At what point in a neophyte programmer's life should he/she switch from the "immediate & non-stop coding" Khan Academy approach recently discussed here on HN to this Ken Thompson "take a moment and think first" approach? Isn't there the danger that they might not be able or motivated to make the switch?
It's important to see that these aren't mutually exclusive, even though they sound quite contrary. On the one hand you have the hack -- the one-off prototype thrown together simply to verify what can or can't be done. On the other hand you have the design -- once you know what perspective is best, to build the system from that perspective and validate that your intuitions are building one coherent whole.
Kay's Law is that the right perspective is worth 80 points of IQ: that is, if you find the right way to look at a problem, you can do things with the ease of someone with a 180 IQ who didn't have the advantage of that perspective. How do you get a good perspective? Well, there are a couple different ways. Some are pre-made perspectives: for example, dynamic programming: "let's just build up all the solutions in order from n=0, reduce everything to some other problem we've already built up." Some are intermediate, like wishful thinking: just imagine that magically, you have functions which you don't have, so that you write an algorithm which is correct, but references a bunch of functions which don't yet exist.
Some approaches to new perspectives are more general: hacking, for example, is when you try a bunch of things which just barely work, to find one which does. Another is duck solving: explain your problem out loud to an inanimate object (like a stuffed or rubber duck), and half the time you'll accidentally create a perspective purely for explaining the problem which is useful for solving it. These are very generally phrased because "problems" are a very general topic for intellectual discourse.
When I'm trying to understand foreign code in less than enough time, I instrument it (almost always at inputs/outputs) and try to treat functions as black boxes. In an ideal world the black boxes would be working.
Data flows are as important as algorithms. See this thing from Guy Steele.
Fred Brooks, in Chapter 9 of The Mythical Man-Month, said this:
"Show me your flowchart and conceal your tables, and I shall continue to be mystified. Show me your tables, and I won't usually need your flowchart; it'll be obvious."
That was in 1975.
Eric Raymond, in The Cathedral and the Bazaar, paraphrased Brooks' remark into more modern language:
"Show me your code and conceal your data structures, and I shall continue to be mystified. Show me your data structures, and I won't usually need your code; it'll be obvious."
That was in 1997, and Raymond was discussing a project coded in C, a procedural language. But for an object-oriented language, I think this aphorism should be reversed, with a twist:
"Show me your interfaces, the contracts for your methods, and I won't usually need your field declarations and class hierarchy; they'll be irrelevant."
I think, however, that practitioners of both procedural and object-oriented languages can agree on Raymond's related point:
"Smart data structures and dumb code works a lot better than the other way around."
"Smart data structures and dumb code works a lot better than the other way around."
An interesting assertion. In your link Guy Steele observes the duality between objects (where it's easy to add new data types but harder to add new operations that work on all of them) and abstract data types (where it's easy to add new operations but harder to add new data types).
Guy says that the former tradeoff is almost always the right one, but I've encountered many situations where the latter was much more convenient. It's far more preferable IMO to be able to choose which tradeoff you prefer based upon the constraints of your particular problem.
This usually is not so much a language-level problem as a cultural problem; many programmers are infatuated with OO (I know I was at one time) and unaware of the tradeoffs OO makes or when it's appropriate to use another approach. Hopefully over time multiparadigm languages like Python will help make "objects vs ADTs" more of an engineering question and less of a religious one.
As a recent neophyte, I will use my anecdotal data of size 1 to hazard an answer. I believe once syntax has become relatively understood is when a programmer should start mapping (mentally or on paper) the steps they need to successfully complete a program.
