Flash of inspiration

5 min read

Non-profit making organisation gears up for mass production of low-cost laptop aimed at bringing schoolchildren in the developing world all the benefits of the IT revolution. Jon Excell reports

In 2005, when Nicholas Negroponte — founder of MIT’s Media Lab — announced his intention to bring out a low-cost laptop for the developing world he was greeted by a chorus of corporate derision.

Two years on, the mean-spirited sniping has given way to admiration (or in Intel’s case imitation) and Negroponte’s not-for-profit organisation One Laptop Per Child (OLPC) is gearing up for mass production of a device that promises to pass on the educative fruits of the IT revolution to the schoolchildren of the developing world. Negroponte and his team are now finalising plans to sell the XO — or $100 laptop as it has been dubbed — in large numbers to governments across the developing world.

According to OLPC’s technical director, Mary Lou Jepsen, the XO has arrived at this promising position courtesy of a design that makes it cheap to produce and operate, while retaining the performance characteristics of far more expensive devices.

Running on an open-source Linux operating system, the machine uses flash memory in place of a hard drive. It has a specially-developed low-power display screen that can be read in the sunlight, a 10-hour battery life, and consumes so little power that it can be charged by hand.

What’s more, separate computers are able to connect to each other via a built-in wireless feature dubbed Mesh networking. As well as enabling laptops to talk to each other, this also allows multiple laptops to piggyback on another computer’s internet connection.

The cost of the machine is around $175 (around £90). It is expected to fall to the $100 mark by 2009, and Jepsen believes that this could ultimately fall to $50. One of the keys to getting the cost so low has, she said, been the group’s development of a new type of display screen. ‘The average cost of a display in a laptop is $120 so that’s a real barrier to a $100 laptop.’

Low-cost wi-fi chips, less than five dollars worth of plastic, and a camera that costs just one dollar have also helped force the manufacturing cost down. but perhaps one of the biggest factors was the decision to do away with a hard disk and replace it with flash memory. ‘Hard disks are expensive, power hungry and the leading cause of hardware failure,’ said Jepsen. ‘So we got rid of it and moved to flash. Thanks to the ipod revolution the cost of flash has plummeted 50 per cent every year and it seems to keep going.’

It is notable, she claimed, that all of these cost reductions have been achieved in the development of a high-quality product that is fast attaining design icon status. Indeed, the XO recently went on display at New York’s museum of modern art.

Jepsen, who formerly headed Intel’s display division, explained that she managed to get the screen cost down to about a third of an average one by re-thinking the pixel layout to match the function of the human vision system. This borrows from TV and the way that an mpeg is encoded, where there is three or four times the black and white (luminance) resolution compared to the colour (chrominance). She said that this has enabled the team to produce a display that is both low power and readable outside.

The XO has trialled so successfully in countries such as Nigeria that chip giant Intel is developing a rival product

Although driven by an AMD chip, XO’s cunning architecture means that the CPU is actually a relatively unimportant component of the system, ‘Everybody thinks that in developing the new laptop it’s all about the CPU — but it doesn’t actually matter that much with our design which allows us to turn off most of the motherboard and keep the screen on.’ Jepsen said this approach dramatically reduces the power consumption of the computer. ‘the average consumption of a typical laptop is around 20W and the XO consumes just 1W. If everybody switched to our architecture we would save $10bn in energy costs worldwide.’

She also suggested that the IT industry is perhaps guilty of an obsession with Moore’s law — and that cramming more and more transistors into a computer is not necessarily the best way to improve performance. ‘Do we really need Moore’s law any more?’ she asked. ‘it still takes me the same amount of time to send emails, open a spreadsheet, open word or whatever as it did 10 years ago.’

The blame for this lies with bloated software, she claimed. ‘It’s not just Microsoft or Linux, it’s everybody. What we’ve done is skinnied down the operating system so that it can run fast with low power. It’s well known that if you double the processor speed you double the power consumption and the cost goes up because you’ve got more transistors. We can do a lot with less and it runs faster and uses less power’.

An important factor within this ‘skinnying down’ process has been the way OLPC has worked closely with the open source community — particularly those involved in the development of firmware.

‘All of these people have come together for the project and they all say the same thing — the software guys never get to influence the hardware, and the hardware guys never get to influence the software. And the firmware people are stuck with bad hardware and bad software: it’s the “Wintel” legacy where there has been such dominance by the Windows Intel platform and it’s been hard as an engineer to push out a lot of ideas that many people have had for a long time. We’ve got a chance on this,’ she said.

So far, OLPC has manufactured around 5,000 laptops, and plans to make a further 2,000 before mass production begins in September.

These existing machines have been scrupulously tested in a series of field trials across the developing world. Citing the example of a Brazilian child who learned more in a couple of months with the laptop than in the previous six years of his education Jepsen believes the trials have proved beyond doubt that the benefits of the machine will far outstrip the developing world’s meagre investments in school textbooks.

‘The average cost of these per year, per child, is about $20. so over five years the XO is cheaper and offers so much more: word processing, email, the world wide web, games, drawing and even a music studio.’

The trials have also been vital in validating and tweaking the machine’s design. ‘In Nigeria we have a test school that has slanted desks and the kids didn’t know that the laptops weren’t supposed to fall on the floor. In three months only two screens have broken. We’re putting in better cushioning to fix that but I think it’s pretty good. Try that with your Apple ibook.’

Jepsen said that while she believes the XO has the potential to succeed as a product here in the West, OLPC is unlikely to go down this avenue. ‘For clarity of purpose we became non-profit and just focused on kids in the developing world. If we tried to do both I think that decisions would change. Every decision I make is about scale and maximising the number of units shipped — not making money.’ Instead, the profits made by selling the computers will be just enough to cover the manufacturing costs, with OLPC taking just $1 per machine to cover expenses.

The organisation is currently gearing up for launching in eight countries, but is still waiting to close orders. It’s a nail-biting time and, with the project’s ambitions dented by the news that Intel, a vocal critic in the past, is now developing a rival product, Jepsen admits to feeling frustrated.

While Intel’s move could be viewed as a flattering vindication of OLPC’s aims, she suggested that the company’s actions are more likely the result of her organisation choosing an AMD chip rather than one of Intel’s. ‘It’s frustrating to be doing a global humanitarian effort where everyone associated with the project works for free or on a pretty meagre salary, and then this giant wants to try to kill the effort because we don’t happen to be using their chip.’ Intel has previously denied targeting OLPC.

But Jepsen’s inside knowledge of Intel gives her cause for optimism, ‘I understand how it thinks and just want to find a way to collaborate with it and stop this nonsense to give the kids some opportunity.’ She added that Intel’s behaviour also brings to mind a famous quote by Mahatma Gandhi: ‘first they ignore you, then they laugh at you, then they fight with you, then you win”. ‘We’re at three,’ said Jepsen, ‘and I believe we will get to stage four.’