A friend asked this question recently and it prompted a really interesting discussion. This post doesn’t really answer that question, though. I’m not sure what “good” means.
The nature of work (and maybe reality) is to train you to filter and discard information all the time. Especially in an environment where information has a short half-life, we end up being optimized (by our environment) for being able to find information. Knowing where to find stuff rather than knowing the stuff. This results in our brains being filled with pointers to information rather than the content itself.
The alternative is a medieval monk, who memorizes things all day. They have everything loaded into their head, but due to the time required to memorize, their information will span an inherently smaller set of say 400 books over their lifetime.
In contrast with the Monk, my information set as a modern webivore maybe spans 400,000 book-equivalents, but the tradeoff is that my brain is full of pointers, snippets to deeper concepts. There’s something unsatisfying about not feeling like an expert in things, though.
Our brains are information caches. Caches are tiered. Thanks to the technology of writing, I can store one level of information in my head, and then ‘page out’ to other external forms of information, my notes perhaps act as an L2, the source material acts as an L3, etc.
Recall
Notes are useful for two things, recall and insight generation. Recall is being able to pull things back up in order to support arguments or ideas that I’m trying to ‘call forth’. Paging in things I’ve seen in the past
I can present as an argument a take like “GDP is a bad metric”, but in response you’ll ask “why?” and now I have to do something to go retrieve back the information that I consumed in the past in order to hold this belief. Further, it’s not enough to just have a pointer to “Piketty’s book” (or whatever), because just holding the book or even having full text search over the book is not really going to help me re-generate the convincing argument for why GDP is a bad metric. I need these intermediate tiers of cached distillation in order to operate. The book is the highest volume, followed by large highlighted chunks, then notes I took alongside reading, and then the broadest but lowest volume concept summaries from the book.
Ideal concept summaries contain links into the deeper tiers of information, allowing for quickly getting back to “that one graph which proves my point”, or whatever.
Or, you could be an economics PhD, and have the whole argument and all its content close at hand. There’s something about researching which is inherently about loading all of this information into mind at once.
Research
My friend Charles works on version control related tooling at the place we work, so he’s got a bunch of crystallized intelligence about version control that is boosted by his job but also diluted by all company specifics, engineering miscellanea kinds of things. If Charles were full time researching version control then you’d be maxed out on that one axis. But even working on hg (eg) full time would not max that out, in that world Charles would be diluted on what the Github issues of the day are for his project, and how the hg build and deployment system works, etc.
So it’s a question of how research-y one’s job is, and not whether their title is ‘researcher’ as such. The degree to which your job is research-y is the degree to which you’re an expert in that thing. Research in effect drags you ever closer to just doing ‘core concepts’ things with a minimal spanning set of skills that you need to get the research done.
But… this brings us back to the Monk. If you’re an economics PhD, you’re only an economics PhD. What happens when you’d like to comment on ideas outside your particular field of fully-cached content? You’re back with the rest of us mortals.
It’s not all rosy, though. From real researchers I’ve talked to, it’s not like their jobs are 100% research. Their lives are full of grant writing, for instance. And to reduce the shininess even further, the researcher may be concerned that they aren’t actually able to do anything, given the minimum spanning skillset effect.
Compression
By virtue of deep (true?) understanding, you’d be able do rehydrate arguments from first principals, the conceptual equivalent of a memory palace, you can walk yourself through the process of coming up with the idea rather than having simply memorized it. Someone who really understands the derivation of an advanced concept need not actually memorize the rule, they can just re-derive it when needed. So there’s a kind of compression factor to crystallized knowledge, where if you have the structure figured out properly, you can really only have the seed of an idea memorized, and then water it and have it blossom forth when needed.
Seeds are the folding lawn chairs, the telescoping rods, the go-go- inspector gadgets of the mental landscape. Along with being full of pointers, my brain should have a good store of seeds too.
Insight generation
Ideas are good for insight generation. Notes are externalized ideas. Notes are also good for insight generation.
We get dreams for free. One of the speculated features of dreams is that we’re doing this kind of perusal and curation of the thoughts in our head. Our brains are self-indexing.
Notes get messy though. A digital gardener does a good job at this, but I find myself letting my collection of notes lapse and become ‘write-only’.
This lapsing hampers my ability to distill new ideas by interfacing with my collection of notes and thinking in concert with it, the way I think in concert with the ideas in my head.
It seems like there may be technological ways to at least improve the situation here. Roam and Obsidian have exploded in popularity, but suffer from the lapse problem. Maybe something like Subconscious will improve the game?
Trees and Branches
Besides pointers and seeds, another piece of the mental landscape is the scaffolding you hang all this information off of. This is maybe the ‘first’ or ur-piece of the landscape. It’s easier to memorize and learn events from history if you can slot them into the surrounding context and web of ideas of other events in history. If there’s a continuity to plug them into. Without that, the ideas just fall away and are lost.
Viraj's Notebook 