THE FORGOTTEN GRANDMASTERS

[Ed Gilbert]

In my opinion its the programmer that makes any particular prog what it is, but consider this where the programmer gets the infomation to insert into it??

Simple answer from Mail players in most part who have worked endlessly on building the line . Consider over the last 10 yrs of the tremendous work put into opening up the unplayable openings (The Star ship Enterprise and many others)that have now been included in the deck for regular Ty play.

Do you think the draw lines were thought up my some magic formula ??? nah, to put it in plain english this play was discovered by committed Mail players (and the occasional master)thru night and day and thick and thin then passed on to programmers for insertion in a program. /

This is not how the opening books are created for either kingsrow or cake. The kingsrow opening book was generated automatically and without any published play input by a program, using several computer-years of effort. The technique is called dropout expansion. It makes very many searches using the kingsrow engine. Each search is for approximately 3 minutes, and the positions and their search scores are added to a database. There is a more detailed description of the process here: http://www.fierz.ch/strategy4.htm. The current opening books for cake and kingsrow each contain more than 1 million positions.

-- Ed

[Alex_Moiseyev]

The ratings and associated titles should be reserved for human crossboard ability

I have in my library ACF rating list from 1996, and I can see Chinook is listed here with rating 2745. Makes alot of sense because Chinook played in many ACF events and got a rating.

Tears are on my eyes. 2600-2700 zone is full of names who is not with us today. Rating list gets more empty from year to year

Alex

[Ingo_Zachos]

...Makes alot of sense because Chinook played in many ACF events and got a rating...

Alex

Veto!

Not, because the programs in questions r strong enough to deserve the title, here you r certainly right that they r at least Grandmasters...

But think twice:
The title is a reward for personal achievements.
What person is this ?
An artifical intelligence ?
Not quite, as the program is not able to feel the honour that a title represents.
So, prosaic, I want to ask :

Why does a machine need a reward?

And what's the use of it?

We could, however, give titles to the authors/programmers, but in that case I prefer a solution like in the mailplay case: to give honorary titles to them, for their achievents, not for their results, as computer play is not comparable to human play:

Does the computer look at the board with his eyes and touches and moves the pieces with his own hands ? Does it push the buttom of the clock?
Does it write down the moves and talks to the referee if a question arises or to the opponent to refuse or accept a draw offer or a cup of coffee that the opponent offers to bring to him?
As long as a machine does not behave like a human, and does not understand what a title is, it should not get a title, as this seems quite useless to me.

Greetinx from finally snow-covered Dortmund, Germany

Ingo Zachos

[Pedro Saavedra]

There have been matches in chess (one between two top players) which were computer assisted matches. Each player brough a laptop with whatever programs he chose and could consult, but within the time limit.

Now here is my question. Suppose we accepted that correspondence or e-mail play (or at least in a given event) could be computer-assisted (i.e. everybody was within his rights to consult computers, books, databases, etc.) there would still be winners and losers. Would there not be a skill still there in deciding whether to follow the computer or which of two equal lines to follow. And the opponent would be making similar decisions.

I was once told (years ago) by a friend that when I played correspondence chess or checkers I was not playing my opponent -- my library was playing his library. And of course, whether to trust a book, what line his library might neglect and all those considerations entered into it. So the issue is not new. It's that reliance on books was accepted years ago and assistance from computers is not yet accepted.

So my question is, if we all openly used computers to assist us, then what? Would all games be drawn? I doubt it. Would it all be chance? No. Perhaps some different skills would come in to play, but that is all. And the level of checkers played would be higher.

[Alex_Moiseyev]

Veto!

Ingo, I override your veto !

Title is given for PERFORMANCE, and it is given to PLAYER. It is not written anywhere that GM nominees should have two hands and foots, noise and eyes. The nominee should show performance at the certain level, thats all.

I would address you to chess at this point. Each program has assigned rating, and the best (?) program today "RIBKA" has rating 3000 or 3100.

