so, the war between human and engine with regard to chess is over and you are on the worng side of history here
How can you say the War is over without firing a shot? Up until now we have been playing chess hamstrung by rules patently beneficial to the computer. To declare the war over reminds me of the great chemist Lavoisier who when a peasant presented him with a meteorite said, "Stones cannot fall from the sky."
Your mind is made up without ever testing the rules. I have quantified how each rules change will impact the final result. Even GM Lev Alburt recognized how important it is to go back to 120'40 followed by adjournment; this one rules change over a 20 game match would add 100-200 ELO rating points. The ability to access opening and endgame manuals would add another 100-200 ELO points. Playing in consultation games would add 100-200 ELO points.
Finally, if we required computers to spend at least two seconds/move with no time delay would add another 200-300 rating points.
Those are the intuitive/logical estimates of how rating points humans could gain my rules are implemented. It is completely unscientific to proclaim computers permanent victor without testing them with human-friendly rules. Other "reasonable" rules changes are possible as well.
Again I don't see the OP's point. Machines can do some things better than humans, but so what? There is no reasonable need to compete with the machines. If we, as humans, could not build devices that exceeded our innate capabilities, we would still all be living in caves and gathering grubs for dinner.
A cannon can send a 14 pound metal ball father than a human can throw one, but we still have an Olympic event for shot putting.
Agree with you. I think changing the rules to level the field is pointless. In chess the objective of Man vs Machine is not to prove who is better but to see how far can the human capability be stretched, how far can the brains ability be pushed and to see what how far can we extend the limits we have come to know through prior studies.
After all the chess engine is a mere product created by humans' cumulative finds and learnings over the years in chess and technology so I dont see any point in proving we are superior when it is glaringly obvious.
I think y'all are overlooking the most important thing. If the rule changes we're to take place often, you would destroy professional human vs. human chess. Because now with top GM's not studying openings and endgames, the competitveness goes way down and you would have to apply these rules to all chess, resulting in a disaster! Maybe once or twice this would be cool against Houdini, but you can't go overboard with it.
sloughter: gm lev alburt is being nice to because you are paying him money ... he is humoring you because he is a nice human being and does not want to hurt your feelings once again, your own words in your own post indicate how far out of touch from reality does your position sit ... your statement: 'how can you say the war is over without firing a shot?" is ridiculous and flies in the face of the reality of the past 20 years because humans beings have been trying their best to beat engines and when kasparov lost to deep blue it was clear which side had gained the upper hand ... since that time one side had become progressively stronger as clock speeds DOUBLE every two years and human capacity does notyour entire position is spurious and based on some weird obsessions you have which causes you to spam multiple websites with the same tired & weak argumentsand seriously, this whole name dropping thing you are addicted to is very sad and has such a level of pathos attached to it .... gm alburt is a nice guy but he runs ads in the back of chess magazines to catch fish like you and to keep income coming in .... he's just too nice a man to tell you the truth about your ridiculous ideas ... this coupled with you constantly mis-applying his actual words is like fitting round pegs in square holes and you just take your little hammer and pound away until you think it fits perfectlyyou are a class b player ... you are not going to reinvent chess no matter how much you desire to do so ... and neither am i for that matter yet so its not like i feel some superiority over you in this regard ... just one of us accepts reality and the other one is detached from it and creating their own fantasy worlds where gms are agreeings with every crazy idea that you can come up withplay ... have fun ... go to tournaments ... meet nice people ... maybe even improve and become a class a uscfbut please for the sake of what is right, stop with this silliness
this one rules change over a 20 game match would add 100-200 ELO rating points. The ability to access opening and endgame manuals would add another 100-200 ELO points. Playing in consultation games would add 100-200 ELO points.
And numbers drop from the sky... I particularly disagree with the last statement.
That said, it is quite impossible to know whether GMs in correspondence can beat the computer under <insert conditions favourable to humans/unfavourable to computers> without actually pointing to actual games.
Also, human-friendly rules are quite unreasonable, and will eventually boil down to issues such as:
-The computer can calculate too many positions. We demand to reduce it to the human limit of 2 positions per second.
-The computer is too careful and will never blunder in <insert creative ways here -- touch-move, parallax errors, etc.>. Either we force it to blunder once every 3 moves or we will henceforth allow takebacks for human entities/groups under <insert 83 eventualities here>.
