Good post. I think the discussion around how one thinks about learning programming is very important. I've noticed everyone approaches it differently and each methodology can have pitfalls.
For example as a self taught programmer I used to just have an idea and throw myself in without reading any other similar code or taking the time to learn some of the fundamental concepts required to write code in that domain. The result of this was that I would spend much too long creating hacky solutions and end up with very messy results, while also failing to master my tools.
It's tempting to jump into a project straight away especially because programmers chronically underestimate the time it will take. However, i think it pays huge dividends to spend some time evaluating current solutions and the methodologies used. Then you can make better estimate and decide if you really want to commit. Once your ready to commit then take the time to get to know your tools and how you want to approach the problem.
Your idea of finding a similar code base to your interests and learning how to manipulate it is brilliant.
RE: Smart Code Learning: Breakdown approach (Inductive reasoning)