wat
Chess will never be solved, here's why
Sort:
Feb 15, 2023
0
#8563
tygxc wrote:
@8896
"Nor if your weakly solving involves hopping from position to position with SF15 it doesn't."
++ It does. At 17 s/move on a 10^9 nodes/s it hops from draw to draw as becomes clear once it reaches the 7-men endgame table base draw or a prior 3-fold repetition. If it would occasionally err, as it is expected to do once in 10^5 positions,
I've pointed out numerous time that you calculation of blunder rates based on an extrapolation that assumes an exponential relationship between blunder rates and think time doesn't work, especially when you don't know what values you're extrapolating from.
I've asked you to stop posting the calculation, which you appear to have done, but I'll ask you now if you would also kindly stop posting the result of your calculation.
Try once again. Here is @cobra91's complete table.
At 2048 seconds a move on my desktop I'm around 7½ times short of 17 seconds on a cloud engine according to your calculation. The nearest values in the table that are 7½ times short of the 2048 second values would be the 256 second values.
If we were to use your exponential assumption we could estimate the blunder rates at 17 seconds on a cloud engine as
(blunder rate at 2048 seconds)²/(blunder rate at 256 seconds)
giving (11/290)²/(4/215) blunders per ply ≈ 1 blunder in 13 ply
(not close to 1 blunder in 10⁵ ply)
then that will show by not reaching the 7-men endgame table base draw and can be easily repaired by retracting that move.
Er, which move exactly? Are you going to solve chess before you begin not solving chess?
If you want to emulate 17 s/move on a 10^9 nodes/s cloud engine using a 10^6 nodes/s desktop, then you need 4.7 h/move.
If you want to reach a win in N moves, then you need time to calculate at 2N depth.
Time to calculate exhaustively at 2N depth, which you're obviously going to get nowhere near at 17 seconds on a cloud engine if N>1000 ply, say.
Hopping from draw to draw is easier as it does not require looking ahead to 2N.
@Elroch can you send me a copy of your laughing peanut?
"it is a legal chess position that can be calculated like any other position."
++ Yes, it is a legal chess position. Yes, it can be calculated. No, it is not relevant to weakly solving Chess. What you present is a set of move sequences from a won position.
Only up to the point where SF15 blunders into a draw, which it does in every game. It's very relevant to your procedure for weakly not solving chess.
Here is another series from the set I posted for you in #5898 (October last year) which you've studiously ignored ever since. This one is a drawn position in KRPP v KRP specially for you since your big red telephone apparently tells you that all games finish up in such positions.
and @cobra91's table
If you're asserting that that SF15 will always hop from drawn position to drawn position if the initial position is a draw, how come 9 of the 11 games get into theoretical wins?
No, it is not a sufficiently large, sufficiently strong tournament, where the Poisson distribution applies.
You're right; it's SF15 v SF15 as you say you plan in your process to weakly not solve chess.
If you could find a set of games with perfect play then a Poisson process with λ=0 would naturally apply, but for the moment SF15 v SF15 is the best we have. (And as I pointed out earlier a Poisson process doesn't apply then.)
@8923
"extrapolation that assumes an exponential relationship between blunder rates and think time doesn't work" ++ I fail to see what is wrong. I have 3 data points: decisive games at 1 s/move, decisive games at 1 min/move and 0 decisive games at unlimited time.
"Here is @cobra91's complete table." ++ Of the 4 cases only the drawn KRPP vs. KRP is relevant.
There clearly are anomalies in your data. The error rate should descend monotoneously and it does not. How come it jumps up and down erraticly? There is something wrong with your data.
What are your sequences for this KRPP vs. KRP position? Please repeat or link.
"Er, which move exactly?" ++ The principle is easy. White tries to win, black tries to draw. If all white tries fail, then Chess is weakly solved. Whenever a draw or a white lossis reached, a white move needs retracting. Whenever a white win is reached, a black move needs retracting. According to my calculations the latter case is exceptional as I calculate being right 99,999 out of 100,000 positions. You doubt that result, OK. I doubt your data. Let us look into your data.
i"f N>1000 ply" ++ Chess is not 1000 ply deep. If there are w non-transposing choices per move, then all white moves and all black moves d moves deep lead to w^(2d) positions. There can be no more positions than there are legal positions, so w^(2d) < 10^44.
