question about a game

[Patrick Parker]

1. 10-15 21-17 2. 11-16 17-13 3. 16-20 23-18 4. 8-11 26-23 5. 7-10 25-21 6. 9-14 18x9 7. 5x14 29-25 8. 4-8 30-26 9. 11-16 24-19 10. 15x24 28x19 11. 3-7 22-18 12. 1-5 18x9 13. 5x14 *

i am curious about the 1-5 move
does it lose?

[Patrick Parker]

or 22 18 before it lose

[Macb]

Hello Patrick,

1-5 is okay. after 1-5 then 18-0 5-14 19-15 10-19 25-22 8-11 32-28 19-24 28-19 6-10 22-18 11-15 18-11 14-18 23-14 16-32 11-8 10-17 21-14 7-10 14-7 2-11 Drawn. Hope this helps.
Mac Banks

[Macb]

Hello Patrick,

I made typo. after 1-5 then it is 18-9. Sorry.

All the best,
Mac

[Patrick Parker]

19-15 kingsrow gives a question mark to and gives 26-22 a -100
is what i am trying to figure out 19 15 is in the book i have
and once i play it. it says it draws

[Patrick Parker]

just dont understand this odd evaluation looking about 37 moves ahead

[Macb]

Hello Patrick, ok I will go 26-22 instead of 19-15 then 14-17 21-14 10-26 31-22 7-10 22-18 8-11 32-28 6-9 13-6 2-9 25-21 9-13 18-14 10-17 21-14 13-17 14-9 17-22 9-6 22-26 19-15 11-18 23-14 26-31 28-24 16-19 Drawn.
1-5 is solid draw for red. Patrick.

All the best,
Mac

[Patrick Parker]

maybe this is what we were talking about in another post
a problem with the heuristics

1. 10-15 21-17 2. 11-16 17-13 3. 16-20 23-18 4. 8-11 26-23 5. 7-10 25-21 6. 9-14 18x9 7. 5x14 29-25 8. 4-8 30-26 9. 11-16 24-19 10. 15x24 28x19 11. 3-7 22-18 12. 1-5 18x9 13. 5x14 26-22 14. 14-17 21x14 15. 10x26 31x22 16. 7-10 22-18 17. 8-11 32-28 18. 6-9 13x6 19. 2x9
25-22
still gives lopsided eval.....who that be whats wrong?

[Macb]

Hello Patrick,

Here is the play after your 25-22 instead of my 25-21 , red then goes 9-13 28-24 11-15 18-11 10-14 11-8 14-17 22-18 17-22 18-15 Drawn. White has 6 checkers against red's 5 checkers and both can get two kings but red has 4 white checkers tied up with 3 so there is no attack for white to win this game. Patrick, 1-5 is a safe draw. I don't know how much plainly I can tell you that it will draw. Look at the board without your computer, what are you afraid of if you had the red side with one piece short. There is no attack. I don't know what program you are using but I doubt any of the major programs out there will say that 1-5 will not draw. I think Kingsrow or Cake++ will agree that 1-5 draws. Sometimes when you look at the board a long time , one can see something that is not there. I have done this before myself. By the way Patrick, are you kin to Bill Parker out of Mississippi? I used to live in La. myself (New Orleans) and we had lots of checker players. Now , I know only of Albert Tucker from Minden, La. and Harvey Shaffer from Greenwell Springs and you. Patrick, I have loads of play on all the openings and you can always email me at mbanks36@hotmail.com and I will gladly give you any play that I may have on the openings as I have compiled a manuscript that has played developed over the past 45 years.

All the best,
Mac

[Ed Gilbert]

Hi Patrick,

Yes the computers have trouble with this line. I tried it with cake and wcc and they all have trouble identifying this as a draw from many moves back. They all make the correct moves to draw, which is the most important thing, but it is hard for them to see far enough ahead to see an equal position. The problem is the 4x3 cramp. Of course this is a well known cramp, and all the programs have evaluations to recognize it, but they can't just always assume that the side that is a man down will eventually even things up. Sometimes there are ways to squirm out of this cramp without giving up a man, and if the evaluation always assumed that the cramp was worth the full value of a man then the programs would eagerly pitch a man every time to assume the cramped position, which would cause them to lose a certain percentage of those games. I think kingsrow gives this cramp the equivalent of about 1/3 of a man, and that combined with the power of the lookahead search is usually sufficient to guide it to make the correct moves.

-- Ed

[Alex_Moiseyev]

Ed, if you don't mind - let me give you an advice how to build heuristic logic for this type of positions (with 3x4 scramp).

All these positions were study and fully covered in Oldbury Encyclopedia, in total 8 or 12 different situations. All you need - enter all this positions with solutions into your book. Then you should have some logic which helps program to recognise transposition into one or another position, covered in his book - and this class of positions will be solved for KR.

DEO called this element as Parameter 27 and it is covered on Volume 5, Pg. 525, and i beleive in some other volumes also. Derek was very interested and paid alot of attention to this class of positions and his contribution in checkers theory was tremendous.

Regards,

Alex

[Ed Gilbert]

Hi Alex,

I don't mind at all, quite the opposite. I don't think opening books are the answer, because there are millions of positions with these cramps, so something more general is needed. But perhaps I can learn some general themes about these from DEO's Encyclopedia. Unfortunately I don't have this, but I will see if I can borrow it or obtain copies of the relevant pages.

-- Ed

[Patrick Parker]

thanks i figured it was something crazy like that

[Ed Gilbert]

More about this position: the main reason that programs have trouble evaluating this particular game to a draw is that this game ends in a draw by repetition, with the 4x3 cramp still intact. Usually these cramps are dissolved after a few moves, as the side with the 3 men holding the cramp will pitch for the 2x1, or else the side with the 4 men will find a way to pitch out of it. But in this case the program must see all the way past a bunch of king moves where neither side can make any progress and thus the draw by repetition. This is a very long distance from the game position back at move 14.