However, I understand your point. You can award programs with "honor titles"

And in addition - if we allow programs to play in human events, they should have a rating. ACF never made a rule that programs are probited to play. And if KingsRow or WCC attend USA National and do well (2 times in 5 years), the programmer has all rights to claim title ... unless other rules will be written in WCDF where this sitiation would be covered clearly.

I think programmers role in promoting checkers is seriously underrated. Checkers are different today than 6-8 years ago and completely different than 15 years ago, when Chinook came in. We should take out our hats against these guys who worked for us, in most cases - for free.

KR and Cake opening books are 30%-40% different from PP which we had just 5-8 years ago.

Alex

[Alex_Moiseyev]

yet there are well over 1000 GM's

John,

thanks for posting this. You can pass my personal greetings to Dan and tell him that chess has very serious problem with "title inflation" which needs to be fixed.

I heard that several years ago when number of chess GM was around 700, G. Kasparov proposed to create a "Super GM" title which covers top 20-30 players only !

It seems like if decide to award our best programs by title - we will be ahead of chess and set a good standards and example they may follow

Respectfully,

Alex

[tommyc]

This talk of computer Grandmasters is side-stepping the real issue and is fantasy talk.

The issue is GM status for real living breathing, working, Mail playing pioneers people, not some machine.

Lets try stay on the subject whatever ones opinion is.

After all the Grandmasters (and others)are now reaping the information Mail play has put before them with relish from new published play.

[tommyc]

The technique is called dropout expansion "says Ed Gilbert"

As some posts have alluded to comp. programmers have given lots of their time free to this pursuit. This is very commendable. Im not that well up on comp knowledge but if these programmers dont use any play presented to them by Mail players(especially) then im wondering why the barred openings werent open long before Starship?? And of the barred opening that remain then it should be no problem for one of these programmes to actually use their ""dropout expansion"" to solve them?? Maybe some one can put me right on this.

[Ed Gilbert]

if these programmers dont use any play presented to them by Mail players(especially) then im wondering why the barred openings werent open long before Starship??

I started doing dropout expansion in 2001. I think this is also the year that Thomas Linke presented his paper in which he described the basic concepts and gave it the name 'dropout expansion'. I don't know much about starship, but when I compare the kingsrow book with Cayton's compilation I see more differences than similarities.

And of the barred opening that remain then it should be no problem for one of these programmes to actually use their ""dropout expansion"" to solve them??

Of course that has been done, both by me and by Martin Fierz. No draws were found. The results are in the opening books of kingsrow and cake. I also re-expanded the lost openings last year using the 10-piece endgame database with the book generator program. No draws were found.

-- Ed

[Alex_Moiseyev]

No draws were found.

Ed, this is really good news, because existing ACF card system which I developed several years ago strongly depends on the mathematical fact that 156 = 13x12

The bad news are - probably I will have to pay $50 to Brian Hinkle very good correspondence player. It was my bet that when Chinook finish it's work - there will be at least one difference in existing deck. Brian predicted no differencies.

Tommy, it might be interesting for you that Brian main point was - "otherwise Mail players won't miss it !"

Sincerely,

Alex

[tommyc]

Alex ive not seen Hinkles post...i dont get that "otherwise Mail players won't miss it !" I wouldnt take a bet w/ Brian Hinkle he nvr pays up!!.

[Alex_Moiseyev]

Tommy,

my conversation with Brain wasn't a part of forum but result of email excahnge. Brian wanted to say, that if (theoretically) we would have an opening which is not a part of the current deck (considered as lost today, but drawn), or opposite situation - we have an opening today in 156 deck which is loses and must be removed ==> then mail players would find this long before.

And this is the reason why he beleives that there shouldn't be any more changes to existing 156 deck.

I think I still have 10-15 years ahead to accumulate this sum ($50) until Chinook finish it's work !

Sincerely,

Alex

[tommyc]

My opinion of B. Hinkle is that he talks alota crap but doesnt walk the walk..Alex...........Hes good on the theoretical but shy on the practicalities In other words he might slavver over how it should be done but dont ask him how he might get there/