♡ 19 ( +1 | -1 ) The ansswerto second question is yes. It has been already discussed in another thread; I don't remember the site where these kind of tournaments are held. Maybe someone does remember.
♡ 74 ( +1 | -1 ) advanced chessThere is a game called "advanced chess," in which the opponents are allowed to use computers. In fact, World Champion Vladimir Kramnik and Vishy Anand will be playing an advanced chess match in Leon, Spain this month (June 20th-June 24th). The players will be allowed to consult databases and chess engines. It will be a six game match. The time control will be: entire game in 1 hour.
Such a match would be interesting only between grandmasters. The point is that the computers will eliminate probably all tactical blunders so that the players can concentrate on deep strategy.
If I were in an advanced chess match, I would simply let the computer play for me.
♡ 104 ( +1 | -1 ) choiceThere seems nothing wrong to me with the desire to use a computer as long as your opponents are using one too. It is a different kind of chess, but who's to say what is better. At least that way, no one can cheat.
At the ICCF and IECG computer use is allowed. It is very hard to compete against very good players who are also using a computer when you don't use one yourself.
My last compterless game in the ICCF we had an extremely complex game with many pieces hanging and both Kings about to be mated. I spent about six days/move with no certaintly I'd played the right move.
My opponent only spent one day/ move and all his moves were perfect and he eventually won. After the game he sent me the computer analysis showing my errors (keep in mind, this was completely legal there).
So I now use a computer in all ICCF games. There is no way I am going to give such an advantage to an opponent who is probably just as good as me without computers.
But I have to say, it is really refreshing to come to a site like Gameknot where computers are not allowed.
♡ 96 ( +1 | -1 ) The great cheaters tournamentHow about once a year we could have " The Great Cheaters Tournament." That's a tournament where anything goes. You could use your chess program, an expert or a master (if you know one), a data base----anything! There would be no divisions in this tournament, everyone is equal! Fritz is the great equalizer. Can you imagine walking into a saloon with that big Fritz hanging on your hip. The bar girls would fawn over you and those mean looking hombres at the bar would move over and give you some room. Why you would be hunting with the big dogs! We could charge a nominal entry fee and get a large turnout. Even Twinkletoes (1257) might knock off Brunetti! Happy days are here again! If there is a large turnout, Gameknot could award the top finishers lifetime memberships. One thing is for sure---almost everyone has a program and when it's percolating on all cylinders they feel invincible.
♡ 41 ( +1 | -1 ) not so clearIt's not that easy to use a program against another opponent who is also using a program when this is legal to do.
My opinion is that usually the better chessplayer will win. But there is a skill to gaining insights from the program and knowing when to ignore it.
It's not like you can just strap a big Fritz on your hip. Fritz is not even the best program to use for correspondence chess. Not even close to the best.
♡ 33 ( +1 | -1 ) notopPlease dont build a monument around my statement, it was mostly in jest. When I use Fritz, it's a synonym for all chess programs and thanks for informing me about not being able to strap on a Fritz----I didn't know that! You seem to know a lot about computers and correspondence chess. What is the best and what do you use?
♡ 111 ( +1 | -1 ) humans vs. computersThere have been a lot of threads talking about computer use. I'm intrigued by the top players' comments on the limitations of chess engines but given that processing power has pretty much doubled every year for at least the last 10 years, surely now chess engines can beat GMs hands down?
It's been a long time since Deep Blue/Kasparov (and no rematch despite opinions that Kasparov was not playing to his normal standard).
Does anyone know how Deep Blue compares with latest PC engines? I know it was a purpose built RISC system but that was when a Pentium processor was state of the art.
I have Chessmaster 6000 which has some good tutorials - is Fritz or any other engine way superior in anyone's opinion? Can the top GK players beat all of these programs on their highest setting? Surely a chess engine consulting a database of all GM vs GM games cannot be beaten?
I think the world's perception is that the Deep Blue victory signalled the turning point in computer AI whereby in chess at least, humans are inferior - I hope not. Any opinions?
