So I've learned a new term from ya'll: scaffolding. Of course I knew it from the namespace of construction, but here it means a framework or prewritten code or auxiliary aids such as diagrams. Anyone want to elaborate? I see links between "scaffolding" and the concept of "immersion", already well established in the language learning community -- human languages that is. The ideal is to immerse oneself in a culture or community that uses the language expertly, natively, in its full-blown form, while at the same time struggling with exercises, tutorials, working to get up to speed. This is how we learn our first language, usually. Sure, we have picture books and a huge amount of literature, gradated in terms of difficulty, but the importance of all that background scaffolding i.e. mature adults speaking and writing the language to each other, kids listening in, reading, shouldn't be underestimated. When that background is missing, one often sees the debilitating effects. In today's world, television is an important part of immersion and language learning. What would Python look like on TV? That's more my focus than traditional book publishing, though I in no way dismiss or denigrate the importance of the latter. I just think that domain is already being rather competently handled by others (I'm even more convinced of that since reading recent posts to this list), whereas we have zero "Python cartoons" on TV (showmedo and such screencasting archives are a first step -- I have some pilots in the queue there as well). Regarding mathematics, I think we should recognize that we teach it with a lot of scaffolding as well. True, constructivism (and constructionism) advocate lots of exploration and self-motivated play (becoming self motivated is almost a prerequisite for eventually getting expert at something, no?), but there's really no way a single person, nor even a single generation, is going to reinvent all those wheels from scratch. Take vectors for example. They actually derive in some degree from complex numbers, which come from polynomials (getting generalized solutions). People were having trouble wrapping their minds around these critters until the Argand Diagram came along (another guy had it even earlier) and showed a + bi as (a,b) on a plane of two axes: real (x) and imaginary (y). That set the stage for integrating with trig's unit circle, Euler's identities, plus showed how two complex numbers multiply in the complex plane (enter fractals, Mandelbrot set, at a later date). Hamilton wanted something similar in 3 dimensions and Quaternions were born. But they were difficult, especially pre-computer, and that's where Gibbs and Heaviside enter the picture, using some of these same concepts to create a rather simpler vector arithmetic, wherein addition and subtraction are defined, along with scalar multiplication, plus these two other operations, dot product and cross product. This marks the beginning of today's linear algebra, with its matrix transformations, eigenvectors etc. (still with links to polynomials though e.g. per the Cayley-Hamilton theorem). That's a *lot* of scaffolding, and it took many many decades to get us to where we are today (not a static picture -- lots of momentum towards Clifford algebra ala Hestenes, exploring other forks in the road ala Grassmann et al -- names mentioned in passing on one of my Europython slides (slide #22)). So let's remember, when we teach Python, that we're always relying on lots and lots of scaffolding, even in the case of "blank canvas" programming (it took a lot of evolution and engineering to give as that deceptively blank screen in the first place (I say "deceptively blank" because Python on bootup is already a richly populated environment, as dir(__buitlins__) reveals)). We're born into the midst of things, a world already highly sophisticated and technological. There's no going back to primitive origins except mythologically. In practice, the status quo is immersion, and there's really nothing we can do about that. So why resist? That being said, I agree that blank canvas or blank screen programming has its place. But maybe not without access to documentation. It's not about memorizing everything and moving to some deserted island, asked to code in Python. There's no need to presume such a deserted island model. It's a very busy planet that we need to emulate and synch with. OK, OK, so maybe I'm putting my own spin on "scaffolding" a little, incorporating it into my namespace in a slightly different way. Consider me a visitor to CS from the neighboring philosophy department, where we twist things a bit, but not so much we're indecipherable (at least some of the time). Kirby References: Good history of the evolution of these linear algebra concepts, with lots of links to recent thinking: http://calclab.math.tamu.edu/~fulling/m629/f03/linalg.pdf Linear algebra still very tied to Polynomials of the Italian Renaissance: http://en.wikipedia.org/wiki/Cayley-Hamilton_theorem Typical use of "scaffolding" in the CS namespace: M. Z. Numan, S. Ali, and J. Giniewicz, "Effective Scaffolding for Problem Solving and Higher Order Cognitive Skills", presented at the Annual Conference for the Advancement of College Teaching and Learning, Harrisburg, Pennsylvania (2002).