Reading through the Bitfrost security specification for the OLPC project - very sensible stuff. I am impressed.
Years ago Microsoft had a project to come up with a more usable version of Windows. The result - Microsoft “Bob” - was a disaster. The OLPC project is after something similar in intent, and looks as though the result is going to be usable, and possibly even brilliant.
Unix is an operating system written by programmers for programmers. Windows is an operating system written by programmers for early adopters. Early adopters were generally programmers and other folk with a similar mindset. I think Unix is pretty cool, and am perfectly comfortable using Windows (though less inspired by each successive version) - but I earn money writing software.
If you sit behind an ordinary non-techie trying to use either system, it soon becomes clear that both Windows and Unix are horribly unsuitable for most folk. If you were trying to come up with a usable, secure, reliable design for use by the bulk of the population - what we have at present is insanely bad.
By aiming mainly at children, the OLPC project looks to be solving the harder problem. The foundation built by the OLPC project may prove the base from which secure, usable, and reliable general purpose computers can (finally!) be built for the bulk of the population. Of course, they have the advantage of starting from a clean slate, and have the cumulative experience from observing Windows on millions of desktops, and common use of the Internet. Also the OLPC project can take for granted hardware and software that did not exist when Windows was conceived.
Looks as though - by design - computers based on the OLPC design will need little or no anti-virus software. Anti-virus software on Windows machines is extremely intrusive, and often takes a large whack out of system performance. If you offer the consumer a choice between Windows (almost guaranteed to become unusable in a few years) and a machine that is secure, reliable - and a bit faster … the trade-off is going to eat large chunks out of the Windows market. Backwards compatibility with Windows applications will only count for a steadily diminishing market.