I once wrote a custom build systems in python to generate Make files. I did this because I wanted to use convention instead of configuration and achieve fast builds on multicore machines. Later, after leaving that company, I wanted to recreate a similar build system. I wrote it in python to generate Ninja build files.
I found it easier and cleaner to write Ninja build files.
I wasn't able to do a direct comparison of the build speeds of the two systems, but I bet the Ninja builds were significantly faster.
I was 15 when I read Issac Asimov's book Understanding Physics. That one book caused a radical change in me. Even though I had always found math and science enjoyable, I was struggling at the time with the Physics 1 class I was taking in high school. I read the book half way through the first semester, and went from a state of struggling to a state of effortless understanding. That experience taught me that for me, the best way to learn was to teach myself. First read a well-written layman's overview of a topic, and then dive deep with well-written academic books.
Extrapolating a bit from that lesson, and with ~40 more years of hindsight, I'd highly recommend that a bright 14 year old read some books that give a broad overview of science & philosophy, and establish a thorough understanding of the scientific method. One that I happen to be reading right now that I feel is remarkable is David Deutsch's The Beginning of Infinity. Deutsch is a theoretical physicist at Oxford working on Quantum Computation. The Beginning of Infinity (despite the new-agey sounding name) is an incredible work.
Another book I highly recommend is Richard Dawkins' The Selfish Gene, not just for how it explains Neo-Darwinish, but because of the chapter on Memes. Deutsch also has a great treatment of Memes.
