I just finished watching a video (sadly in Real) of a talk by Seymour Papert and Alan Kay, speaking about math education using computers and programming. The really the neat part is about halfway through where Alan Kay starts doing demos using Squeak and the always-mysterious Morphic (and they seem to have a new site, Squeakland, embodying legendary and only recently truly revealed Etoys).
I used Squeak quite a bit several years ago, but found it extremely difficult to achieve that last 5% of the development process -- great for making toys, but hard to make real tools. And Morphic was really a big part of what pushed me away -- besides the performance issues (which are very significant to this day, despire Moore), Morphic is a deeply cryptic system (classic ravioli code) with little effort expended at the time to make it accessible to a programmer. The rest of Squeak, while not particularly well documented, had a sort of simple beauty that made it accessible, the quintessential example of good object-oriented programming and a class hierarchy that served as better documentation than narrative could have. Morphic isn't like that -- to many small features only become sensible when viewed in the larger framework, and its own object model which you could only see in relief. It was obviously born of a strong vision in the creator's mind, and perhaps it is well communicated in the informal lessons that happen in the Squeak core team if not to the wider public... and perhaps with its maturity you don't have to understand it -- it's concrete, you needn't program it, you can simply manipulate it. But this is obviously not satisfying to a programmer.
Still, I find myself intrigued, and I have a childish desire to make my own race car. Everytime I've tried to go back, I find myself both intrigued and frustrated; maybe it's still worth one more go. Despite the pessimism of the last paragraph, it's an awesome demo, and it leaves me a lot of things to think about. More later...