Dylan Sharp Vs. Harvey Williamson (G4)

Discussion of computer chess matches and engine tournaments.

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
User avatar
Ovyron
Posts: 2749
Joined: Tue Jul 03, 2007 2:30 am

Re: Dylan Sharp Vs. Harvey Williamson (G4)

Post by Ovyron » Sun Nov 10, 2019 9:57 pm

Harvey Williamson wrote:
Sun Nov 10, 2019 11:35 am
I was hoping for 18. Rc2 as I found Re6!
Yes, I think 18. Rc2 leads to a "black to move and mate in X" position :) 18. Rcc1 is the only move that still holds the draw (except that one misstep and I'm toast, let's see if I don't trip.)

We can finally show how a position where white is lost looks like:



Black plays 18...Re6! and wins! 0-1

MikeB
Posts: 3467
Joined: Thu Mar 09, 2006 5:34 am
Location: Pen Argyl, Pennsylvania

Re: Dylan Sharp Vs. Harvey Williamson (G4)

Post by MikeB » Mon Nov 11, 2019 3:19 am

Ovyron wrote:
Sun Nov 10, 2019 9:57 pm
Harvey Williamson wrote:
Sun Nov 10, 2019 11:35 am
I was hoping for 18. Rc2 as I found Re6!
Yes, I think 18. Rc2 leads to a "black to move and mate in X" position :) 18. Rcc1 is the only move that still holds the draw (except that one misstep and I'm toast, let's see if I don't trip.)

We can finally show how a position where white is lost looks like:



Black plays 18...Re6! and wins! 0-1
Bf5 also wins after Rc2. White is holding on by a string in the current position.

User avatar
Graham Banks
Posts: 33168
Joined: Sun Feb 26, 2006 9:52 am
Location: Auckland, NZ

Re: Dylan Sharp Vs. Harvey Williamson (G4)

Post by Graham Banks » Mon Nov 11, 2019 5:21 am

MikeB wrote:
Mon Nov 11, 2019 3:19 am
Ovyron wrote:
Sun Nov 10, 2019 9:57 pm
Harvey Williamson wrote:
Sun Nov 10, 2019 11:35 am
I was hoping for 18. Rc2 as I found Re6!
Yes, I think 18. Rc2 leads to a "black to move and mate in X" position :) 18. Rcc1 is the only move that still holds the draw (except that one misstep and I'm toast, let's see if I don't trip.)

We can finally show how a position where white is lost looks like:



Black plays 18...Re6! and wins! 0-1
Bf5 also wins after Rc2. White is holding on by a string in the current position.
Yeah - a G-string.
My email addresses:
gbanksnz at gmail.com
gbanksnz at yahoo.co.nz

User avatar
Ovyron
Posts: 2749
Joined: Tue Jul 03, 2007 2:30 am

Re: Dylan Sharp Vs. Harvey Williamson (G4)

Post by Ovyron » Mon Nov 11, 2019 5:48 am

MikeB wrote:
Mon Nov 11, 2019 3:19 am
Bf5 also wins after Rc2. White is holding on by a string in the current position.
I think I can defend Rc2 Bf5.

Of course it'll all be moot if Harvey beats me.

But think about it, white MUST have a draw with best play, lest Harvey wouldn't care if I played Rc2 or Rcc1, because he'd know a line that beats both. The line that beats Rcc1 doesn't exist. A line that defends against any black attack exists. But can I find it?

Uri Blass
Posts: 8594
Joined: Wed Mar 08, 2006 11:37 pm
Location: Tel-Aviv Israel

Re: Dylan Sharp Vs. Harvey Williamson (G4)

Post by Uri Blass » Mon Nov 11, 2019 6:10 am

I am afraid that the game is going to get dark again if people suggest alternative wins.

I did not give a lot of computer time for the relevant position but I did not see a forced win after 18.Rc2 not with Re6 and not with Bf5.

Note that even evaluation of +3 by stockfish I do not cosider as a forced mate because maybe there is a fortress and certainly not evaluations that are close to +2 pawns.

Uri Blass
Posts: 8594
Joined: Wed Mar 08, 2006 11:37 pm
Location: Tel-Aviv Israel

Re: Dylan Sharp Vs. Harvey Williamson (G4)

Post by Uri Blass » Mon Nov 11, 2019 6:18 am

Ovyron wrote:
Mon Nov 11, 2019 5:48 am
MikeB wrote:
Mon Nov 11, 2019 3:19 am
Bf5 also wins after Rc2. White is holding on by a string in the current position.
I think I can defend Rc2 Bf5.

Of course it'll all be moot if Harvey beats me.

But think about it, white MUST have a draw with best play, lest Harvey wouldn't care if I played Rc2 or Rcc1, because he'd know a line that beats both. The line that beats Rcc1 doesn't exist. A line that defends against any black attack exists. But can I find it?
The conclusion is wrong.
If Harvey does not see a forced win at this point of the game it does not mean that there is no forced win and it does not mean that Harvey cannot win the game regardless of what you do.

it is possible that harvey cannot be convinced that he is going to win against every line because there are too many lines but practically can find a win against every line that you will practically play.

Jouni
Posts: 2010
Joined: Wed Mar 08, 2006 7:15 pm

Re: Dylan Sharp Vs. Harvey Williamson (G4)

Post by Jouni » Mon Nov 11, 2019 7:46 am

I played yesterday some SF long games after move 18. All were 0-1! Still I am not 100% sure of black win.
Jouni

User avatar
Ovyron
Posts: 2749
Joined: Tue Jul 03, 2007 2:30 am

Re: Dylan Sharp Vs. Harvey Williamson (G4)

Post by Ovyron » Mon Nov 11, 2019 8:44 am

Uri Blass wrote:
Mon Nov 11, 2019 6:18 am
The conclusion is wrong.
You meant to say "the conclusion could be wrong", for the reasons you stated. But after I found 18. Rc2 Re6 it took me less than an hour to see black was winning, and I didn't need to solve it to a mate in X. My conclusion is only wrong if Harvey and me can't tell the difference between a lost position and one that isn't (say, the opening position is one that isn't). If we can, then on a given position, we know what move(s) save the game and which ones don't. To know that the conclusion is wrong, you'd need to know that the position is a mate in X yourself.

