He wants to learn about American Checkers

General Discussion about the game of Checkers.

Re: He wants to learn about American Checkers

Postby Chexhero on Wed Nov 04, 2015 1:26 pm

Well my point is that no matter where the opposition moves, the program will always be able to counter with moves that will lead to a clean database draw. It doesn't have to choose opponent moves. If there was some way the opposition could force a crazy fortress scenario or seesaw situation then I might agree with you. But the proof is actually saying, any move you make, our program will make a counter move which will force you in a database draw. Now, I am no program wiz like yourself, but I would highly doubt there is a way for either color to somehow force a fortress or seesaw draw considering how rare they are. I can't prove it, but apparently they say their program did.

It personally doesn't matter to me what their conclusions are. It doesn't affect out game or make it any less worthy. But then again it is a reason that people use to dismiss the game, which is why I suggested that you and others could challenge Schaeffer on this notion.
Chexhero
 
Posts: 561
Joined: Sat Jan 08, 2011 10:11 pm
Location: PA

Re: He wants to learn about American Checkers

Postby chipschap on Wed Nov 04, 2015 5:03 pm

Did Dr. Schaeffer actually claim to have "solved" checkers? I don't recall such a claim; he has claimed and demonstrated that GAYP is a draw, but that is very definitely not the same thing (see Richard Pask's recent discussion of this).

Sam Sloan is an interesting and very controversial character, fairly well known in Chinese Chess (Xiang Qi) circles as a higher-level American player. Whatever you think of him personally, Ishi Press has published numerous worthwhile books. I too would welcome a new print edition of Reinfeld.

Before releasing my PDF version (most of which is thanks to Mel Tungate) I got permission from Mr. Reinfeld's heirs, Donald and Judith Reinfeld. Mr. Sloan ought to do likewise.
chipschap
 
Posts: 123
Joined: Sun Nov 27, 2005 12:54 pm
Location: Honolulu, Hawai'i

Re: He wants to learn about American Checkers

Postby Chexhero on Wed Nov 04, 2015 5:46 pm

Bob,

Here is the link to the Chinook website where you can read about checkers being solved. https://webdocs.cs.ualberta.ca/~chinook/project/. Also be sure to check out the "News" section to watch the video and listen to the podcast. Very interesting!
Chexhero
 
Posts: 561
Joined: Sat Jan 08, 2011 10:11 pm
Location: PA

Re: He wants to learn about American Checkers

Postby George Hay on Thu Nov 05, 2015 2:41 pm

Good question, Bob! "Did Dr. Schaeffer actually claim to have "solved" checkers?" In the link in the just above post by Joe (Chexhero) the answer is yes!
Under Milestones the 2007 entry is: 2007: Checkers is solved. Perfect play leads to a draw.

Bob, you wrote to "see Richard Pask's recent discussion of this." Where would we find Pask's recent discussion of checkers being "solved" by Chinook?

--George Hay
George Hay
 
Posts: 806
Joined: Sat Jan 01, 2011 7:41 am
Location: Grand Rapids, Michigan, USA

Re: He wants to learn about American Checkers

Postby liam stephens on Thu Nov 05, 2015 9:38 pm

The article on the Chinook website (referred to above) states that "Checkers is now solved", but makes no reference to the level achieved whether Ultra-Weak, Weak, or Strong.
In Wikipedia, there is the following statement:
"Draughts, English (Checkers)
This 8×8 variant of draughts was weakly solved on April 29, 2007 by the team of Jonathan Schaeffer, known for Chinook... From the standard starting position, both players can guarantee a draw with perfect play."

This is confirmed in an article in SPECTRUM July 2007
"Schaeffer’s proof solved checkers for 19 different openings, all of which end in draws...
...Solving checkers has taken a big monkey off Schaefer’s back. The fact that the game wasn’t solved for every possible position and then tucked away in a database doesn’t seem to bother him. 'Well, the checkers players would love it, because [then] you’ve got this oracle that can tell them everything—answer every single unanswered question in the game of checkers,' he says. 'But first of all, I don’t have the patience to do it. And second of all, I don’t have the technology to do it.' Even with the best data-compression techniques, Schaeffer says, the amount of storage required to solve all possible positions of checkers would exceed even the capacity of the world’s biggest supercomputers with tens of petabytes of storage by an order of magnitude."
=======================================================================================================================================================
So, the game has only been partially solved.
If all that has been proved is that with perfect play the game is a draw, then as pointed out in a previous posting above,
that has been known, at least empirically, for 100 years or more.
To prove it so, is hardly a world shattering event (unlike walking on the moon, for example).
As an analogy, in this day and age is there a necessity for a scientific proof that the earth is not flat ? :D
Incidentally, I believe the “unanswered question” was answered in 1908 by Charles Ives!!
Last edited by liam stephens on Sat Nov 07, 2015 11:08 am, edited 1 time in total.
liam stephens
 
Posts: 912
Joined: Sun Nov 27, 2005 2:56 pm
Location: Ireland

Re: He wants to learn about American Checkers

Postby Ingo_Zachos on Fri Nov 06, 2015 7:43 am

Checkers is solved for positions up to 10 pieces. In chess, the postions for up to 7 pieces have been solved, as far as I know.

There is no database for 24 pieces, so the game checkers as a whole is not yet solved. Nine Men Morris or Four In A Row have already been solved, but they are far less complex.