-The computer has too much memory. We all know memory is important in chess. The human record for memorising digits of pi is 67,890 digits (Wikipedia). Providing that it's more reasonable to spend a lifetime on chess than memorising digits of pi, let's reasonably cap the memory a computer can use to 1MB, equivalent to slightly over a million digits of pi (or well over 2 million, if you optimise).
-We end up with a computer that evaluates 2 positions per second, with 1MB of memory (say good-bye to extensive calculation trees; no RAM to remember all "interesting" positions).
So the contest will eventually be: given the same processing power, who can make better moves in a game designed to be intuitive for humans?
Rather than change the game, I think a match of "humanity vs the machines" is prudent here, where humans are allowed counsel among peers, but no technological assistance, and machines are allowed all its technological advantages, being machines.
Will a collective band of Top GMs trump the engine I wonder?
We don't know the limitations of the human mind; when a player like Sophia Polgar can have a performance rating of 2850 in Rome 1989 as an IM, think of what would happen if a Carlsen, Anand or Topalov had a similar massive jump in playing ability!
Bobby's performance rating in a 19 game winning streak leading into the candidates tournament was about 3100; remember he defeated the second highest rated player of the western world, Bent Larsen, 6-0 and beat one of the strongest Soviets, Taimanov, 6-0 as well as 7 games in a row in the Qualifiers.
You will find many instances even now where humans will find some long-term positional compensation in a position the engines will miss until much later.
What if in a 10 game match with these new rules, humans were able to draw 5 games in a 10 game match? What humanity would be treated to is the highest quality chess that has ever been played by a human. With the right rules, here is how I arrived at my guesses for the rating points:
In a 10 game match with unlimited access to opening and endgame theory, humans would draw one or two games they otherwise would lose,
In a 10 game match in a consultation game, humans will draw one or two games they would otherwise lose,
By returning to the old rules i.e. 120'40 followed by adjournment humans will draw one or two games they would otherwise lose. This was an unsolicited comment from GM Lev Alburt.
By eliminating time delay and forcing the computer to take at least 2 seconds/move (not computation), humans will win a game as they run the engine out of time in a simple endgame.
We will never know how good humans are at chess unless we have an objective yardstick---the computer is the yardstick. The only way the yardstick makes sense is if we give humans every meaningful fair rule to level the playing field. So far we haven't even done this.
Think how simple it would be to test just one rule! Two or three pros could try out a consultation game to see if they improve; it is easy to do and costs nothing. Wouldn't it make sense to provide a fair, objective test of human abilities? It is not a fair test to force humans to complete a game in one session. Humans make fatigue errors; computers don't. This was known fifty years ago hence the rule 120/40 followed by adjournment.
Simply writing off humans without leveling the playing field is a defeatest view. Even if the humans don't do well, it will still be the finest chess ever played by human beings.
In a humans vs computers matchup, I think the peer consultation aspect for the human side should account well enough for issues of fatigue and blundering.
I don't look at engines as a single entity, even though you're playing against a single machine. Computer engines are the totality of technological advances, including brute performance benchmarks and games databases, and so the only "fair" matchup between humans and computers is a peer consultation game, since that will represent the totality of human capabilities.
It's already assumed that the world's strongest GMs have their opening theory down, so we don't need to allow opening book references, since it's probably inclined on the aggregate through assignment of team members. We want to keep the human side as "organic" as possible.
If machines are still found to be winning matches against human fatigue, then longer time controls, where players are allowed adequate rest should be enough for a "level playing field", without radically changing the rules of the game itself. Not accounting for fatigue is already manipulation in itself, and so we need to rule that out if we are to come to a more useful conclusion as to the state of human chess. I want to see humanity's optimal best vs the computer engine's best.
Rule changes, like setting arbitrary limits on natural chess capabilities is just voluntarily adding confounds to the game.
sloughter, your ideas are beyond supercilious ... lev alburt should refund your money your inability to accept what the majprity of people are trying to tell you really begs the question of whether or not the money you spend on chess lessons would not be better spent with a mental health care professional just because the engines of today have given you an inferiority complex is no reason to keep mind raping people on forums with your bizzaro-world fantasiesif you really need to beat a computer in order to realize your humanity or to feel better about yourself or the human condition than engage it in a kick boxing match as that arena still favors human beings
Houdini 3 Pro can be beaten if it chooses the wrong openings. Here it chose the Two Knights Defense. I invite you to demonstrate equality for Black at any time in the main line; Black made only one second best move and that was all it took for me to secure an advantage. Due to a programming defect it lost on time. When it lost on time I asked the Pro to provide best play for White and it found a winning plan for White. Don't you think if I can beat Houdini 3 Pro when it makes one second best moves that top professionals could beat it?
