View Thread > Internet & Society > Low-Cost Computers for the Developing World > about design
This week's question by Ethan Zuckerman:
http://cyber.law.harvard.edu/home/ethan_zuckerman
You are the technology advisor to the president of Berkmania, a large, diverse developing nation located somewhere in the global South. 40% of your 400 million citizens are under 15 years old. As a result, education is an enormous national priority and the subject of much debate in the press and in the streets.
Digital technology has had an increasing impact on your nation - several small, but high-profile companies have been hiring Berkmani citizens for jobs in callcenters, software development houses and other low to mid-level technology jobs. As a result, many young people are enrolling in technical training programs in the hopes of finding jobs... and many more are demanding that the Berkmani public schools offer technical training as well as traditional coursework.
The Minister of Education is enthusiastic about the idea of computerizing the primary and secondary schools, but is understandably concerned about costs. Not only will schools need to train and hire IT educators, they'll need to buy an enormous number of computers. One estimate suggests that, to purchase a computer for each classroom (which the Minister considers the bare minimum!), Berkmania would need to purchase 2 million computers. At $500 apiece for low-end computers imported from the US, the billion-dollar cost would be several times the national education budget and a far greater figure than a donor would ever commit to Berkmania.
Seeing these numbers, the Minister of Education has publicly called for the creation of a sub-$100 computer for use in the schools. Her vision of a portable, programmable device with keyboard, monitor, and full email and Internet access has captured the public's imagination, and newspapers are starting to write stories about how Berkmania will change when "cyberchildren" are learning their lessons from their "e-schoolbooks".
There's just one minor problem - this device doesn't exist.
Two camps are emerging with different strategies on creating the e-schoolbook. One camp, led by patriotic Berkmani IT professionals, is suggesting that the nation sponsor research on low-cost computing and build one or more local companies that could produce e-schoolbooks. They point out, correctly, that PC prices in the US and Europe have hovered near $500 for half a decade and show no sign of ever reaching a sub-$100 price point. They also point out that Berkmania, itself, is a large market, and that the developing world as a whole is an enormous market. Finally, they make a patriotic argument that Berkmania needs to demonstrate it's "digital independence", it's capability to take on IT challenges independently, and lead, rather than following, the global IT community.
The other camp also includes many IT professionals, but they are more skeptical and less overtly patriotic. They point out that, historically, electronics have only become affordable when millions of users in developed nations purchase devices. They are deeply skeptical of the ability of the patriots to create a sub-$100 device. They suggest that Berkmania would be better served by investing, short-term, in recycled computers and long-term in multifunction cellphones now being developed in the US and Europe, which have keyboards and WiFi access to the Internet. While these phones currently cost several hundred dollars, the skeptics point out that GSM cellphones once cost many hundreds of dollars and now cost a couple dozen dollars - as a result, more than 10% of Berkmanians own a cellphone. They suggest the government fund a project to research these American and European-made devices and start building software and curicula for them, anticipating that they will become affordable in the next few years.
The president is willing to invest $10 million in a two-year research project to "solve" the e-schoolbook problem. He's asked you to recommend how to spend the money.
Do you fund the e-patriots to create a uniquely Berkmani sub-$100 computer, or do you fund the e-skeptics, who want to bet Berkmania's technical future on hardware developing in the North?
I would argue that this is not an either or choice. For many reasons, it is important for Berkmania to look closely at existing technologies. Chip design, chip fabbing etc. are hardcore technical skills that are not only expensive to learn, but also to do. It would also be too easy to just ignore the years of work that have gobne into existing platforms such as Intel, and thereby, at least partially, ignore all the work that has gone into developing software for these solutions.
On the other hand, just picking a Northern corporations ready-made tech of the shelf, limits all the great experiences and technical knowledge that might come from working to build locally suitable computers for local conditions.
I would advocate designing a Berkmanian computer based on existing processors, memory modules, and even Board designs, but constructed of individual parts to meet the specific needs of the local population.
Build it locally, but from foreign designed parts.
Hi: This sounds good to me, up to a point. I think the largest issue raised in this scenario might be the dearth of local expertise available in Berkmania. On the software side, there is a huge volunteer support network to guide development of local and regional needs, but hardware and manufacturing are still limited in resources to address such problems. Waiting for parts to become cost-effective will delay such an approach, I think. How can we get the cost of existing parts down to an affordable cost that Berkmania can work with? Tough question.
The numbers are important - Berkmania has 400 million inhabitants - if it has enough IT professionals to fuel a debate, it has them in large numbers, in absolute terms.
While it is true that fab is expensive, VLSI design is not - and commercial fabs encourage "offshoring" of design using their libraries. Spending some of that $10 million on evolving a chip that does most of what the more expensive commercial chips do, but at a lower cost, albeit losing a sliver off the speed or power of the most powerful new computers, makes a lot of sense.
Such chips can be packaged to sit on the same motherboards as obsolete computers from the developed countries, thus enhancing the usability with Open Source OSes and applications.
Both intiatives need encouragement - VLSI design and OSS development to meet Berkmania's needs.
I agree that getting the cost of existing parts down is an issue. I do think, however, that you can go a long way by choosing a low-end system such as a VIA C3 based system, and standardizing it for the project. i.e. if you purchase motherboards, harddrives etc,. in large quantities and make do with slower hardware, and then actually construct the computers locally in Berkmania you're a long way towards affordable prices. I do think scale is the issue here. Buy more units.