... (or another random country in the developing world) I wouldn't have bought an XO for ever kid in my country this year for a simple reason. The software is too raw. I've always imagined that a successful early prototype of their inexpensive low-power dual-mode display, or at least the underlying technology, was the tipping point for publicly launching the project. As I recall, this previously unheard of technology was mentioned from the beginning, and I doubt they pulled the concept out of thin air or off a wish list, since exactly what they envisioned appeared right on schedule.
They just didn't have the same hook on the software side. Sure, they knew that they could use Linux and other free software so the whole implementation wouldn't be dependent on the good will of an American corporation. But it should be noted none of the key people who launched this project are stalwart free software advocates. Anyhow, the designs and overall scope of ambition of the software to be used in the XO unfolded late in the process, especially compared to the hardware. The innovative security specification (that's the plan, not the implementation) didn't appear until this year.
When I last checked the development snapshot of the XO software, it looked like they'd streamlined and tightened things up to the point where the software they ship later this year will work reasonably well. But a month or two ago, it was still hard to tell if it would work at all, and fundamental aspects of the architecture still seem unsettled, making it difficult for third parties to develop applications for the platform.
If OLPC really gets rolling, I'm not worried about educational applications and/or content appearing. The whole world can contribute their own bits and pieces to that effort. The bottleneck though is implementing an entirely new desktop computing paradigm, which primarily has to be done by a relatively small, cohesive team.
The Sugar development team has done an amazing job up to this point, and I think just about all their decisions have proven to be good ones, but for the XO to be "I'll order a million" ready today, they would have needed as much of a head start in writing the software as they seem to have had on the display.