Forums
Most Recent
 Exquisite Chess set hand crafted in prison.
 FACT: You can't improve at chess
 What Does "gg!!" Mean?
 Who is the greatest chess player of all time ?? Bobby Fischer ??

Need help figuring out what these chess symbols mean!
NM eltenedor
 cant castle in 960
 Stuff NonChess Players Say
 6/22/2017  Cutting Off The Escape Square
 That's exactly what's going to happen to "MyRatings 1523"
 Race to 100: Bishops VS Knights: Promotion
Forum Legend
 Following
 New Comments
 Locked Topic
 Pinned Topic
A few years ago, in the distant past for some, a Canadian company by the name of Chanuk used multiple processors in a supercomputer to go through all the possible variations of Checkers (Draughts). By knowing all the possible variations that can result from certain moves, they can effectively always win the game by knowing which moves win and which moves lose. In recent news, I discovered that Chess will undergo a similar study by this company, and will be beaten within the next 10 years.
I have heard many speak of this event with negativity, and some with positivity. How do you feel about a computer successfully "Beating" chess?
Will chess still be played with the passion it is today if the game has been "beaten" by Chanuk?
Will you, personally, continue playing if the game is beaten?
My answer is that I love the game of chess, and will play it despite the supercomputer's involvement, and I hope the game of chess will continue to be played with passion and brilliance.
A few years ago, in the distant past for some, a Canadian company by the name of Chanuk used multiple processors in a supercomputer to go through all the possible variations of Checkers (Draughts). By knowing all the possible variations that can result from certain moves, they can effectively always win the game by knowing which moves win and which moves lose.
This isn't exactly true. Checkers has been solved by computational proof, but that doesn't mean the computer will always win, only that it will never lose. A very skilled player could play the computer to a draw.
It's been recognized for some time that checkers is much less mathematically complicated than chess, and tournament play has responded. A standard tournament match begins with three random moves before they players even start, so that the game doesn't get played out.
Have people stopped playing Checkers? No. Have people stopped playing Chinook (Chinook, not Chanuk, is the name of the computer) at Checkers? Yes.
In recent news, I discovered that Chess will undergo a similar study by this company, and will be beaten within the next 10 years.
I believe this is false. Can you cite a source? It just doesn't make sense numerically that chess could be solved in the same manner as checkers in that short of time.
Here are some numbers for those who think about solving chess this way. I'll give you the benefit of the doubt that there are only 3 moves to look at for each side on each move (obviously there are more than this). That means that in 20 moves (40 half moves) there will be 10^19 positions to generate and evaluate. If you are able to generate and evaluate a billion positions per second, it would take you 316 years to generate and evaluate the end positions of the first 20 moves.
Given 3 moves to consider each half move, "solving chess" 30 moves deep would require generating and evaluating 10^28 positions. If you had a billion computers that could each generate and evaluate 10 billion positions per second, this would take over 31 years.
7man endgame tablebases are expected to take another seven years to finish, so I figure we are much more than 10 years away from 16man tablebases (a solved game). Barring aliens, time travelers, or Gods giving us new technology:
"So, what does all this mean for chess? Can we extrapolate these results from one checkers playing programs to chess playing programs? Certainly, Chinook and Fritz use similar search algorithms. They each have a positional evaluation function. They each take advantage of table bases for evaluating endgames. The big difference is the number of positions possible in each game: 10^{20} for checkers and 10^{40} for chess. To get some idea of this, if a computer could solve checkers completely in one nanosecond (a single cycle of a 1 GHz computer), it would take this computer 3000 years to solve chess."
Also, Chinook's team leader was quoted as saying, during an interview:
""Given the effort required to solve checkers, chess will remain unsolved for a long time, barring the invention of new technology.""
Perhaps you are correct loomis, I dont see how anyone could know that for sure though. In any event Marion Tinsley is incredible! I am no checkers fan but just imagine being at the very top for more than 40 years!!? I doubt this has ever been done in in any other sport/game. It makes Lasker's 27 year chess reign pale in comparison.
"I dont see how anyone could know that for sure"
Reb, do you believe it can be done for tictactoe? If so, why not checkers? In principle it can be done for chess, but the time and resources necessary are too large for the current or foreseable technology of humans.
Here are some numbers for those who think about solving chess this way. I'll give you the benefit of the doubt that there are only 3 moves to look at for each side on each move (obviously there are more than this). That means that in 20 moves (40 half moves) there will be 10^19 positions to generate and evaluate. If you are able to generate and evaluate a billion positions per second, it would take you 316 years to generate and evaluate the end positions of the first 20 moves.
Given 3 moves to consider each half move, "solving chess" 30 moves deep would require generating and evaluating 10^28 positions. If you had a billion computers that could each generate and evaluate 10 billion positions per second, this would take over 31 years.
Ah, but here is the brilliancy. It simply plays itself continously using multiple processors, and the data is fed into a bank for the computer to analyze. Using multiple processors means it is playing itself at least 100 times simultaneously (this is what I have read, but it could be an untruth), will allow it to solve chess by evaluating each of the games and seeing which move won. If each game takes an hour, and the computer is on continously, that means that it will evaluate 24,000 games a day, each with multiple positions. It will then, using the data it learned while it was playing, begin another set of a hundred games. It is simply playing itself continously to learn which moves won the game. So, each year it is evaluating 8,760,000 games, which means at least that many positions. As I understand it, the company released a statement when they came out with Chinook Checkers saying they would be doing chess next. It would however, be extremely amusing if that was PR and did not mean anything definite.
Unbeliever, that still doesn't even get you close. Notice in my post that I allowed for 10 billion computers running simultaneously, so to say the least I am not impressed at 100 simultaneous games.
You say 8.76 million games a year. Let's give you the benefit of the doubt and say 10 million games a year and 100 moves analyzed per game. That equates to 1 billion moves per year. But as I stated above, you won't even make it at 1 billion moves per second or even 100 billion moves per second!