> Sure. Speak French without learning French. Or maybe learn French by reading a picture book.
That's basically the approach Rosetta Stone takes, and it works fairly well.
Rosetta Stone for math? I'd like to see it.
I have a friend who was able to buy a laxative in Italy without knowing a word of Italian. It's a funny story. I doubt the same approach would work with math.
One example: Using Robinson infinitesimals allows you very easily to write code for forward mode exact differentiation (not symbolic, not approximate). But justifying these simplifications is hard. The question is how much math can be simplified by analogous means without wrecking the foundations.
Another example: Many really useful systems have to deal with uncertainty. I have yet to see a system that allows programmers to easily build such models. I have seen some nice ideas probability monads, Bayesian networks, etc. But how many non-specialists are prepared to use such tools? Happy to have HNers prove me wrong on this one.