[coldtea]

>His 'theory' (highly flawed, I think) is that one needs excellent memory to be able to make decisions now, without needing to take the time to research them.

I don't see any flaw in the theory. When you code for example, if you don't know in advance about several idioms, data structures, algorithms etc that's (most of the time) not something that you will make up later by researching and changing your program. It's simply something that will take you down a narrower path and constrain your programming.

I'm not talking about knowing the details of algorithm X, or how to implement it from memory. But if you don't know it's existence even, it wont be an algorithm you'll consider when you write your program.