>>482192187The books don't solve the innate problem though. That's my point.
Efficiency is just "best practices" and "probably good enough". And it provides zero insight into specific moments. Like chasing a yakuman.
For example, if chasing a yakuman is like a 10% winrate for +32,000 points, that's EV 3,200 points. But then you have to consider chances of ryuukyoku for like say 20% to get 0 to +3,000 points, and dealing in for say 15%, and also being forced to fold by somebody else and losing to tsumo / bleeding noten payments. There are actually a lot of possible outcomes.
And then if chasing a yakuman is stupid but you can break it up for say a 3 han 4,000 point hand for like 30% winrate, that'd be an EV of 1,200 points. Objectively, the numbers in this hypothetical would say "chase the yakuman even if it's not efficient" while a lot of people would start rambling about efficiency and how yakumans are too rare to matter; and that you should've gone for the shit hand no matter what even if EV says otherwise. That's the problem here. People have opinions, and no objective way to compare and also prove themselves right or others wrong. And no data either. If the yakuman is actually 1% winrate, players wouldn't know because there's no way to know that. If the yakuman is actually 50% winrate, players wouldn't know either. The only numbers we have are tenhou stats from years ago because they locked away recent game results because they're selling it to Naga and other bot devs for money. There's Amae but the data is locked.
Of course, looking at the wall doesn't solve this either because it's a per-instance situation, and doesn't evaluate all situations but that goes back to my original point: Players don't have an anchor or reference point for "all situations". Blindly following somebody else's instructions only go so far if they don't know WHY it's better other than "works on my machine". Sounds like survivor bias rather than objectively correct proofs.
