Yow! Ther's an interesting idea. But you know, I want to argue against it -- LambdaMOO is such a nice system already, and has a very, very nice language (though, appropriately, much more limited in domain than Python), what I think we maybe need is Python-MOO integration... maybe a MOO-Zope integration. THAT would be truly cool.
I hope you will elaborate on the latter, but I still believe that an *entirely* Python MOO would be an extremely useful tool for teaching programming, precisely because Python is not "limited in domain." Beginning programmers struggle with language syntax and semantics - it is only later, after one or more languages have been mastered, that people begin to abstract away the differences. Learning multiple languages is powerful and important, but needing to deal with multiple languages *right away* is a big problem. Python's strength for CP4E is that it is both easy to learn and general-purpose. Python is powerful enough to be the *implementation* language for a MOO server, as well as the scripting language. I think that a MOO could be carefully designed so that as students advance they can go from simply exploring the world ("take apple") to extending the world ("class apple:") all the way to extending the implementation ("class emailPager:"). Learning a MOO language would teach someone the fundamentals of programming, but it would leave them powerless to apply them in the REAL world. That's fine if you are a highly motivated person who will pick up and learn other programming languages - but "Everyone" will not do that. On the other hand, it's a small step from using Python to manipulate a MOO environment to using it to manipulate other things.
Bruckman's dissertation (http://www.cc.gatech.edu/~asb/thesis) has some amazing material covering adolescent girls programming interactive agents. The CP4E community really needs to be aware of this work.
Indeed. I had seen this before, but couldn't remember the reference in support of my previous post. Dave