if i were to develop software for kids, i'd most likely do it using pygame, even if i'm not a huge python fan. the early jumpstart products were built using an engine built around a proprietary scripting language, so i wouldn't have any major performance concerns. (what would be considered a 'base' machine has certainly moved way beyond what it was seven years ago.)

the real cost of developing this sort of software is in the content (art, sound, music, educational content). done right, the programming should be a rather modest percentage of the total development cost. (and later products should benefit from the groundwork laid by the earliest ones. unless you do something stupid like start them over from scratch.)