Originally posted by mateuloseTo my knowledge there's only 1 person who has ever used the term 'partial cheating',that person would be me.This has absolutely nothing to do with this thread.A serious question was asked here,as wulebgr already said.
What ppl in RHP need to break from is the habit of claiming something is only "partial cheating", or "I am only a cheater sometimes". There is no partial cheating or a sometimes cheater, you are either a cheater or not, if you use an engine for just one move, YOU ARE CHEATING, you may never get caught and there would be no way to prove mathematicly you ...[text shortened]... ing, but in your concious, you know it's wrong and you ARE a cheater, caught or not! Nuff said!
You've attacked me personally at least 3 times,last few days.This is a clear case of such an attack.Once again I ask,why are you doing this?
You rant a lot about people attacking you,yet you do the same!
Originally posted by Ragnorak[Event "?"]
Hey Wulebgr,
do you have Legal's game in pgn format?
[Site "Paris"]
[Date "????.??.??"]
[Round "?"]
[White "de Kermeur, Legall"]
[Black "St Brie"]
[Result "1-0"]
[ECO "C23"]
[PlyCount "13"]
[EventDate "1750.??.??"]
1. e4 e5 2. Bc4 d6 3. Nf3 Bg4 4. Nc3 g6 5. Nxe5 Bxd1 6. Bxf7+ Ke7 7. Nd5# 1-0
Originally posted by SirLoseALotI was going to suggest that my actions would the identical to those of any database with all possibly moves in them at the start of a game. Based on the positions of the pieces at start, a database will have a finite list of possibly immediately next moves.
In theory,I would agree it's cheating.Yet,if one does what pcaspian says,after a game is completed,and use it for another game that has an identical or similar position,it would be ok.
What's the difference?
If per se, I have a chess program that computes these moves prior to me playing a game (recursive function in programming), it could tell me how many times, when it played itself, did it win by making such a move and how many times it lost by making such a move. Ofcourse it is not 'suggesting' I make that move, merely showing me the statistical percentage that it won after it had made that move.
Simply put, all I am doing is using mathematics to my advantage. It is logically impossibly to determine every possibly move from start to finish, however as I am using a unique board position and calculating this each time I need to move, I am giving the computer a statistically easier chance of determining the position it found to be the statistically most sound.
The way I see it, for atleast the first 4 moves, the computer will be a very large database of 'existing' moves and as such puts me on par with anyone using a large database of more plausible moves.
cheers
Originally posted by pcaspianIf you relied only on such recursive analysis, and not chess history, your computer would need be much stronger than any existing today to compete with even the smallest databases. If I understand you correctly, you are suggesting a form of calculation that chess engine programmers abandoned 25-30 years ago. Only then did chess playing engines enter the C category. It would still be cheating, but it would not help you. On the other hand, if your computer had in fact played all those games, chess as we know it would be finished--fortunately we're a long way off from that.
I was going to suggest that my actions would the identical to those of any database with all possibly moves in them at the start of a game. Based on the positions of the pieces at start, a database will have a finite list of possibly immediately next moves.
If per se, I have a chess program that computes these moves prior to me playing a game (recursive ...[text shortened]... d as such puts me on par with anyone using a large database of more plausible moves.
cheers
There are 20 possible positions after white's first move; 400 after black's; 1862 after white's second move; then 9825, 53516, 311642, 2018993, 12150635, and after 4 1/2 moves 69284509. So, it takes a mere 3 1/2 moves to reach 2 million possible positions. Your computer must generate an incredible amount of data if it relies on statistics alone.
correction:
The numbers I gave above are for uniquely realizable positions; that is the possible positions that can be reached only one way after the given number of half moves.
The total number of possibilities after each of the first 8 half-moves are:
1: 20; 2: 400; 3: 5362; 4: 71852; 5: 815677; 6: 9260610; 7: 94305342; 8: 958605819. These numbers include only legal positions.
Thanks to François Labelle of UC Berkeley for this data.
Originally posted by Wulebgr
correction:
The numbers I gave above are for uniquely realizable positions; that is the possible positions that can be reached only one way after the given number of half moves.
The total number of possibilities after each of the first 8 half-moves are:
1: 20; 2: 400; 3: 5362; 4: 71852; 5: 815677; 6: 9260610; 7: 94305342; 8: 958605819. These numbers include only legal positions.
Thanks to François Labelle of UC Berkeley for this data.
