#851
That is correct
Tic Tac Toe, Nine Men's Morris, Checkers and presumably chess are draw.
Connect Four and Antichess are wins for whoever plays first.
Chess will never be solved, here's why
Sort:
Feb 7, 2022
0
#843
mitsubishieclipse wrote:
As for "computer Poker", no. They can be programmed to give playable hands. For example, 3 players, it can be programmed to give 2 people a pair in the dealing. 3rd and 6th cards are 4's. The whole deck can be programmed. If you played Zynga Poker on Facebook, or Xbox Live, you will know that very rarely is there not a decent hand given to someone. When you play with REAL cards, you can get many more boring hands.
Almost exactly 50% (49.9%, to be more precise) of hands are better than a high card, so at least 2 out of 3 hands are this good 50% of the time.
P(2 or 3 out of 3 better than high card) = 0.5 * 0.5 * (1 - 0.5) * 3 + 0.5^3 ) = 0.5
I suggest you record your hands over 100 hands or so and see that it is indeed consistent with random dealing.
Feb 7, 2022
0
#844
"You told me literally “More time doesn’t help you catch mistakes.”
Did not say that.
But by trying to insist that I did - you can disagree with your own idea.
Its very neat - and nobody can stop you doing so except the original poster.
I doubt even the moderators would stop you doing that.
I doubt you'd be able to compete with 'the other guy' though. 
But maybe you will be able to 'eclipse' him. With a fast foreign car ...
playerafar wrote:
Optimissed wrote:
Are you manic-depressive?
Probably not reportable to the moderators - as nobody is named.
'somebody' has admitted to being 'perturbed'. But he keeps showing up here. Perhaps that's an objective of 'somebody'. Although its not the objective of the forum. Nothing in the opening post to the effect 'get perturbed here'.
But as always - it can be posted around.
And just now - we got new posters.
Who also posted around the provocations.
A good thing. Progress.
Someone is watching the monkey-house from afar. That's progress too.
TsetseRoar wrote:
Strange thread...most people seem to have no idea what "solving" a game means.
Solving a game does not mean being dominant over all other humans...otherwise chess was "solved" by Morphy, even though we can find many suboptimal moves in his play now.
Solving also doesn't mean calculating every permutation.
What it means is a mathematically best strategy has been shown -- either an unstoppable strategy that always wins for one player, or a strategy that forces a draw (where an unstoppable can be proved to not exist).
At this time, the game of chess has not been solved, but there is no reason why it is impossible, or intractable.
I think Go will be solved before chess, and when that happens we can suspect chess is coming soon. Go has many more permutations than chess, so is sometimes described as a more complex game, but I think, given only one kind of "piece" and "move" it looks a better candidate for finding an unbeatable strategy.
Deepblue was Champion in 1997, while AlphaGo Zero made it to beat the World Champion only after 2015 (It had won WC Lee Sedol and Highest-Rated-at-the-time Ke Jie by 2017), which points out a possibility for Go being later solved than Chess.
It could also be that the larger Go board with 361 places to drop pieces increases its complexity by exponents. Seen some arbitrary mechanical calculation on chess, Shogi, Go, checkers, etc. and Go rates about 50+% higher than the others in terms of logarithmic complexity.
#857
That is right Go is more complex than chess.
Chess is not nearly as complex as people used to think.
Even 10^37 is a massive overestimation of the number of legal and sensible chess positions.
Nevertheless chess is still much more complex than checkers.
#859
Well Mr. 2300, do you think the following randomly sampled positions without excess promotions play a role in solving chess? Those are in the 10^37 count.
Feb 7, 2022
0
#851
tygxc wrote:
#859
Well Mr. 2300, do you think the following randomly sampled positions without excess promotions play a role in solving chess? Those are in the 10^37 count.
I understand that any proclamations by me on whether they can be ignored are as inappropriate as any by you. I feel this is not a result of my rating, but independent of it (more related to having two maths degrees. "Solving" is a mathematical activity, also found in chess problems and tablebases but not in play in general).
#861
There are two different things: 1) solving chess and 2) assessing the feasibility of solving chess.
Mathematically chess is a finite game and hence it can be strongly solved.
However, people argue solving chess is not feasible as it would take too much time.
To make such statements needs the number of positions involved.
Tromp counted 8726713169886222032347729969256422370854716254 possible positions and found 5% of these legal after sampling and thus arrived at his 10^44.
However the vast majority of his positions contain 9 excess underpromotions. That never occurs in any real game between humans or engines.
Hence the newer estimate 10^37 without excess promotions is a better estimate.
