I'm beginning to think that the "code walkthrough" style is just way too involved for many of these tutorials/demos.  Most of them can be reduced to "define scenegraph", maybe set up a few event handlers to manipulate it.  I think I'll go back to the low-level tutorials when I get a chance to work on them again.  There the low-level focus means you get that immediate "and this is what that complex explanation means in real code" feel.

With the scenegraph implementations I find myself saying things like "We create a Teapot and make it shiny"... somewhat redundant when the code says Teapot() and shininess = .9 a few lines later.  Or maybe it's just that I've been doing VRML for so long it just "is" and I can't feel the difficulties for a new user there.