Hey thanks Alex, I appreciate it — the two biggest things I do are 1) use audiobooks so I can quickly pick up a book when I’m not doing something else (probably 80–90% of my ‘reading’ is while walking somewhere or otherwise in transit), and 2) listen at increased speed (generally 2.5x). I find that debate helped with being able to process the…
Yeah I think that’s plausible. The bit that still gives me pause is when you see folks describe certain openings/variants as ‘playable at X level’, and that GMs converge upon certain paths. This is Magnus, after all — but I guess one way to test it is, once model in-hand, run its moves back through PlayMagnus and see the outcome.
One question that comes to mind is whether 2Mn games is too many — are these all GM games?
What you’ve implicitly defined here is that a good position is one that appears in games with a good outcome — that, absent other information, a position was probably good for White if White ended up winning the game.