Feb 15, 2023
0
#8565
tygxc wrote:
@8923
"extrapolation that assumes an exponential relationship between blunder rates and think time doesn't work" ++ I fail to see what is wrong. I have 3 data points: decisive games at 1 s/move, decisive games at 1 min/move and 0 decisive games at unlimited time.
Then you obviously also can't tell the difference between 1 in 13 and 1 in 10⁵. Perhaps you could work on that first.
"Here is @cobra91's complete table." ++ Of the 4 cases only the drawn KRPP vs. KRP is relevant.
Relevant has to be relevant to something. To what? Your big red telephone?
There clearly are anomalies in your data. The error rate should descend monotoneously and it does not. How come it jumps up and down erraticly? There is something wrong with your data.
No. There's something wrong with your understanding. Try it yourself; it may reduce the crap rate on the thread for a while.
What are your sequences for this KRPP vs. KRP position? Please repeat or link.
"Er, which move exactly?" ++ The principle is easy. White tries to win, black tries to draw.
I usually try to win as Black, SF15 just follows it's algorithm.
If all white tries fail, then Chess is weakly solved. Whenever a draw or a white lossis reached, a white move needs retracting. Whenever a white win is reached, a black move needs retracting. According to my calculations the latter case is exceptional as I calculate being right 99,999 out of 100,000 positions. You doubt that result, OK. I doubt your data. Let us look into your data.
Well, amend you calculations. I've pointed out they don't work. Never was any reason to think they would. By all means look at my data. You might have tried producing your own before filling the thread with crap.
i"f N>1000 ply" ++ Chess is not 1000 ply deep. If there are w non-transposing choices per move, then all white moves and all black moves d moves deep lead to w^(2d) positions. There can be no more positions than there are legal positions, so w^(2d) < 10^44.
Here is a 269 move game Ivan Nikolic vs. Goran Arsovic (538 ply)
According to you they've reached every single legal chess position many times over. As an exercise you might try locating this one somewhere in the game.
Feb 15, 2023
0
#8566
It's worth remembering that it would have been just as easy to believe engines were near perfect when they were a few hundred points weaker. i.e. very weak.
Likewise world championship matches had a lot of draws when the players were about 800 points weaker than a top engine now. It did not mean they were perfect: it meant they were not good enough to obliterate the opponent almost 100% of the time.
@8925
"Relevant has to be relevant to something. To what?" ++ To weakly solving Chess.
"I usually try to win as Black" ++ 'weakly solved means that for the initial position a strategy has been determined to achieve the game-theoretic value against any opposition' Hence we look for a strategy for black to draw against all white opposition.
"look at my data" ++ There is something wrong with your data.
"Here is a 269 move game Ivan Nikolic vs. Goran Arsovic (538 ply)"
++ I know that game. They repeated some positions twice. There were many forced moves. The game reaches a 7-men endgame table base draw on move 108. For the purpose of weakly solving Chess it ends there on move 108.
ICCF WC Finals games lasted between 13 and 119 moves, 42 moves average, with standard deviation 16. For the purpose of weakly solving Chess all lines end around 119 moves and most before 42 moves.
@8926
"it would have been just as easy to believe engines were near perfect when they were a few hundred points weaker. i.e. very weak." ++ For the same time/move they erred more than present ones and future ones will err less. To weakly solve Chess it is not necessary to have a perfect engine, it only has to get enough time to calculate until the 7-men endgame table base.
A weak engine can also weakly solve Chess, but it will take more time.
"world championship matches had a lot of draws" ++ Matches have different dynamics from tournaments. In a match only 2 entities compete. One strategy e.g. in Karpov - Kasparov was to physically wear down the opponent with many consecutive draws. Another strategy in e.g. Spassky - Fischer was to deliberately play subpar openings to dodge preparation. Tournaments provide a better measure as more entities compete and more games are produced for statistics.
