion Bobcat finds its way into the successor to the XO-1. It's only a question of whether or not AMD can make the kinds of profit they're shooting for from this project that remains to be seen.

The Bigger Picture

Given the possibly lucrative market for cheap computers in developing nations, AMD isn't the only company making a push for that market. Similarly, given the mindshare offered by a program such as OLPC, other groups are looking at making laptops for distribution.

At least initially the biggest competition for AMD and the OLPC program has been Intel, who established their own OLPC-lite project called the World Ahead Program shortly after the OLPC program went public. The result is the Classmate PC, a laptop similar to the XO-1 but slightly more traditional in design and more importantly built using Intel chips throughout. Rather than taking a hands-on approach with the whole chain like the OLPC organization is doing, the Classmate PC is simply a proposed laptop that Intel will provide the plans and chips for and let other groups build. At least one OEM began production of the machine earlier this year, with a price tag of $225.

As of July 13th, Intel joined the OLPC organization, putting the future of the Classmate PC in jeopardy. Intel had previously dismissed the XO-1 laptop for various reasons, proposing their Classmate PC design as the more appropriate laptop. Given their movement to the OLPC organization, this is a notable if not sudden change in direction for Intel.

To put things in perspective, OLPC head and MIT professor Nicholas Negroponte was accusing Intel of selling the laptop below cost as recently as May of this year. For all of the for-profit parties involved, there's enough of a potential future market in for-profit machines to be fighting over a non-profit machine today. With Intel having joined the OLPC organization, much of the public fighting for access to the developing world's educational market should subside.

With that said Intel isn't completely out of the picture yet. Intel is working on its own competitor to AMD's Bobcat processor, Silverthorne, which would be similarly capable in low-cost, low-power computers for the developing world. Although we're left to speculate on just what made Intel finally join the OLPC organization, we can't rule out that they're being given a fair shot at getting into the successor to the XO-1 (whenever that becomes available). At the very least they will still have access to the same markets AMD does in terms of selling for-profit computers that are not in competition with the XO-1.

ASUS is the other major competitor to the XO-1 and the OLPC organization. ASUS will be selling its Eee PC, which is similar to the XO-1, later this year for a price suspected to be around $200. It uses Intel chips for its underpinnings, but it's not a Classmate PC. Unlike Intel and the OLPC organization, ASUS is not intending to directly get into education with the Eee PC; rather it's going to be a widely sold laptop for anyone looking for a cheap portable computer.

First Thoughts

As the XO-1 is still in beta production and beta field testing, there are still many questions left to answer; we're looking right now at what can be, not what will be. At the forefront of questions are the intangibles, such as if the educational theory behind the OLPC organization is sound and these laptops will make as much of a difference as the organization believes it will. We're optimists and believe from our own experiences that the laptops will help out, but we also don't hold PhD's in education, so time will tell.

Education still remains as one of the greatest factors in creating and ensuring prosperity, and expanding it is one of the best ways to improve a developing country. At the very least the OLPC organization will succeed in making information more available; the deciding factor will be how it's used to improve education.

On a more technical note, there are still some issues we've seen with the XO-1 and its surrounding infrastructure that need to be worked out. The first XO-1's will likely be going to nations with some of the required infrastructure to use them, so it's not power that worries as much as it is internet access, which is especially sporadic in Africa. Full utilization of the XO-1 will require regular internet access, and this isn't something the OLPC organization has strongly focused on addressing. We'd expect many initial sites to be using either satellite or telephone connectivity.

We'd also like to reiterate our concerns about the software we're seeing at this time. On paper the software is great, but it's hard to translate that promise into the real world when the demonstrations we're looking at aren't showcasing the full software suite and there's so little time left. The hardware on the other hand is quite impressive on paper, and the field test results we've seen back that up. The only catch for the hardware will be getting the price of the unit down to the $100 goal from the current $175.

Last but not least, there are also practical issues of abuse with the XO-1 that bear mentioning. It was widely reported last month as offbeat humor that some of the XO-1 units being field tested were being used for browsing pornographic material, to which the OLPC organization has decided to include filtering on the shipping models. This still leaves open other concerns such as spamming and scamming, though the number of primary school students capable of and interested in such feats would naturally be low. Instead the issue will be adults who get their hands on the laptops, something the unique green design is supposed to deter through shaming since it would imply the laptop was stolen from a child.

Shifting away from the OLPC initiative for a moment, this represents the emergence of a new market that all sides believe is ripe for development. For the CPU industry, this may be a new front in the battle between Intel and AMD, vying for the new bottom end of the market along with the traditional top end. Although AMD is in the better initial position, Intel is plenty capable as we've seen in shifting resources toward solving new problems, so they are by no means out of the picture. For Intel this would merely be icing on the cake, given their current control of the processor market; for AMD the future success for the company is at least partially dependent on rapid growth here.

As for the success of the OLPC initiative and the XO-1, with production starting later this year we should have an idea soon of how much of an impact the initiative will really have. There is precedent for great change, and access to technology can be a catalyst. Time will tell if OLPC can make that happen for the developing countries.