Strong Chess

The "revealed" account is named PeterThiel

HOW BIG IS CHESS? There are three numbers: Atoms in the universe: 10^80 or so Next, the "Shannon number" 10^120 is famously associated with chess. It comes from a 1950 paper Claude Shannon, prolific researcher and namesake of that other Claude. The Shannon number estimates the number of possible chess *games* under some very simplifying assumptions: - That every full move, there are about 1000 different branches, which corresponds to about 32 legal moves per position (in reality, this varies from 1 to a max of 218) - That the average master chess game lasts 40 moves before resignation - and 1000^40 = 10^120 Of course, a chess game can last hundreds or even thousands of moves before the 50-move rule kicks in. The true number of possible games is therefore much larger--some silly number with tens of thousands of zeros. BUT! Of course all of these games would transpose into each over every which way. Imagine the vast number of ways a game might arrive at the exact same queen versus king checkmate. The number of legal *positions* is a very different story. A Dutch computer scientist John Tromp (a little chess reference in his name!) estimated this number using statistics (and Haskell!). He created a confidence interval around the number of positions that can be reached in a legal game of chess. It is approximately 4.8 x 10^44. So, still pretty big, but there are enough atoms! It is also interesting to note that chess *is* strictly solved with 7 or less pieces on the board, and for a small slice of 8-man positions too. But on a more practical level, computers already play chess in a way that looks a lot like perfection to the naked eye. So we are both many orders of magnitude away, and kinda close, to solving chess, depending on how you look at it. Will chess be solved in our lifetimes or not? Who knows. Either way, I'll still be in the 3+0 pool trying to flag you.

My #1 complaint with commercial gyms by a mile. I can kinda understand shitty barbells, because I get that clients quickly abuse and damage them, but this exact brand of bench is a ubiquitous butthole cancer

is this performance art, or what the fuck

Comparing LLM codegen to chess engine history: I think models still fall short of elite humans but are astonishingly fast at low-medium difficulty tasks. Much like 80s-early 90s engines weren't GM level, but vastly faster at tactics. But they're still cringe at writing English

delete your account

delete your account

i always thought it would be funny to tattoo my entire opening repertoire all over my body like Michael from Prison Break

About 1% exactly by the Elo EV formula. But in practice, big underdogs actually score a little higher than expected. A couple years ago, I (2200) returned to the board and realized in horror I was dead lost against a 1200. I applied ferocious effort for hours and won--narrowly!

First contraction must start with I L eliminates I'LL --> first word is I'VE I'VE GOT A GOOD FEELING ABOUT THIS is one of the most ready I'VE phrases and fits

lately I am staying very hydrated and what the fuck who has time to pee this much?? I have a family and a job

SPECIAL THANKS to @Kaldas90 for the line I used to brutally massacre my 12-year-old son in daily with 100.0 accuracy

i watched some Marie Kondo videos, now i am folding everything in the house with an autistic fury

The Middle by Jimmy Eat World came on at the chess meetup and the 24 year olds were like, "wait what song is this?" and it was TODD'S TIME TO SHINE

The problem is this mixes up sequences of moves with positions! 10^40 is already very close to the number of unique positions legally reachable ever in chess, while move sequences transpose all over the place and reach vastly higher numbers

Assuming that everyone picks randomly, which judging by timeline is approximately the case, a vote for blue is: Suicide: 49.999554% Irrelevant: 49.999554% Saves half the world: 0.000892% Similarly, red kills half the world 0.000892% of the time, otherwise irrelevant.

The Scheveningen. Huge in the 90s, now almost extinct

Exact parallel with @SS_strength here. "How does LeBron train?" seems like a good question, but turns out not to be. The boring, "which methods have been shown to repeatedly work across most of the population?" is the one you should be asking.