> something you can almost self-teach from widely available free online content.
I think you massively underestimate just how much data is online for everything, especially once you include books which are freely available on every possible subject (illegally, perhaps, but if Meta can download them for free then so can everyone else).
There's less noise for many other subjects than for software engineering, there's often just a couple rather than 100s of competing ways to do everything. There might just be one coursebook rather than 1000s of tutorials. But the data for self teaching is absolutely there.
Consider two fields with vast amounts of literature: medicine and law.
Medicine faces two key challenges. First, while research follows the scientific method, much of what makes a good doctor—intuition, pattern recognition, and clinical judgment—is rarely documented. Second, medical data is highly sensitive, limiting access to real-world cases, images, and practice opportunities. Theory alone is not enough; hands-on experience is essential.
Law presents a different problem: unknown unknowns. The sheer volume of legal texts makes it nearly impossible to be sure you’ve found everything relevant. Even with search tools, gaps in knowledge remain a major risk.
Compounding this is the way law is actually practiced. Every judge and lawyer operates with a shared foundation of legal principles so basic they are almost never discussed. The real work happens at two higher levels: first, the process—how laws are applied, argued, and enforced in practice. Then, at a third, more abstract level, legal debates unfold about interpretation, precedent, and systemic implications. The first level is assumed, the second is routine, and only the third is where true discussion happens.
Self-teaching is easier in fields where knowledge is structured, accessible, and complete. Many subjects are not.
A significant chunk of human knowledge is not publicly accessible. You cannot self-teach how to make a modern aircraft, jet engine, nuclear reactor, radar tech, advanced metallurgy etc.
Similarly, I would wager most of the useful economics and financial theory that humans have come up with is only known to hedge or prop trading firms.
For some subjects, the entire journal-published academic body of knowledge for it is probably some useless fraction of the whole and university academia is operating nowhere close to the cutting edge. People are probably doing PhDs today on theses that some defense contractor or HFT firm already discovered 20 years ago.
Even things like specialized medical knowledge, I would wager is largely passed down through mentor-mentee tradition and/or private notes as opposed to textbooks. It's unlikely that you can teach yourself how to do surgery just from textbooks. I once had a pathologist's report use a term for a skin condition that was quite literally ungoogleable. The skin condition itself was fairly ordinary, but the term used was outright esoteric and yet probably used on a daily basis by that pathologist. Where did he learn it from?
Taylor Wilson built a nuclear reactor at his home when he was 14. People build jet engines and put them on modern model aircraft every day.
If the instructions aren’t immediately available, the internet provides connections and forums to find anything your heart desires.
Information wants to be free.
Arbitraging micro-opportunities (or far more likely, deploying insider information masked as HFT or some secret sauce arbitrage) is not economically useful.
I think you massively underestimate just how much data is online for everything, especially once you include books which are freely available on every possible subject (illegally, perhaps, but if Meta can download them for free then so can everyone else).
There's less noise for many other subjects than for software engineering, there's often just a couple rather than 100s of competing ways to do everything. There might just be one coursebook rather than 1000s of tutorials. But the data for self teaching is absolutely there.