The part that stuck with me is femto's 'slow programming' line. I never hand-assembled anything, I came up too late for that, but I still write the tricky parts of a program on paper first. Last month it was a little state machine for a game I'm tinkering with. Two pages of boxes and arrows, and I caught three logic errors before I ever opened an editor. I think the paper forced a kind of discipline that's easy to lose now, when the cost of trying something is one keystroke. The flip side is that Woz had to be right before he ran anything, and that's a brutal way to learn. I'm not sure I would trade the fast feedback loop we have now for it, even if the paper habit is worth keeping.