A while ago this talk was recommended to me (most likely through some Hacker News comment thread), and I finally got around to it. I did not know the author, Edward Kmett, before this; he seems to come from a computer science background (lots of Haskell, category theory, etc. which is all fun!), and he is well aware of the continuous learning needed for such a field.

I have only a few loose notes on this, and I'll tackle them in the order mentioned in the talk. Maybe in the future the ideas will be a big more coherent!

In the brief introduction about himself, Kmett mentions how valuable it was to come back to college after having had some professional experience (from my humble and naive perspective, the advantages are not so clear, but maybe I don't have enough professional experience, or depth in a given field for it to be worthwhile);

Almost immediately there is a reference to Richard Feynman, which comes across as a signal for relentless curiosity, playfulness, and a possible appreciation for academic ties; I suppose I am myself very influenced by Feynman, and while it clearly doesn't show neither in my capabilities nor output, it triggered a small shift in my perception of what a genius really is.

Later on, Kmett references one particular Feynman practice: to always hold a dozen problems in mind, and always check newfound solutions and strategies against those problems.

Another sound piece of advice: do research and development simultaneously. Assumming these are opposite ends, it increases the likelihood of arriving at common ground. In more concrete terms, I feel like it helps avoid spending all of one's time in the domain of theory: one ought to solve problems, after all.

Then, a few techniques and behaviors that could enhance being a good learner:

- spaced repetition — a natural strategy. This is most likely the most significant habit I have adopted in my life; if nothing else, it makes me feel like I can learn anything, given enough time (and appropriate conceptual breaking down);

- iterative deepening — to always go a little further than last time. Catching up with past work is, itself, a form of spaced repetition; but one should always try to go one step beyond.

- write stuff down, or: don't be too smart. Everyone knows the feeling of something being obvious one day, and inpenetrable shortly after. Write clearly, insightfully, and be willing to explain the jargon (better yet: do a preliminary glossary!)

There were also a few references to some mathematicians (in particular, the works of Serre and Alexander Grothendieck), and a great list of follow-up resources.