♡ 25 ( +1 | -1 ) about chessmaster,chessmaster is not very strong,unless mine has something wrong with it.I can beat a so called 2100 player in a blitz game!I think they use a different rating system,either that or I am an amazing player against computers.
♡ 11 ( +1 | -1 ) same with minemy cm8000 did the same to me also. but now it takes longer to analyze things and it seems to be fairly close to FIDE.
♡ 199 ( +1 | -1 ) bullmoose..To answer your question regarding chess engines, the top engines are playing at around 2700 Fide right now. Recent matches saw the engines( Chess Tiger, Fritz 7, Hiarcs, Junior, etc..) defeat Gulko and Gurevich, both rated in the 2600s.. however, Smirnin was able to beat the computer conglomerate... he is ranked around 2700 I believe.
The last version of Deep Blue which played Kasparov searched 200 million positions ( or "nodes" ) per second.. which is still a good deal faster than today's progs which on the fastest hardware , which search around 5 million/sec or so. It is important to note that there are many factors which influence the strength of a search engine; many pundits believe an engine like fritz has a much better "evaluation" factor than deep blue, meaning if fritz could search as fast as deep blue it would beat it hands down.
According to Moore's law, CPU speed roughly doubles every 18 months. A doubling of CPU speed does not dramatically increase the strength of a program; in general a program is able to add one "ply" to its search depth every 2.5 - 4 CPU multiples, that is, a CPU of 1200 searches about 1 ply more than a CPU of 400 MHz or so. So, in reality, strength based on hardware gain will be a steady but fairly slow increase in strength.
It's also important to note that the latest installment of deep blue only played 6 games; this is hardly enough to gauge its strength. I personally believe that Kasparov would not lose a match to that version of deep blue had he had a few practise games or had the tournament conditions been more "normal".
I talked with Frans Morsch ( programmer of the world's strongest program, Fritz 7 )a few months back, and he told me he was working on a new breakthrough that would make his engine, in his words, "virtually unbeatable".. but went on to say that this is 5 years away. Time will tell.
♡ 49 ( +1 | -1 ) Wewill soon know, were the humans stands against programs. Kramnik is meeting Deep Fritz on October this year, that should give us a fairly good idea were humans are on the line!
Have a look here: www.chessbase.com/newsdetail.asp?newsid=363
Best wishes Cairo
♡ 30 ( +1 | -1 ) CairoWhy don't you play a computer here on Gameknot? I'm sure we have lots of people who would volunteer to man the computer. I think it would be extremely interesting to see how an excellent player confronts a chess program. You could be talking to us while you play (the computer can't hear you). What an excellent training device!
♡ 2 ( +1 | -1 ) sure---why not show me how to wip cm8000?
♡ 48 ( +1 | -1 ) JimI always like a good challenge :-)) so why not, on the following conditions:
1) The game don't take place before beginning of August 2002.
2) I want to know up front what kind of program I'm playing against.
3) Also, who is the computer operator.
4) I'm playing the white pieces.
5) Minimum 6 days per move.
Good idea with the "talking" on the way. We can open up a thread for that!
So, what you say :-))
Best wishes Cairo
♡ 18 ( +1 | -1 ) your on!you can play my chess master.and to save you from losing rating if I win or draw,I will resign no matter what the outcome of the game.the time control can be 7 days per move.OK?
Tell me some more about your program and yourself. You can read most about me in my profile!
♡ 7 ( +1 | -1 ) Sounds goodThat gives you about six weeks to get everything set up---Should be fascinating!
♡ 99 ( +1 | -1 ) I think,with all my respect to tulkos, that there's no match between his CM8000 and cairo allowed to think a week per move. Cairo could beat CM8000 even in a 2 hours game! As I wrote before, in the slow games the human player is advantaged, especially if he's a 2400 ICCF player :) The only way to have an interesting game with cairo is to play it as fast as possible (3 days/move), to use the best software/hardware possible combination (Deep Fritz on a dual Pentium 1600, or Fritz 8 on a Pentium 2000, at least), allowing the computer to analyze for hours (at least 4, 5 per move). Even so, the human player keeps an advantage. Otherwise there is no match at all :)
♡ 22 ( +1 | -1 ) I agree, Brunnetti.Many strong players here on gameknot could defeat CM8000 if given enough time, and especially if they knew they were playing a program, (programs have certain weaknesses.)
