How important is it really to be able to recall information using ones long-term memory these days though? With my phone I pretty much have access to all the information in the world from one single place, from almost anywhere in the western world. Why would I want to memorize most of it? Books are not only for sharing knowledge but also remembering it? These days, just memorizing core concepts such as finding the correct information, how to evaluate it, and how to apply it should be enough? Leave the rest of the recalling for our computers.
I think it's at least very helpful in building emergent knowledge, knowledge of connections between things that, only after experience / long-term processing, you realize are instructive or related or at least similar or, very commonly, "Y is really just a special case of the more general X" idea.
It's still completely possible to learn some of the outgoing connections or equivalences after a specific search of the internet, but the more things you spend time ruminating on, the more connections/realizations seem to almost "bubble up" out of your subconscious/whatever. I think this is what [inspiration](https://en.wikipedia.org/wiki/Artistic_inspiration#Ancient_m...) is: the result of subconscious processing or drawing relationships between thoughts in your head that are memorized. That, I think, is what you wouldn't have anymore if you were to completely outsource your knowledge to the Internet. I have witnessed some brilliant statisticians come up with completely new ways for researchers to think of their data/analysis, without ever consulting Google...though that's a relatively poor example, and more related to developed Mathematical Intuition than anything else. I would argue that said developed intuition is a perfect example of drawing underlying connections between things by letting them stew around in your brain, connections that would be very much harder to find overtly.
It's a difference in speed. I know people who can give an estimate of a scientific/technical idea working within about a minute by calculating things like signal strength by Fermi estimate.
It moves conversations along with a much better flow, because now you can focus on how to improve that, make a new estimate and iterate, and discuss the real-world implications from this, such as could you provide that much power or can you buy the components somewhere.
So, enough knowledge for Fermi estimates is really useful.