In Zürich 1953 there were more decisive games than in e.g. Madrid Candidates'.
That reflects the lower error rate now.
Feb 15, 2023
0
#8569
An engine that can calculate to a tablebase AGAINST ALL DEFENSES can be known (rather than guessed) to be perfect.
Such an engine does not exist, nor will such an engine exist in the forseeable future.
i know
@8925
I have looked at the KRPP vs. KRP endgame. It is not a draw, but a win, albeit with more than 50 moves without capture or pawn move. This position could be claimed as a table base win in ICCF. Table base win claims never happen in ICCF, as opposed to table base draw claims, that are frequent. So this position cannot arise from the initial position by optimal play from both sides. It is thus not relevant to weakly solving Chess: only drawn positions are relevant. Please come up with a drawn KRPP vs. KRP.
Stockfish plays it table base perfectly until it makes a mistake leading to a draw 25 Ke4?.
@8929
"An engine that can calculate to a tablebase AGAINST ALL DEFENSES can be known (rather than guessed) to be perfect. Such an engine does not exist"
++ Any existing engine can do that if given enough time.
Incorporating game knowledge is per van den Herik beneficial in speeding up the process.
It is not necessary to calculate 1 e4 e5 2 Ba6? to a win for black in all lines, we know 2 Ba6? is no optimal play. We can also use logic. 1 Na3 cannot be better than 1 Nf3. So once a draw in all lines against 1 Nf3 is found it is trivial to do the same for 1 Na3.
1 a4 cannot be better than 1 e4 or 1 d4. So once a draw in all lines against 1 e4 and 1 d4 is found, it is trivial to do the same for 1 a4.
pit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finleypit finle
pit finley
------------------------------------------------------------
FINDING NEMO Transcript v1.0
Copyright 2003 Walt Disney Pictures, Pixar Animation Studios
------------------------------------------------------------
Transcribed by BaD_BURN
email : markgonzalez154@hotmail.com
------------------------------------------------------------------
| Okay, this is the work-in-progress FINDING NEMO film transcript. |
| Why is it 'work-in-progress' you might ask? Well for one, this |
| isn't a 100% accurate transcript: some words might be missing, |
| may not be right. Second, some lines may or may not have been |
| spoken by the right character. There are instances in the film |
| where a line is spoken but the character isn't on screen, which |
| makes things complicated. But I'd say this transcript is about |
| 98-99% accurate. Dialogue for each scene is seperated by a line |
| of equal signs (=). |
| |
| This transcript is open for corrections, additions if you have |
| any. What you CAN'T do, however, is to edit it and take credit |
| for it. Although I do not own the movie or it's screenplay, this |
| transcript was made with no intention of copyright infringement |
| and the like. Enjoy. And remember: 'Fish are friends, not food'. |
------------------------------------------------------------------
======================================================================================
MARLIN
Wow.
CORAL
Mmm.
MARLIN
Wow.
CORAL
Mmm-hmm.
MARLIN
Wow.
CORAL
Yes, Marlin. No, I see it. It's beautiful.
MARLIN
So, Coral, when you said you wanted an ocean view, you didn't think that we we're gonna
get the whole ocean, did you? Huh? [sighs] Oh yeah. A fish can breath out here. Did your
man deliver or did he deliver?
1
CORAL
My man delivered.
MARLIN
And it wasn't so easy.
CORAL
Because a lot of other clownfish had their eyes on this place.
MARLIN
You better believe they did--every single one of them.
CORAL
Mm-hmm. You did good. And the neighborhood is awesome.
MARLIN
So, you do like it, don't you?
CORAL
No, no. I do, I do. I really do like it. But Marlin, I know that the drop off is desirable
with the great schools and the amazing view and all, but do we really need so much space?
MARLIN
Coral, honey, these are our kids we're talking about. They deserve the best. Look, look,
look. They'll wake up, poke their little heads out and they'll see a whale! See, right by
their bedroom window.
CORAL
Shhh, you're gonna wake the kids.
