The two species are complementary; hedgehogs have the knowledge and expertise necessary to develop fine-granularity models of their domains, whereas foxes find commonalities and connections between many coarse-grained models. Foxes rely on hedgehogs to form their knowledge base; hedgehogs can use foxes' insights for new directions of research and development. Innovation (as opposed to optimization) is the result of cooperation occurring across vertical and horizontal model boundaries.
All this discussion is with the implicit assumption that only humans can handle knowledge. But we have machines to help us.
So today it's easily possibly to talk with people from diverse fields, to search for knowledge of specific characteristics in different fields, or to reuse knowledge using code or patterns from a far field.
So it feels that with the right tools and technology and training, people could connect ideas from multiple fields, without deep investment, and being a polymath could be "democratized", similar to how many other complex tasks have been democratized.
And this field seems somewhat under-invested, relative to the potential.
Until computers get a lot smarter in understanding research papers, you'll always have the problem of unknown unknowns. How do you discover things in other fields if you don't even know the terminology? For example a medical researcher reinvented (or at least claims to have reinvented) some calculus: https://academia.stackexchange.com/questions/9602/rediscover...
If you never heard of calculus, googling for solutions can be difficult.
I think that Paul Otlet may be your man for that sort of thing. He was obsessed with how to organize and catalog not only things, but knowledge and information.
Interesting guy, and ahead of his time. It's too bad that his work (along with the rest of the world) was mostly dropped on the floor in the chaos of WWI and II. There's a book about him called, "Cataloging the World: Paul Otlet and the Birth of the Information Age."
This is a paper from 1994, before the Internet etc we're popular.
But let's try:
1. Google to find what the relevant fields are - you get math. Search a forum about math -> discover math.stackexchange.com -> ask for help or even just terminology regarding "area under my graph/curve"? Once you get that , you could follow on it(say ask for a library of the most accurate integration method we know of today).
2. Google that, find relevant results and dig for terminology, and follow on it.
And sure , this isn't a bulletproof method, and you won't get 100% coverage. But it help find connections .
Its still people answering the stack overflow questions, and its still people writing the posts Google finds.
Not to mention knowing what to search for and how to find it is itself a very broad skill that requires ploymath type knowledge to really get the most out of it.
Same reason librarians seem to know a bit of everything.
I'd see the optimum as the opposite type of "democratization": replacing most narrow experts with smarter and smarter AI/ML systems (that will sooner or later surpass human-level specialist knowledge in most areas), and have like 90% of human beings educated to be integrators/"polymaths"/expert-generalists instead! That would make 1000x more sense!
And of course, humans can be augmented with higher bandwith interfaces when these becomes available, then have their minds uploaded and become "software", when that becomes possible etc. The general idea is that human-pattern minds (whether human or future human-like ones) are one of the best type of minds for integrating/generalizing/being-polymaths etc.!
Imho our whole educational systems are 100% wrong for focusing on training narrow-experts and not of a mix of mostly integrators and "T-shaped people" instead! It's like we're training fishes to climb trees! Human brains are inherently bad at focusing and specializing by nature of how evolution built their freakin hardware even.