♡ 18 ( +1 | -1 ) In answer to the original question...Everyone thinks it is cheating....even those players who do use "programs"....if they didn't think it was cheating, they would be honest and up front about their use of "aids".
♡ 42 ( +1 | -1 ) alright,butcairo would probably not think about his moves for 6 days non stop,while I can give the computer that much time to think about each move! If cairo doesn't mind,I could look through the chessmaster databases as well.this might be taking to much of an advantage however.and if brunetti has any doubts about the strength of the program while i check the databases,he can challenge me.
♡ 32 ( +1 | -1 ) I have no doubtsabout CM 8000's strength! It's a fast-play strength, and can't match human strength in correspondence chess. I accept your challenge; you can use books and databases as you like, but you won't, since I gonna play a totally unorthodox opening: this way it's me and CM only. Can we play via forum (like the GKCM-The bunch game)?
♡ 77 ( +1 | -1 ) Kramnik vs Deep Fritz / Fritz 7I don't think is much of a contest. As I mentioned in my previous post, we've already seen Smirnin beat a team of the best progs, under conditions less favourable than Kramnik will have... and Smirnin is definitely weaker than Kramnik.
Originally the conditions of the match allowed Kramnik access ( for a month or more I think) to a copy of Fritz which would be identical to the version he would play, for practice purposes. Unless this condition has changed, this should result in a Kramnik crush, since the program operators are only aloud to make small modifications to the opening book, and no modifications to the engine's code.
I suppose anything could happen.. perhaps Einstein Group should hire Don King?
♡ 4 ( +1 | -1 ) what do you mean?start a linkand then post moves there?
♡ 3 ( +1 | -1 ) Yes,a thread. If you like.
♡ 2 ( +1 | -1 ) all right!thats what I'll do!
♡ 149 ( +1 | -1 ) ??Seems like we have a lot of computer experts here, I'm not sure I understand the most of it :-)) However I accepted a game against a program called CM8000, with tulkos as an operator, on agreed terms. But it seems like brunetti and _mda_ don't think this is a match at all!! First of all I think you guys overestimate my strenght. Secondly I'd hope I could beat the program, so maybe other players here on GK, wouldn't think, it wasn't "so fancy" to play with a chessprogram against humanplayers.
All this: "The only way to have an interesting game with cairo is to play it as fast as possible (3 days/move), to use the best software/hardware possible combination (Deep Fritz on a dual Pentium 1600, or Fritz 8 on a Pentium 2000, at least), allowing the computer to analyze for hours (at least 4, 5 per move). Even so, the human player keeps an advantage. Otherwise there is no match at all :)" I have a little understanding of, but since you guys are so "hot" on this, I will leave "the scene" to you, its all yours and I'm withdrawing my acceptance for the game.
tulkos thanks for your challenge on behalf of your CM8000 :-))
Best wishes Cairo
♡ 89 ( +1 | -1 ) seems a little unfairWith all due respect to Brunetti it strikes me as a little unfair the way he has usurped Cairo as regards the computer match. The original challenge seemed to involve Cairo versus cm8000.
Cairo seems to have accepted the changed circumstances with good grace, but even so, it does seem a little unfair that Brunetti has decided to replace Cairo with himself.
Perhaps you could both play cm8000 ?
Best wishes to you both,
♡ 23 ( +1 | -1 ) Let bothGK GMs play against the program. There is no rule against that. I am rubbing my hands together at the prospect because I have never beaten CM5000 (thats what i have) beyond moderate playing strength.
♡ 88 ( +1 | -1 ) Unfair replacement?Cairo accepted to play not before August; now we're in June. Where's the replacement? Can Tulkos play only one game per year?
Tulkos challenged me too, and I don't think this is an usurpation.