MARLIN
Oh, right. Right.
CORAL
Aww, look. They're dreaming. We still have to name them.
MARLIN
You wanna name all of 'em, right now? All right, we'll name this half Marlin Jr. and then
this half Coral Jr. Okay, we're done.
CORAL
I like Nemo.
MARLIN
Nemo? Well, we'll name one Nemo but I'd like most of them to be Marlin Jr.
CORAL
Just think that in a couple of days, we're gonna be parents!
MARLIN
Yeah. What if they don't like me?
CORAL
Marlin.
MARLIN
No, really.
CORAL
There's over 400 eggs. Odds are, one of them is bound to like you.
CORAL
What?
MARLIN
You remember how we met?
CORAL
Well, I try not to.
MARLIN
Well, I remember. 'Excuse me, miss, can you check and see if there's a hook in my lip?'
CORAL
Marlin!
MARLIN
2
'Well, you gotta look a little closer because it's wiggling'.
CORAL
Get away!
MARLIN
Here he is. Cutie's here! Where did everybody go?
MARLIN
[gasps] Coral, get inside the house, Coral. No, Coral, don't. They'll be fine. Just get
inside, you, right now.
MARLIN
No!
MARLIN
Coral! Coral?
MARLIN
Coral? Oh!
MARLIN
Ohh. There, there, there. It's okay, daddy's here. Daddy's got you. I promise, I will
never let anything happen to you...Nemo.
======================================================================================
NEMO
First day of school! First day of school! Wake up, wake up! C'mon, first day of school!
MARLIN
I don't wanna go to school. Five more minutes.
NEMO
Not you, dad. Me!
MARLIN
Okay...huh?
NEMO
Get up, get up! It's time for school! It's time for school! It's time for school!
It's time for school! Oh boy! Oh boy!
MARLIN
All right, I'm up.
NEMO
Oh boy--whoa!
MARLIN
Nemo!
NEMO
First day of school!
MARLIN
[gasps] Nemo, don't move! Don't move! You'll never get out of there yourself. I'll do it.
All right, where's the break? You feel a break?
NEMO
No.
MARLIN
Sometimes you can't tell 'cause fluid is rushing to the area. Now, any rushing fluids?
NEMO
No.
MARLIN
Are you woozy?
NEMO
No.
MARLIN
How many stripes do I have?
3
NEMO
I'm fine.
MARLIN
Answer the stripe question!
NEMO
Three.
MARLIN
No! See, something's wrong with you. I have one, two, three--that's all I have? Oh,
you're okay. How's the lucky fin?
NEMO
Lucky.
MARLIN
Let's see.
MARLIN
Are you sure you wanna go to school this year? 'Cause there's no problem if you don't.
You can wait 5 or 6 years.
NEMO
Come on, dad. It's time for school.
MARLIN
Ah-ah-ah! Forgot to brush.
NEMO
Ohh...
MARLIN
Do you want this anemone to sting you?
NEMO
Yes.
MARLIN
Brush.
NEMO
Okay, I'm done.
MARLIN
You missed a spot.
NEMO
Where?
MARLIN
There. Ha ha! Right there. And here and here and here!
======================================================================================
MARLIN
All right, we're excited. First day of school, here we go. We're ready to learn to get
some knowledge. Now, what's the one thing we have to remember about the ocean?
NEMO
It's not safe.
MARLIN
That's my boy. So, first we check to see that the coast is clear. We go out and back in.
And then we go out, and back in. And then one more time--out and back in. And sometimes,
if you wanna do it four times--
NEMO
Dad..
MARLIN
All right. Come on, boy.
NEMO
Dad, maybe while I'm at school, I'll see a shark!
MARLIN
4
I highly doubt that.
NEMO
Have you ever met a shark?
MARLIN
No, and I don't plan to.
NEMO
How old are sea turtles?
MARLIN
Sea turtles? I don't know.
NEMO
Sandy Plankton from next door, he said that sea turtles, said that they live to be about
a hundred years old!