It is not true that there are twice as much positions as diagrams: every position with black to move can be converted to a diagram with white to move by switching colors: up/down symmetry.
Every position with lost castling rights can be converted to its left/right mirror image.
Randomly sampled positions without excess promotions like the four #860 probably play no role either in solving chess. This might reduce the count by another factor 10^6, leaving an estimate of maybe 10^31.
Feb 7, 2022
0
#853
mitsubishieclipse wrote:
Tic Tac Toe is not a game. You've been lied to. A quick Google search will tell you that whoever goes first and has first move, the same square always wins. It cannot be defended. If you have ever gone first, and lost, in tic tac toe, it is your own fault. If you pick the right square first, nothing the opponent does can win. First turn person should ALWAYS win 100%. If they pick that square. No supercomputer needed.
Is chess a game? The same may apply.
Incidentally, in tic tac toe, "whoever goes first and has first move, the same square always wins", is a bit optimistic. With perfect play it's a draw.
Most kids are introduced to it at about the age of five and strongly solve it the same day, so they know that at least.
Feb 7, 2022
0
#854
Seems quite hard to explain to @tygxc that while when you are playing a game you can ignore options for the other side and truncate analysis lines, this is _absolutely_ _not_ true when you are solving a chess problem rigorously (and the same goes for the problem "solving chess", which can be described as "From the standard starting position exhibit (or prove the existence of) a forced mate by one or the other side, or exhibit (or prove the existence of) a forced draw for both".
Feb 7, 2022
0
#855
mitsubishieclipse wrote:
No idea what you are talking about.
You told me literally “More time doesn’t help you catch mistakes.”
Which is completely false. No argument to it what-so-ever.
You told me literally “More time doesn’t help you catch mistakes.”
Which is completely false. No argument to it what-so-ever.
No argument to it whatsoever?
On the face of it SF and other engines use a discrete number of possible evaluations for their positions rather than the "real values" considered in the article.
tygxc wrote:
#861
There are two different things: 1) solving chess and 2) assessing the feasibility of solving chess.
Mathematically chess is a finite game and hence it can be strongly solved.
However, people argue solving chess is not feasible as it would take too much time.
To make such statements needs the number of positions involved.
Tromp counted 8726713169886222032347729969256422370854716254 possible positions and found 5% of these legal after sampling and thus arrived at his 10^44.
However the vast majority of his positions contain 9 excess underpromotions. That never occurs in any real game between humans or engines.
Hence the newer estimate 10^37 without excess promotions is a better estimate.
It is not true that there are twice as much positions as diagrams: every position with black to move can be converted to a diagram with white to move by switching colors: up/down symmetry.
Every position with lost castling rights can be converted to its left/right mirror image.
Randomly sampled positions without excess promotions like the four #860 probably play no role either in solving chess. This might reduce the count by another factor 10^6, leaving an estimate of maybe 10^31.
The bolded statement above is characteristic of your arguments in general. Imprecise and "selectively detailed" (i.e. you try to use realistic numbers in some stages, then summarily make leaps of faith in other stages).
I guarantee you that the "vast majority" of the aforementioned positions do not contain 9 (no more and no less) excess underpromotions. Which is exactly what you just claimed.
You then go on to toss out your usual offhanded comment about throwing out another 6 orders of magnitude in a manner that seems to defy any realistic understanding of exponents .
This is the difference between you and Mr. Tromp, who had to keep shooing your extrapolations about his study away. His careful and accurate study eliminated between 1 and 2 orders of magnitude after all his work, while you routinely eliminate 6-19 orders of magnitude on a personal whim with seemingly no more than some pie-in-the-sky musings, loosely tied to somebody else's estimates. You're like a conspiracy theorist that takes X number of good scientific studies, then mashes them together into a set of ridiculous conclusions by taking wild liberties with how those numbers correlate.
Never realised chess was a problem.
TheHarbingerOfDoom wrote:
Never realised chess was a problem.
I didn't think I'd see you in the monkey house.
Tic Tac Toe is not a game. You've been lied to. A quick Google search will tell you that whoever goes first and has first move, the same square always wins. It cannot be defended. If you have ever gone first, and lost, in tic tac toe, it is your own fault. If you pick the right square first, nothing the opponent does can win. First turn person should ALWAYS win 100%. If they pick that square. No supercomputer needed.
Ummm...no. You are all over the map here. I changed your accurate statements to blue, and incorrect statements to red.
Tic tac toe cannot be lost if you move first and know how to play correctly, but you cannot win against a player that knows how to play correctly either. Tic tac toe is a draw with best play. I'm surprised this post even made it 6 hours without being refuted
.