More 3-move openings solved by Schaeffer

[Ed Gilbert]

The chinook website has a news brief saying that 6 of the 3-move openings are now solved. They seem to be making much faster progress now. There was a 1 year gap between the first and second solutions, then a 3 month gap to the third, and now the latest 3 were all finished a few weeks apart in July. Link to chinook website.

-- Ed

[Ingo_Zachos]

lol!

I believe Dr. Schaeffer did not "solve" this opening. He took an electonic device to do all calculations and to keep the lines in memory.
It is just like cycling the Tour de France with a motor bike.
And that they r proven to be drawish is no great thing, just like proving that a circle is round.
Wo is really surprized ?

Greetinx from rainy old Europe,

Ingo Zachos

[liam stephens]

Yes Ingo, I think that time and experience has shown that all of the 156 allowed 3 move openings are sound for the draw, given perfect play. To prove it so, comes as no great surprise.

If you have a free half hour or so and wish to amuse yourself, try the following: Set up the pieces on computer, back to back, i.e. black on squares 17 to 28, White on squares 5 to 16. Now set the computer on Autoplay, preferably with the timer set at 1 or 2 seconds or even instant, and see what happens. The pieces scurry all over the place like demons possessed. It's very funny to watch. Using the instant timing setting I found that Black eventually won.

Regards - Liam.

[Martin Fierz]

It is just like cycling the Tour de France with a motor bike.

Hi Ingo,

that is not true at all. It is more like building a motor bike and then competing in the Tour de France!

Of course, nobody is suprised that these openings are draws. The whole excercise is about proving that the game of checkers is a draw, and the emphasis is very clearly on proving. It is one thing to (like Fermat...) write a conjecture in the margin of a book, but it is another thing to actually prove it. It is of no practical interest to the checker player, but it is of interest to computer scientists.

cheers
Martin

[Ed Gilbert]

I believe Dr. Schaeffer did not "solve" this opening. He took an electonic device to do all calculations and to keep the lines in memory.
It is just like cycling the Tour de France with a motor bike.
And that they r proven to be drawish is no great thing, just like proving that a circle is round.
Wo is really surprized ?

It's probably true that this is rather ho-hum news to most checkers players. But it is an achievement in computer science, and some of the techniques that were used to prove these ballots drawn may have other useful applications. For example, I know that they had to solve the 'graph history interaction' problem in order to do this properly. This is a problem that checkers programs and opening book generators have in some form or other, and I would be interested to see his solution to this problem.

-- Ed

[Ingo_Zachos]

Hi Martin,

...
that is not true at all. It is more like building a motor bike and then competing in the Tour de France!
...

The point is: You r not allowed to use a motor bike in the Tour de France, even if you built it...

And I am not allowed to use a database if I play in a match or tournament.
So the practical influence of it is close to zero, and
I still doubt that Dr. Schaeffer would draw any of these lines against a master player, cause he is not allowed to use the database.

That was all I wanted to say.

Greetinx from rainy Germany,

Ingo Zachos

[mday]

Hi Martin,

...
that is not true at all. It is more like building a motor bike and then competing in the Tour de France!
...

The point is: You r not allowed to use a motor bike in the Tour de France, even if you built it...

And I am not allowed to use a database if I play in a match or tournament.
So the practical influence of it is close to zero, and
I still doubt that Dr. Schaeffer would draw any of these lines against a master player, cause he is not allowed to use the database.

That was all I wanted to say.

Greetinx from rainy Germany,

Ingo Zachos

And you wouldn't be able to draw them vs. a Master either.

[Ingo_Zachos]

And you wouldn't be able to draw them vs. a Master either.

Well, I tried, Michael, and drew and beat masters . Did you do the same ?

[Ryan Pronk]

Dr. Schaeffer is to be commended for all the work he has done for this game. As a checkers player, I am fascinated with all work Chinook has done thus far on the openings it has solved. My hope is that Chinook will "add" more openings to the 156 deck - even if the draw is incredibly microscopic.

I've kept relatively quiet on the topic of checkers programs (mostly because I do not understand how any of them work). However, I have concluded (since I do talk with other checkers players about this), that IF Chinook were a "commercial" program, it would be the superior (and the favorite) program by far.