MARLIN
Well, you know what, if I ever meet a sea turtle, I'll ask him. After I'm done talking
to the shark, okay? Whoa, whoa, whoa! Hold on, hold on, wait to cross. Hold my fin,
hold my fin.
NEMO
Dad, you're not gonna freak out like you did at the petting zoo, are you?
MARLIN
Hey, that snail was about to charge. Hmm, I wonder where we're supposed to go.
FISH KIDS
Bye, mom!
FISH MOM
I'll pick you up after school.
CRAB KID
Come on, you guys. Stop it! Give it back!
MARLIN
Come on, we'll try over there.
MARLIN
Excuse me, is this where we meet his teacher?
BOB
Well, look who's out of the anemone.
MARLIN
Yes. Shocking, I know.
BOB
Marty, right?
MARLIN
Marlin.
BOB
Bob.
TED
Ted.
BILL
Bill. Hey, you're a clownfish. You're funny, right? Hey, tell us a joke.
BOB/TED
Yeah, yeah. Come on, give us a funny one.
MARLIN
Well, actually, that's a common misconception. Clownfish are no funnier than any
other fish.
BILL
Aw, come on, clownie.
TED
Yeah, do something funny.
5
BOB
Yeah!
MARLIN
All right, I know one joke. Um, there's a mollusk, see? And he walks up to a sea, well he
doesn't walk up, he swims up. Well, actually the mollusk isn't moving. He's in one place
and then the sea cucumber, well they--I mixed up. There was a mollusk and a sea cucumber.
None of them were walking, so forget that I--
BOB
Sheldon! Get out of Mr. Johansenn's yard, now!
KIDS
Whoa!
MR. JOHANSSEN
All right, you kids! Ooh! Uuh, where'd you go? Where'd you go? Where, where'd you go?
NEMO
Dad, dad...can I go play too? Can I?
MARLIN
I would feel better if you go play over on the sponge beds.
MARLIN
That's where I would play
PEARL
What's wrong with his fin?
TAD
He looks funny!
SHELDON
Ow! Hey, what'd I do? What'd I do?
BOB
Be nice. It's his first time at school.
MARLIN
He was born with it, kids. We call it his lucky fin.
NEMO
Dad.
PEARL
See this tentacle? It's actually shorter than all my other tentacles but you can't really
tell.Especially when I twirl them like this.
SHELDON
I'm H2O-intolerant. [sneezes]
TAD
I'm obnoxious.
MR. RAY
[singing] Oooh, let's name the zones, the zones, the zones. Let's name the zones of the
open sea.
KIDS
Mr. Ray!
SHELDON
Come on, Nemo.
MARLIN
Whoa, you better stay with me.
MR. RAY
[singing]..mesopolagic, bathyal, abyssalpelagic. All the rest are too deep for you and
me to see.
MR. RAY
Huh, I wonder where my class has gone?
KIDS
6
We're under here!
MR. RAY
fsvbv
Feb 15, 2023
0
#8576
tygxc wrote:
@8925
"Relevant has to be relevant to something. To what?" ++ To weakly solving Chess.
Thank you. Not relevant to what you propose then.
"I usually try to win as Black" ++ 'weakly solved means that for the initial position a strategy has been determined to achieve the game-theoretic value against any opposition' Hence we look for a strategy for black to draw against all white opposition.
Assuming your big red telephone has decided the starting position is a draw then you should be looking for a strategy for White to draw against any play by Black as well. One of the flaws in the definition you quote, but it's immaterial because you don't intend to take any notice of the definition.
Those of us without a big red telephone would also consider a strategy for one side that wins with reasonable resources (including time available to move).
"[reinserted; By all means ]look at my data" ++ There is something wrong with your data.
Without looking at it. Typical of you to decide if the facts don't fit your theory then the facts are at fault.
"Here is a 269 move game Ivan Nikolic vs. Goran Arsovic (538 ply)"
++ I know that game. They repeated some positions twice. There were many forced moves. The game reaches a 7-men endgame table base draw on move 108. For the purpose of weakly solving Chess it ends there on move 108.
