"Punt!" When your options have all run out, sometimes the thing to do is just kick the ball as hard as possible and hope that it goes somewhere helpful. While this may seem a nasty comment about the One Laptop Per Child's "buy two, give one" approach, it really is not.
In the early days of the personal computer (1975 - 1980) we did nothing but punt. We put out hardware that only early adopters could love in the hopes that someone would write software to make them useful. Eventually, someone did.
It required a large base of early adopters, but we had that, and it seems as if the XO-1 laptop will have an even bigger base. In order to result in development there must be a matrix of communication, but that's not hard with wikis, etc. (we had none of those back then and had to rely upon club meetings and newsletters).
There also being a significant body of experience with open source software development, it should be much less difficult for people to work out how to get programs developed as a group effort. What will be necessary is some significant input on educational software and courseware - the experimental stuff as well as the established approaches.
Here much work remains to be done - all the software geeks in the world won't necessarily have the critical mass of clues about what to try. Developers will need to bring into their matrices of communications a lot of people who are students and faculty in colleges dealing with education (where an awful lot of "the same old stuff" gets taught but who may have some radical ideas that could now be put to the test), as well as lots of teachers and ex-teachers who hunger to break out of the ways they have to implement "the same old stuff".
With luck, some daring professors will use the "$100 laptop" as the platform for their students to use in developing class projects. The "Give 1 Get 1" or G1G1 program offers the opportunity to break open the issue of courseware development for the XO and implement the ethos of "fast, cheap and out of control". Anyone who can get together $400 can play.
A good proportion of what results will be relatively worthless, except as inspiration to others to do better. They laughed at our early efforts, too. But somewhere in there will emerge the Visi-Calcs - programs and courseware that has never been done before and that turns drudgery into fun.
But please remember - it's not just about writing software. It's about creating courseware and the software that runs it - and this takes the intimate involvement of people who do very different things than software geeks do, in different ways.
While Nicholas Negroponte previously derided pilot projects as "ridiculous", he is now forced into a position to sell into a user environment consisting of nothing but pilot projects. Don't worry, he will take full credit for any desirable results.
That's the way the game is played, especially around MIT. Even if they have to punt.