My computer will ofcourse not be based purely on all possible moves and even the most elementary 'intelligence' will significantly decrease those numbers. A computer that continuously plays against itself (non-stop) and learn from its previous mistakes (AI) and charts those in a database. Should a certain line result in a high percentage loss, it will abandon that approach in future.
Irrespective of the intelligence in the program, what I am suggesting can theoretically be achieved in any program. If you chart the positions in Fritz and have it play itself (from that point onwards), it should be able to inform you (once its finished all its games), which next move proved successfull (in the long run) and which proved unsuccessfull. Simply put, as it goes along, it charts a new database before I actually make the move. Technicality ofcourse, but still a database.
You mean prove which moves are successful against a computer. All it takes is for the opponent to find a move that the computer didn't analyse (maybe because it's inferior if you can see 15 ply deep). However when that happens you have to work out the way. That's far from easy in many cases. Find yourself a collection of puzzles and see if you can solve a mate in 7. Most will be almost impossible. Now what if instead of mate in 7 it was win a piece in 7. Hard? Yes.
Humans play differently to computers. Humans play different to other humans. This method will not be a foolproof win. Without a good level of chess skill it won't work.
Originally posted by XanthosNZ
Humans play differently to computers. Humans play different to other humans. This method will not be a foolproof win. Without a good level of chess skill it won't work.
I doubdt that not. Yet computers make Kasparov very edgy at times. Sufficiently intelligent computers are very difficult to beat and they're getting bigger memories, faster brains, and are designed more intelligently. Ofcourse Kasparov attempts to play specifically to 'fool' the computer based on the logic it was programmed in. As computers become wiser, this will not always be possible.
Either way... theoretically speaking. What is the difference at the start of a game between using a database to make your first move.. and a computer ?
Originally posted by pcaspianThe fact that the databases is stored knowlege and that the computer creates a move. One you look up the other is made from scratch :-)
Originally posted by XanthosNZ
[b]
Humans play differently to computers. Humans play different to other humans. This method will not be a foolproof win. Without a good level of chess skill it won't work.
I doubdt that not. Yet computers make Kasparov very edgy at times. Sufficiently intelligent computers are very difficult to beat and they ...[text shortened]... e at the start of a game between using a database to make your first move.. and a computer ?
[/b]
Originally posted by pcaspianI have hundreds of thousands of computer vs computer games in my databases, and am constantly adding more. In fast, Arena has been running an engine tournament in the background 24/7 since Saturday. It is a thematic tournament using an opening I'm fond of (and for which I have a shelf of books). All these games will be added to my main database, and I'll learn from them. No doubt I'll be playing a "novelty" that is not, in fact, a novelty because it was suggested by my own private store of chess history.
My computer will ofcourse not be based purely on all possible moves and even the most elementary 'intelligence' will significantly decrease those numbers. A computer that continuously plays against itself (non-stop) and learn from its previous mistakes (AI) and charts those in a database. Should a certain line result in a high percentage loss, it will abando ...[text shortened]... s a new database before I actually make the move. Technicality ofcourse, but still a database.
[/b]
I believe this process is entirely different than having the engines run while I play. It is a form of study, training, practice, research. Anyone who wants to work at it can double the size of the largest commercial databases--adding games from RHP, ICC, RWBC, and so on. The complexity of chess remains such that computers are still a long way from removing all the risk, although Bobby Fischer believes otherwise.
My point is, so long as you are creating databases before playing, rather than using engines to calculate while playing, you are within the bounds of acceptable correspondence play where using engines to calculate moves is forbidden (some correspondence organizations, such as ICCF, permit engine use).
The quality of the databases you can create remains limited by processor time, engine strength, and the time control/ply depth you specify. One computer can create hundreds of games in a few days or only a few games per day, or even one move per day.
An example: in a game I posted elsewhere in this thread, I found an idea five minutes after offering a draw to my opponent, who accepted the offer, that convinced me I might have winning chances. Hiarcs 9 confirmed my assessment within a few seconds, but then favored a different move after 30 minutes of analysis; after 90 minutes it was back to favoring the original move. I'll likely never see this position on the board again because I gave myself some uncomfortable problems with the opening moves, and will avoid that line. But, the changes in Hiarcs's analysis raises some questions regarding how long I should let the engine run for each move while I'm exploring new lines in my pet openings.
I've heard statements to the contrary, but I believe 1500 rated players are much stronger today than 1500 players were ten years ago. Training with computers has contributed to stronger chess players and tougher competition.