Wayan and I and others have wrangled over the last two years over what OLPC programs would cost, and whether countries can afford it.

How much would XO deployment

Hardware Costs

I want to have another try, beginning with the promise of a $75 laptop, the XO-2, with a new screen from Pixel Qi, in 2010, perhaps a year from now.

We also need school servers, something like one per hundred students. These can be extremely low-cost computers, even donated computers discarded by businesses, without monitors. Real sysadmins prefer to work at the command line anyway, and can just ssh into the box, while others can use a laptop as a visual X terminal.

If we budget $200 each, that adds $2 per child, less than the noise in this model. Similarly, each server will need a wireless access point (WAP). These cost as little as $20 new, and are also available used for free in many places.

Other Costs

But we all agree that laptops and servers are not the only expenses in a laptop project. We need electricity, Internet connections, and teacher training at a minimum to make the program work. We need a lot of other things to end poverty and oppression, but we can see a path to all of them once children are routinely educated properly and able to get jobs and also to join together politically.

For a start, that will mean huge increases in both developing country tax bases and articulate voting populations. This in turn will halt the current brain drain in which more than 50% of doctors trained in some African countries, such as Cameroon go to Europe or the US to get jobs. And so on.

How much do these essential program components cost? Or rather, how much more does a country have to spend to get from its current situation, with some electricity, some Internet, and some teacher training, to enough of each to make a laptop project successful nationwide?

Teacher Training

I am going to suggest that the additional cost for teacher training will be so small on a global scale that it won't make any difference to our calculations here. It is necessary to train an initial group of trainers in each country, and to integrate laptop training into the curriculum at teachers' colleges. This is a one-time cost.

After that, training will continue to cost no more than before. Several teacher training programs exist, and it would not cost much to translate the available materials to other languages.

Of course, for countries that do not have enough teachers, where there are costs for recruiting, training, and above all paying more of them, this is not the whole story. I don't have the data to make an estimate on the appropriate levels of spending, so I will have to leave it out for now. If anybody else can provide credible data on all of the countries concerned, we can discuss it. That means how many teachers there are, how many are needed, the current pay scale, teacher training facilities, and other factors.

Electricity & Internet

That leaves electricity and Internet. Solar panels currently cost less than $10 per watt, which is enough to generate perhaps 8-10 watt-hours (wH) of power daily. XOs operate in a range of less than 1 watt (reading books, backlight off) up to about 8W. Children using XOs in school and for homework might use them for five or six hours a day, sometimes for reading at less than a watt, sometime for computing.

Lets say 6 hours at 4W on average, or about 25 wH, so they need three watts of generating capacity, divided between school and home. At some point we should be able to get data from deployments. Perhaps $30 per child for solar panels, installed, with essentially zero operating cost.

This does not take into account existing electric grids, so it may be a bit overstated. Or not. We are not talking about just slapping a panel on the roof and plugging it in. School systems need customized designs according to climate, terrain, other available power, and student needs.

This does not provide for much electricity outside the schools, although presumably any surplus can be shared or sold, perhaps as a student-run business that charges mobile phone and car batteries. At which point we can consider organizing microfinance for local enterprises, where schoolchildren could provide computer and information services to their families.

WiMax Internet for whole countries has been priced at $10 per person (Sprint, for example, proposing to cover about 300 million people in the US with an outlay of $3 billion) for several years, with 90-95% coverage. It will cost more to reach the most isolated farms and such, but let's start with the $10 figure and see how far that gets us.

Children make up about one sixth of the global population, although in some countries they are as many as a third. You can't provide wireless coverage to only a subset of a population, so we have to provision everybody. Let's be conservative here, and say $60/child for Internet, but everybody else gets to use it, too.

Again, we have ignored existing Internet facilities. In the absence of affordable international fiber optic connections in most of Africa and parts of Asia, they haven't amounted to much compared with demand. Now that enough fiber optic and satellite capacity is in the pipeline for the first time, this should start to change soon, and accelerate.

$165 for One Laptop Per Child is cheap

Per Child Costs

What do we have so far? Cost per child is something like

• Laptops @ $75
• Electricity @ $30
• Internet @ $60

for a total of $165 per child, or $165 billion overall, as a rough estimate. Considerably less than a war, or a major financial bailout.

Now I hope you don't believe in this number. I made a lot of assumptions there. I don't know how many costs I may have underestimated, and how many I may have overestimated. But it's a number we can work with. It's a number that has foreseeable consequences for global development. And any other number we might find has similar consequences.

It is worth noting that this is not $165 billion per year, or anything like that. It is $170 billion installed and done with, except for the working of Moore's Law. The electrical power and Internet towers and antennae have a lifetime of several decades, but the computers and microwave electronics will be obsolete in three or four years, no matter how long they can keep operating.

If we replace them every four years, issuing new laptops in, say, first, fifth, and ninth grades, at the same price, we are down to less than $20 per student per year, or $20 billion overall. Considerably less than the cost of textbooks, which some of us propose to replace with Free digital learning materials. We could say that the net cost of laptops will be zero, or even negative, but I will let that pass for now.

The sum of $20 billion a year is still real money, but compare it with some other things that we are quite willing to spend money on. The oil market worldwide is $5 billion per day or more, depending on current prices.

The US alone spends more than $10 billion annually on school buildings, and more than $50 billion each on electricity and communications, including telephone and Internet. US foreign aid is more than $25 billion annually (unfortunately, most of it military).

Barack Obama campaigned in part on a pledge to double foreign aid, but that was before the financial meltdown. The added $25 billion or so, if we could do it, would be enough, in principle, to fund the entire global rollout of educational laptops over four years, along a good chunk of village electricity and Internet.

So if we are successful in ending the current major recession, perhaps in Obama's second term we can actually do all this. Or maybe sooner, if we wanted to put together a genuine global stimulus package.