I've expressed my thoughts about the outcome of a human-computer match: is this unfair? My opinion is that the human player can easily win.
Cairo: why you got offended? You can play your match, and if my forecast is that you will easily win where's the problem?
Corimaje: have you ever played your program making a move a week?
♡ 218 ( +1 | -1 ) What The Heck Playing Vs. Computers!!!In my personal honest opinion, using computer softwares while simultaneously playing a GK correspondence game is absolutely not CHEATING in the strict sense! WHY ???
1. The ultimate objective of GK is to promote goodwill, camaraderie and friendship among its members. 2. There is no monetary reward involved in winning games here. 3. Prestige is not in the offing here because, as a matter of fact, majority of the players do not disclose their true names. 4. High ratings do not carry much heavy weight because what counts most is how you train and test your practical approach to the game. 5. The player playing against the computer is the one enjoying the game and the one actually benefitted because he is mentally trained. It'll be everything to gain. On the other hand, the player using the computer is totally on the losing end, because a) he did not enjoy a single moment of the game due to absence of tension and pressure; b) operating a computer in chess entails a lot of waiting time, hence, he loses valuable time instead of concentrating on more productive undertaking; and c)even if he wins with the computer's aid, he still remains a weak player.
THEREFORE, don't be afraid of playing against an opponent using a computer!!! You'll definitely gain extensive benefits even if you loses.
Computers are computers, and they calculate innumerable variations in seconds. I agree with that reality. HOWEVER, they have vulnerable weakness such that in a situation that offers several equally good moves, they cannot decide, PSYCHOLOGICALLY or INTUITIVELY, which move that offers the most practical chance taking into consideration the style of the opponent.
Finally, in these CYBER ERA, almost all chessplayers are equipped with computer softwares. We cannot deny that!!! NOBODY KNOWS WHO uses computer while simultaneously playing here! If a player plays like a grandmaster, you cannot just accuse him of employing the aid of softwares. In correspondence chess, great games and masterpieces are generally created.
MY POINT IS ... don't be afraid to play an opponent using softwares. Face it HEAD HIGH !! After all, PERSONAL TRIUMPH will flow in your heart when you beat IT !!!
I just don't se any logic, for me to play against a computer on agreed terms, when you playing against the same computer, on terms there is less favourable! no logic see :-))
You have done and doing some great things for this site Alex and I really and I'm sure a lot of others appreciate that, just to mention the Thematic Tournaments, GK-Olympics and the match against "The Bunch" + you have be elected to The Hall of Fame, as the 1st. player here on GK. You are a "wisard" on the Forums, not many questions, you don't have an answer to or a oppion about, very helpful I would say! On top of that you playing an "outrageous" number of games, at the same time and not at least, your leading the rankinglist by more than 100 points to the nearest!! All very impressive achievements!
Here comes a little friendly advice, in a positive spirit :-)) When you read thru my previousley thread and you apperently sitting back, with the impression, that I should have bean offended in any means, I think its maybe time for you, "to slow down" your activities a little bit :-))
Keep up the good work Alex!
Best wishes Cairo
♡ 15 ( +1 | -1 ) Just for the record..Hi cairo..
When I said the match wouldn't be much of a contest, I was referring to Kramnik vs. Deep Fritz, not you vs CM 8000... see the title of my previous post.
bye for now..
♡ 51 ( +1 | -1 ) CairoI like your professional approach to playing a computer. Getting all the agreements up front and laying all the details on the table. Perhaps I missed something but where are the details on the current game? Were we told what kind of computer tulkos has? what is his chess program settings? How long is his program thinking on each move? What if he has it on novice level? It would be more enjoyable if more information was given. Perhaps someone with a strong program will challenge you to a game in Aug.
♡ 33 ( +1 | -1 ) JimThanks, and I like your attitude towards these issues :-))
I will be open for a challenge, as long as terms are agreed up front. Maybe somebody with a good program would like to challenge me, we will see!
Best wishes Cairo
♡ 13 ( +1 | -1 ) _mda_I apologise and stand corrected!
