It can be hard balance to strike, especially when you've got a book that's really good and dense with quality information.

When that happens, sometimes I find that taking it a step further and only grabbing that you use for your workflows can be helpful, then taking what you just learned and practicing it as soon as possible afterwards to help cement it in your mind.

Tim Ferris has actually talked about this a little bit as the difference between JIT (just in time) and JIC (just in case) learning.

In this context, most of what we're doing while mining books for data is JIC, but I have definitely found situations where I'm copying the whole book like you mentioned, so I'll practice hard, and then rely on JIT (just looking it up as I need it) for the rest of it.