If you can beat the engine so convincingly, as you seem to be suggesting through fault of programming and lost time, that doesn't support your case for modified rules.
The point is that it would just be plain silly to reduce the computers strength just because it hurts the human pride and ego. Thats the point to instead say: "F...ck the human pride and ego - lets instead move on beyond all that nonsense!"
Now what could be cool on the other hand is to ENHANCE the human playing strength and test it against the machine. Not with the purpose of beating it , but as a scientific experiment to see what happens.
How could this be done? First of all you must give the human more time on the clock. Maybe even something like 5-6 hours for 40 moves. Also human consultation without engines would be very interesting because we would learn much about group decision making. And it would be great fun and entertainment for making video and putting everything out on the web. Another thing one could do is allowing the humans a few takebacks during the game, lets say they can have three takebacks. And maybe each time they can take back up to 5 moves backwards ( many variations are possible here ). And it seems fair to maybe play the computer without its opening book, since this is mostly the work of humans anyway. And playing them without book increases the chance of interesting innovations from the computer: It thinks by itself.
This could be a really cool contest IF the computer played at full strength with all processors firing and permanent brain with the strongest programs. Because the point must never bee to limit the core computer strengt, but to create super quality human chess to test it against. And have great interesting games as a result.
My prediction is that in such circumstances the computer would be clear favorite to win. But an even clearer favorite to win would be everyone involved both as players and spectators. Becuase if viewed as fun and as a scientific experiment it would be a great learning experience about both human and computer intelligence. We would learn something more about ourselves , about the example of chess and about intelligent decision making and resource management in general.
Win or lose some specific chessgames - the humans playing with and learning from the machines , will all totally win in any case :)
For a somebody who is not an active chess player (but just interested in chess every now and then), shanahanahan's post looks very wise and well based. A game presenting the best of both worlds - best computer calculating power and the best human creativity and intuition - would surely offer new views and information on the thinking process in a puzzle solving situation, and intellect in general. So, therefore I hold with those here who see it worth while to give the BOTH contenders the best possible conditions and see what happens.
I don't want humans to be able to beat a machine at all. Humans build machines to do things better than we can do. To cripple the machine so that a human can beat it is an insult to our imagination and creativity.
An insult to our imagination and creativity... means nothing to me. In fact its good that humans are still able to create and destroy "technologics", adore and love the human mind and its capabilities.
A pocket calculator from the dollar store can outcalculate any human. I dont hear anybody complaining about that. A chess program is a human tool, get over it.
Actually I thought this thread would have ideas about how to change chess so that computers could no longer outcalculate human players, the way they can't do in the game "Go".
That would have been interesting. I don't know enough about Go to know why the best humans (and even the not so best, apparently) still beat computers handily. But whatever that something is, would be cool if chess had a little of that.
I think there was a chess variant called "Aarima" or something like that, that humans also apparently dominate. I think that was its whole reason for existing, actually.
Just doing a bunch of artificial stuff like limiting the size of computers to a human head (whose head?) isnt very interesting.
Pitching machines throw faster than human pitchers. Oh NO! Cars can move faster than olympic runners! What is the world coming to.
As far as game databases, opening books, and endgame table bases the OP is right. It's not just that computers "calculate better", they are loaded up with multiple books of easy knowledge and shortcuts which humans are strangely denied access to when playing against computers...
Great post! This is my prediction as well---we will see the highest quality chess ever played instead of the error filled games of today where fast time limits are the norm. We have no idea of the synergy of two players discussing a position. We may see them have the ability to extend analysis in the middlegame 20 moves ahead i.e. over the horizon of the computer. We will never know unless we start giving consultation games a try.
What is great about a consultation game is that we can be part of the decision-making process. We get to step inside the minds of the chessplayers and think as they think, see what they see. This will make a spectacular spectator experience.