Rishab Aiyer Ghosh of Electric Dreams fame, wrote an interesting article about License Fees and GDP per capital. He expressed the cost of buying/owning Ms Office & Windows with the average income in countries. In Viet Nam it takes 16 month of average income to earn enough to buy this software. In American terms this would mean over 48 thousand dollars for something you can get basically for free if you download OpenOffice and install Linux. No wonder 94% of all software in Viet Nam has been pirated. His conclusion is that Open Source is a big chance for ICT development in the third world and thereby a huge implied thread to Microsoft's kingdom. There's a point to this, but it isn't that straightforward.
GDP per head with the cost of software isn't the most relevant comparison. The proportion of people using computers in the Third World is often much smaller than in the West. The people that actually work with computers are much richer than the people that don't. Of course, a programmer in Bangalore makes much less than one in San Francisco, but it has been argued that their living standards are not that different. There are lots of cars too in the Third World, though they are comparatively very expensive (a 50 thousand dollar Mercedes in the US compares to over 4 million in Viet Nam, and still you see expensive cars over there).
Another thing is of course the piracy. In Viet Nam only 6% of the software is legal, in the US it is 75%, which means that on average a copy of Windows/Office brings Microsoft 420 dollar, while in Viet Nam it only brings 33.5 dollar, which evens out things nicely. Fighting piracy nail and tooth in poor countries could change that, so this might very well the reason why the BSA is aiming at richer countries so far.
The third thing is of course that the marginal cost of Windows/Office for Microsoft is negligible. Given a good working mechanism of market separation, Microsoft could easily charge 20 dollars for Windows in Viet Nam and 200 in the US. Here the Redmonians have two options: either license certain versions of Windows only for use in certain countries or discount the Vietnamese version heavily.
The second option should work; not a lot of Americans will learn Vietnamese to avoid the cost of a copy of Windows, but it is rather restrictive. The first option would work within the EU for example, because Microsoft could block parallel import of cheap windows versions from Viet Nam with reference to their trade mark (there is free movement of trademarked products within the EU, but not into the EU). Books are cheaper in India than in the UK, but are stamped: to be sold in India only.
But what about the point of the article? Does Open Source open a route for Third World countries to build up expertise cheaply so that they can join the world market for software? May be in some cases, but outsourcing is just another market, with market forces at work. If I were a programmer in the Third World I wouldn't necessarily concentrate just on gcc, but would also include Visual Basic and it's other * Studio cousins. In all fairness, Microsoft is not so bad for starting developers. Their tools are usually pretty easy to get going with and they share a host of good documentation.
For governments it might be a different story. Releasing all software developed by the central government under the GPL isn't a bad idea; the people paid for it, so they should have rights to the results too. Also, getting a localized version out for Open Source products is usually not that much work and can bootstrap the local IT industry faster than waiting for Microsoft to come out with something.