You said, "++ Chess is not 1000 ply deep. If there are w non-transposing choices per move, then all white moves and all black moves d moves deep lead to w^(2d) positions. There can be no more positions than there are legal positions, so w^(2d) < 10^44."
Not directly related to solving chess, but the rest of your response seems to be some stumbling steps toward explaining your error (it is rather a silly error to make). Progress?
ICCF WC Finals games lasted between 13 and 119 moves, 42 moves average, with standard deviation 16.
Which of course has no relevance to anything much.
For the purpose of weakly solving Chess all lines end around 119 moves and most before 42 moves.
'Fraid not. We seem to be going backwards again.
Feb 15, 2023
0
#8577
tygxc wrote:
@8925
I have looked at the KRPP vs. KRP endgame. It is not a draw, but a win, albeit with more than 50 moves without capture or pawn move.
That's a draw with the 50 move rule in force as I believe you yourself have pointed out more than once earlier in the thread.
This position could be claimed as a table base win in ICCF. Table base win claims never happen in ICCF, as opposed to table base draw claims, that are frequent. So this position cannot arise from the initial position by optimal play from both sides. It is thus not relevant to weakly solving Chess: only drawn positions are relevant.
Don't be ridiculous! You're using SF15. It wanders between win and draw every time it makes a half point blunder. It's conceivable it may be irrelevant to weakly solved chess (nobody knows) but not weakly solving chess using SF15 for forward searches (nor weakly not solving chess in that case).
Please come up with a drawn KRPP vs. KRP.
I did. If you want any more do them yourself.
Stockfish plays it table base perfectly until it makes a mistake leading to a draw 25 Ke4?.
Plays what tablebase perfectly? And what tablebase?
There are 11 games with 17 blunders altogether (compared to your supposed total expectation of .00861).
Naturally the moves between blunders will be perfect.
tygxc wrote:
@8926
"it would have been just as easy to believe engines were near perfect when they were a few hundred points weaker. i.e. very weak." ++ For the same time/move they erred more than present ones and future ones will err less. To weakly solve Chess it is not necessary to have a perfect engine, it only has to get enough time to calculate until the 7-men endgame table base.
A weak engine can also weakly solve Chess, but it will take more time.
That is not possible. It isn't a case of time. Enough time does not exist.
"world championship matches had a lot of draws" ++ Matches have different dynamics from tournaments. In a match only 2 entities compete. One strategy e.g. in Karpov - Kasparov was to physically wear down the opponent with many consecutive draws. Another strategy in e.g. Spassky - Fischer was to deliberately play subpar openings to dodge preparation. Tournaments provide a better measure as more entities compete and more games are produced for statistics.
In Zürich 1953 there were more decisive games than in e.g. Madrid Candidates'.
That reflects the lower error rate now.
Oh, and games relevant to solving chess are fundamentally different from those between humans in matches and tournaments. Therefore this is not a consideration. It is not relevant to solving.
You do not understand "weakly solving" and that exponentially more lines must be examined than you believe. All your "solution" does is to provide a guess, which is not what people here wish to call "solving".
did i ask
@8896
"Nor if your weakly solving involves hopping from position to position with SF15 it doesn't."
++ It does. At 17 s/move on a 10^9 nodes/s it hops from draw to draw as becomes clear once it reaches the 7-men endgame table base draw or a prior 3-fold repetition. If it would occasionally err, as it is expected to do once in 10^5 positions, then that will show by not reaching the 7-men endgame table base draw and can be easily repaired by retracting that move.
If you want to emulate 17 s/move on a 10^9 nodes/s cloud engine using a 10^6 nodes/s desktop, then you need 4.7 h/move.
If you want to reach a win in N moves, then you need time to calculate at 2N depth.
Hopping from draw to draw is easier as it does not require looking ahead to 2N.
"it is a legal chess position that can be calculated like any other position."
++ Yes, it is a legal chess position. Yes, it can be calculated. No, it is not relevant to weakly solving Chess. What you present is a set of move sequences from a won position. No, it is not a sufficiently large, sufficiently strong tournament, where the Poisson distribution applies.