Some things don't need proof to be true.

Uri Blass
Posts: 8594
Joined: Wed Mar 08, 2006 11:37 pm
Location: Tel-Aviv Israel

Re: Dylan Sharp Vs. Harvey Williamson (G4)

Post by Uri Blass » Mon Nov 11, 2019 9:54 am

Ovyron wrote:
Mon Nov 11, 2019 8:44 am
Uri Blass wrote:
Mon Nov 11, 2019 6:18 am
The conclusion is wrong.
You meant to say "the conclusion could be wrong", for the reasons you stated. But after I found 18. Rc2 Re6 it took me less than an hour to see black was winning, and I didn't need to solve it to a mate in X. My conclusion is only wrong if Harvey and me can't tell the difference between a lost position and one that isn't (say, the opening position is one that isn't). If we can, then on a given position, we know what move(s) save the game and which ones don't. To know that the conclusion is wrong, you'd need to know that the position is a mate in X yourself.

Some things don't need proof to be true.
When I wrote the conclusion is wrong I meant that you cannot get conclusion B from knowledge A(not that B has to be wrong).
The position may be a draw with perfect play but you cannot get this conclusion from the fact that harvey did not find a win and there may be positions that there is a win but harvey is not sure if it is a win or a draw.

When I played correspondence chess and won games often I did not know what was the losing blunder of my opponent even when I could say some moves of the opponent were not good.

I think about moves as better or worse and not as drawing or losing because I simply cannot solve chess even with the help of engines.

Suppose you play white and win and one move of black changed your estimate for expected result from 60% draw 40% win for white to
40% draw 60% win for white.

In this case it is not clear if this is the losing move or the losing move was earlier or maybe the losing move was later.
There are of course cases when you have 90% confidence that you win after a move but even in this case I prefer to be careful if it is not very close to 100% not to say that you win.

User avatar
Ovyron
Posts: 2749
Joined: Tue Jul 03, 2007 2:30 am

Re: Dylan Sharp Vs. Harvey Williamson (G4)

Post by Ovyron » Mon Nov 11, 2019 11:44 am

Uri Blass wrote:
Mon Nov 11, 2019 9:54 am
When I wrote the conclusion is wrong I meant that you cannot get conclusion B from knowledge A(not that B has to be wrong).
The position may be a draw with perfect play but you cannot get this conclusion from the fact that harvey did not find a win and there may be positions that there is a win but harvey is not sure if it is a win or a draw.
Following wrong premises (if I did) to reach the right conclusion doesn't mean the conclusion is wrong. It's like I try to find what's 2+2 and I follow a set of directions that are unrelated, and end subtracting 986 from 990 and getting 4. You could say what I did was nonsensical and unrelated about 2+2, but if I arrive at 4, then the conclusion is right, even if I don't know if 900-986 equals 2+2.

Suppose Harvey had magic access to an oracle that tells him if a position is won for one side or a draw, and he sees Rcc1 is a draw and Rc2 was won for black, then my conclusion is right, and it doesn't matter how I arrive at it. But a magic oracle isn't needed, it'd suffice for Harvey to be able to win from any chess position that is won, and he's unable to win from this position against himself, so he'd much rather have me play the move that he knows loses, because he doesn't want me by chance to play the same drawing moves that he sees. Then my conclusion would also be right.

Therefore, if my conclusion is wrong you wouldn't be able to know if it's wrong, unless you have more information than me about it (like, you had a magic oracle and it says after Rcc1 black mates in something.) It's like person A murders person B, and I'm investigating the case. I follow red herrings and by mistake conclude that person A murdered person B. You take a look and say my conclusion is wrong because I made mistakes that nullify my logic and send person A free. But the fact was that he was the murdered, so the conclusion was right, even if with my knowledge I couldn't have proven it.
Uri Blass wrote:
Mon Nov 11, 2019 9:54 am
When I played correspondence chess and won games often I did not know what was the losing blunder of my opponent even when I could say some moves of the opponent were not good.
I can't play bullet chess, even in won positions I can't play fast enough to mate my opponents. I don't think my abilities to play bullet chess or your abilities to find the blunders you opponents played are relevant here.
Uri Blass wrote:
Mon Nov 11, 2019 9:54 am
I think about moves as better or worse and not as drawing or losing because I simply cannot solve chess even with the help of engines.
I can. The opening position is draw with perfect play. The position after Rc2 is a mate in something in favor of black. This has nothing to do with solving chess because you don't need the solution to be able to reach right conclusions about positions.
Uri Blass wrote:
Mon Nov 11, 2019 9:54 am
Suppose you play white and win and one move of black changed your estimate for expected result from 60% draw 40% win for white to
40% draw 60% win for white.

In this case it is not clear if this is the losing move or the losing move was earlier or maybe the losing move was later.
There are of course cases when you have 90% confidence that you win after a move but even in this case I prefer to be careful if it is not very close to 100% not to say that you win.
As with other people on Talkchess you need 100% proof of something to believe it. As I said in the other post, some things can be true without proof. And if I make a statement that is true, it's a right conclusion and it doesn't matter how I got there, and it doesn't need proof. If I make the statement that a position is a mate in X for one side and it is, it is a right conclusion even if nobody would be able to prove it.

Post Reply