Don't get me wrong, Mr. Fierz and Mr. Gilbert have contributed tremendous amounts of years to making great programs, and we should all be grateful for that. But I also believe that without Chinook, checkers programs would not be where they are today.

--Ryan

[Gene Lindsay]

Ryan, yes if it where not for Dr. Schaeffer computer checkers would not be where it is today. And at one time Chinook was probaby the best checkers program. But, I will put my money on cake or kingsrow ( either one u pick) against Chinook on equal computers with equal ram with equal endgame databases either 8 or 10. Chinook uses a brute force approach to find the best move(in other words it runs each move all the way out) whereas cake and kingsrow( as does WCC and Nemisis) use a pruning tree to discard some moves that it can see are obviously bad like giving away 2 checkers with no positional advantage. Therefore cake and kingsrow will be able to seach deeper into a position than Chinook in the same amount of time. In my opinion right now Cake and Kingsrow are the 2 best checker programs in the world. Ed and Martin if I misstated anything please correct me.

Gene Lindsay

[matthewkooshad]

Thanks for the link, Ed.

[EdTrice]

People forget that before Chinook came around, all previous checkers programs (in the United States) that competed were hardly capable of sustaining their own in the minors section of play. While Chinook did not have access to strong checkers players to craft its evaluation function, it did do something that ALL subsequent checkers programs have copied:

It demonstrated that run-time probing of pre-computed endgame databases could positively impact the play of embryonic positions that were many, many moves away from these positions.

This had never been done before, and it was the quintessential contribution of the Chinook team. True, a great deal of the "invincible" aspect of Chinook began when Martin Bryant's opening book was acquired (and then subsequently improved upon by having the 8-piece database review some of the 3-move ballots), but the groundbreaking achievement of the endgame databases cannot be underestimated.

As an analogy:

The history of mankind's attempt at flight was dotted with poor attempts to construct machines that would flap mechanical wings like a bird. Then the Wright Brothers demonstrated that a lightweight, high-horsepower engine coupled with a low-drag propeller could produce enough forward thrust to lift the weight of the pilot plus the craft.

Chinook was like the successful Kitty Hawk plane, its predecessors were the flapping contraptions carving out inverted "U's" in the sky before crashing.

All pre-jet engine aircraft that were built next used this metaphor to propel passengers all over the world.

Schaeffer's solution was "copied" in much the same way, although by programs with very different approaches to the game (Colossus, WCC, Wyllie, Nemesis, Cake, and Kingsrow.)

You have to give the pioneer credit for "paving the way."

[mday]

And you wouldn't be able to draw them vs. a Master either.

Well, I tried, Michael, and drew and beat masters . Did you do the same ?

No you didn't. Yes I have.

[Gene Lindsay]

Ed, By the way I have WCC Plat. and when it came out I think it was the best at that time. Your copy of WCC that you have continued improving maybe as good as cake & kingsrow, you are probably the only one who knows. The last time I talked to Gil he told me he would not work on WCC again. I still use WCC along with cake & kingsrow to analyze positions, but nemisis is to slow for me.

Gene Lindsay

[Martin Fierz]

Hi Ryan,

However, I have concluded (since I do talk with other checkers players about this), that IF Chinook were a "commercial" program, it would be the superior (and the favorite) program by far.

this is totally wrong. I have no idea what you - and other checkers players - do to conclude this. Perhaps you can explain? (I think Gene's version is much closer to the truth)

But I also believe that without Chinook, checkers programs would not be where they are today.

And this is completely right. Ed Trice has already explained it very well above. BTW, I'm a chess snob and don't play checkers myself. The only reason that I wrote a checkers program is a short article in the Time magazine on the Chinook-Tinsley match. AFAIK, Ed Gilbert dusted off a 20-year old checkers program after finding my interface (CheckerBoard) on the internet - game programmers usually don't want to bother with writing interfaces, and so perhaps he wouldn't have produced Kingsrow if it had not been for CheckerBoard. So indeed, Cake and Kingsrow wouldn't be where they are today - they wouldn't even exist if Chinook hadn't played that match against Tinsley!

cheers
Martin