Mr. Schaeffer has achieved something less sophisticated. He has proved that the game played in GAYP is a draw and that he can produce a device that can not fail in reaching at least this draw from the starting position.

That is called a "weak" solution, but it is not even half of the pudding, so to speak. The prove of the pudding is still incomplete, in fact only a fraction of the pudding has been eaten so far.

It is like having a map of a country.

If you know the way to the capital and back, you still did not mapped the whole country.

Mr. Schaeffer found a route to stay on known paths to savely cross the undiscoverd continent of checkers, but he still has not yet drawn a map of the whole continent. In fact most of it, more then 90 percent of all possible postions, are not yet solved.

You just need no map of the whole continent to cross it. To know one save path is enough for that. So, to know the way through an undiscovered land is a weak solution, no mean achievement (especially for a discoverer) , but a strong solution requires much, much more: you have to draw the whole map.

I hope this clears the matter.

And yes, programs calculate stronger then grandmasters, even stronger then a World Champion, like in chess and many other games. And once they are in a database win or draw, opening database or endgame database, they do not fail.

But it is like in other sports: can u run faster then a car or lift more then a crane? It does not mean that the sport is at its end, because a machine can do it better then a human.
It does however mean that you can cheat using a machine. Like in the 2nd Tour de France already the first four used a train to ride much faster then their opponents...

Greetinx from cloudy Dortmund, Germany

Ingo Zachos
You can rent this space for advertising, if you like!
Ingo_Zachos
 
Posts: 1269
Joined: Sun Nov 13, 2005 7:41 am
Location: Dortmund, Germany

Game On!

Postby Richard Pask on Sat Nov 07, 2015 3:41 am

In another section of this forum there is a reference to an interesting article in The Economist entitled 'Game Over': it is related to this discussion.

Here is the essence of the letter I wrote to the editor:

The computer programs which have been developed in recent years have, far from destroying the game, been a boon to serious players; particularly as they provide an ever-willing grandmaster sparring partner. I, for one, am certainly not 'angry' with the programmers, but very grateful to them and now even more aware of how deep and profound our great game is.

Since 1756 it's been believed by all parties that the game is fair - a theoretical draw - and, with the development of 3-move, a condition for inclusion of a ballot is that it is sound. The conclusion that GAYP is a theoretical draw is 100% expected and guarantees the game's eternal soundness. (A forced win would have killed it off!)

For me, the confusion arises because the Chinook team's two goals: proving themselves superior to MFT and solving the game are somewhat muddled. Had they determined to solve the 3-Move version, the 2 goals would indeed be hierarchical. As it is, Chinook has 'solved' GAYP in terms of the definition very carefully explained in Dr Schaffer's book. However, in practical terms, astonishing achievement though it is, it's a very lame oracle. And of course, against the best 'version' of MFT it could lose a 3-move match. (According to its latest website Chinook is unbeatable - my preferred word - at just 28 of the 156 sound ballots.) According to work by Gill Dodgen and Ed Trice, there are also winning database positions that Chinook, not possessing the perfect-play databases, might not bring to fruition. Finally, against a genuine oracle - 24-piece perfect-play databases - in a match of indefinite length on the 3-move restriction, Chinook, [i]as I understand it, would be certain to lose. [END OF LETTER]

There is a danger, of course, that all of the above sounds like sour grapes towards Dr Schaeffer and his team, so let me clarify things:

The Chinook team's achievement was incredible and reflected enormous skill and dedication on the part of the whole team; Dr Schaffer was always extremely kind and generous in his dealings with me.
Ignoring the technicalities of the word 'solved', Chinook, and other programs, play at a near-perfect level, which is above that which any human can aspire to.
The current computer programs are a great boon to serious players (indeed, players of all standards).
I don't think Chinook's achievements have had, or will have, any negative effect upon the general public's perceptions of the game: they thought it was child's play before, and are simply stunned that so much effort had to be put in to 'solve' it! Here, I feel very sorry for the Chinook team: they will never get the credit they deserve from the general public, because they don't think they've done anything clever!

Finally, the competitive game will never die, but it's currently in a very poor state. One day, when it rejuvenates, I hope that my 21st Century Checkers series for example (thanks again Bob Newell for your great work) will prove useful to serious players; along with the Parragon Draughts/Checkers book which has been designed to be virtually timeless.

The problem the fraternity face is the eternal one: overcoming the game's image, seemingly set in stone because of its unadorned appearance, that it's not worthy of serious study.
Richard Pask
 
Posts: 129
Joined: Wed Nov 17, 2010 3:15 pm

Re: He wants to learn about American Checkers

Postby liam stephens on Sat Nov 07, 2015 7:29 pm

Well done Richard ! – a fine and informative response to the article in the Economist.
Mention of “the oracle” brings to mind a heartfelt cry from Jocasta in the ancient Theban plays.
“Cave oracula quae semper mentiantur.”
(Beware of the oracles which always lie.)

I have recently being reading the 11-16s in the 21st Century Checkers series.Excellent as ever.
A big thank you to Richard and Bob for providing such a wonderful service to the Checkers community, and all for free.
liam stephens
 
Posts: 912
Joined: Sun Nov 27, 2005 2:56 pm
Location: Ireland

Previous

Return to General Discussion

Who is online

Users browsing this forum: clintolsen and 2 guests

class=