I've been fascinated for a long time by just how much effective autodidactism there is in software vs. other fields. There are tons of people who have made majo

I've been fascinated for a long time by just how much effective autodidactism there is in software vs. other fields. There are tons of people who have made major contributions here that are completely self-taught. Software is uniquely suited to autodidactism for three reasons: (1) The tools are easy to obtain and easy to start using. Capital cost is low to non-existent. (2) The learning feedback loop is nearly instantaneous and the results are almost always perfectly objective. Things either work or they don't. There is not much room for delusional or wishful thinking. (3) Resources for learning are readily available and are mostly written in a style that is utilitarian and straightforward rather than cliquish and arcane. Theoretical physics passes on point #1 until you hit the need to do serious experimentation, but it fails on points #2 and #3. There is no command prompt that will tell you in 10ms if a theory is at least rational and internally consistent, and advanced mathematics has an arcane symbology and jargon that seems almost intentionally designed to resist penetration by those outside the academic circles where it is used and taught.