To know and not to do is not yet to know.
This idea has been attributed to many sources. Let’s assume it is essentially Buddhist. The same idea is expressed below in terms of neuroscience.
These items are saying that you need to do things in order to really get them. There are others: Malcolm Gladwell’s 10,000 hours, getting to Carnegie Hall, and the legendary practice regimes of athletes like Jerry Rice, Cris Carter, and Ted Williams.
Of course, it isn’t just volume that builds skill. Practice doesn’t make perfect; perfect practice makes perfect. You have to figure out what to practice, whether and when to practice to your strengths and weaknesses. You might have to mix things up to stay fresh. (By the way, anyone who tells you that computer work doesn’t take a toll on the body hasn’t done it in earnest. People who sit too long need to mix it up more than anyone.)
A lot of other factors come into play as well. Natural talent is one. Baryshnikov worked like crazy but he would never have been who he was without some natural talent. Opportunity is another. Would Bill Gates have been a runaway success in the 1930s? Will a poor but smart and diligent young person in Africa or India be able to do the same things as a graduate of MIT? (That idea is both thrilling and scary.)
Another issue is whether one is trying to learn an entirely new skill, like learning to program for the first time, or a related skill, like learning to program in a new language after already knowing how to program in general, or leaning a new API after having learned many before. Each new endeavor has its own “is-ness” but picking up related skills is far, far easier than picking them up the first time.
Finally, the breadth of the skill is important, and not all skills are equal in this regard. If you know how to play piano then learning to play guitar might be somewhat easier, but learning to play organ or harpsichord would have to be a lot easier. Learning CSS is one thing, learning to leverage SASS and SCSS within CSS, if one already knows HTML and how to program in depth, is far less taxing. I’ve heard it said that one can employ Hadoop proficiently in about two weeks (presumably if one already has a background in data center operations).
I’ve enjoyed and will continue to enjoy getting extra reps in the new languages and tools I’m learning. They are interesting on a direct level because there are specific things you have to know, and interesting on a meta-level because of how those individual bits of knowledge fit into the larger picture.