Best wishes Cairo
♡ 69 ( +1 | -1 ) using a computerusing a computer is absolutly cheating. I thought it was cheating before i even read that it was legal in some places. With a computers help the game of chess and the sport of it is gone. Why even play if you let the computer play the whole game for you? The point of the game is to think about each move and develop your skill. I also play at itsyourturn.com and i am not sure whether or not they allow it,no they dont in fact, but as i have seen in most tournaments in their last stages it is very obvious that they are using a computer. As i said before and will say again a computer is totally cheating. Chess is no fun when you let the computer play for you.
♡ 18 ( +1 | -1 ) So now we know...how Kramnik did against Deep Fritz. Now could someone explain to me how it is so easy to beat programs in correspondence chess. And I would like some proof that it really is.
♡ 32 ( +1 | -1 ) well,I never heard of it being easy to defeat a top program, but if it's easier in correspondence chess, I should think the reason would be that since a computers greatest advantage is it's ability to calculate large quantities of moves, this advantage would be somewhat nullified by the time given to the human to calculate.
♡ 95 ( +1 | -1 ) Computers are FISH!!I can tell when my opponents are using computers. They play like tactical powerhouses that could only come from years of practice--and yet they make moves that are positional nightmares, utter greed!
To see how much of a fish Fritz 7 really is, try this: turn off it's opening book and play the queen's gambit. It will accept the pawn and play into some gawdawful line involving ...b5 in a desperate attempt to cling to its ill-gotten gains.
OK, so Kramnik drew with Fritz in that match, and I'll be the first to admit that I'm no Kramnik. But here in correspondence land the rules are different. We are certainly allowed to use books and websites to look up the opening variations. In every game of chess somebody plays the "novelty," right? Computer novelties usually come at the very instant their opening book runs dry. And you know what? Their novelties usually STINK.
♡ 92 ( +1 | -1 ) IMOthere is no chess without humans playing it. No novelties. No interesting positional play, I think only then chess becomes interesting when humans are playing, long sight plans involved, good plans or bad plans. Tiny little errors in plans usually provide the most interesting positions.
Exchanging Queens is good against a computer. Otherwise it is good to keep much material so there is much to calculate for the computer.
Computer's weakness in positional play is best demonstrated with the move 12. - Bf8? in the game Kramnik vs. Deep Fritz
note the 9. Kf1! move, safely out of opening book! Computer has to start calculating early on.
♡ 121 ( +1 | -1 ) Brunetti v. ComputerI have always criticized those who use computers, and have posted to this end (being severely criticised myself for the effort). Just to recap, this is what I posted on another thread a few days ago. Although I was misinterpreted, I intended the content in a general sense:
'[M]ost have chess books, they surf the net, visit chess websites, they have computers, and when there is a difficult decision to make, the temptation to “get a little advice” is high. And I am not the only one to state this; Michael Goeller’s website* on chess resources has this to say about GK: “Play against others or their computer.”'
However, I find this challenge curious and strangely interesting (perhaps because it is open, and not just some player pretending to play himself whilst letting the computer do a large part of the hard work). In any event, to stir things up, although Brunetti is probably stronger ratings-wise (just how strong are these computers?), I am sticking my head out and stating that (if the guy/girl running the computer knows what he/she is doing), in a three-game match Brunetti will get his tail whipped. :) Gramario
♡ 17 ( +1 | -1 ) Of course it is....cheating to use a computer to analyze and suggest moves in a game that is still on-going. No, I would not like a forum to play in games that allow computers.
♡ 14 ( +1 | -1 ) Quite right,but here I seem to understand that there is a challenge that Brunetti and the Computer manager are both interested in playing out. Gramario
♡ 247 ( +1 | -1 ) I always find it interesting that some continue to claim beating computers is just a matter of having the right strategy, or when someone claims that having extra time will allow them to play at the same level as or better than top GMs like Gelfand or Kramnik (that is, they assume they can beat the computers that have humbled the best in the world, as long as they have enough time to analyze!). Time and time again top GMs have kept on losing or escaping with draws against the top programs and we still think there is a way to combat computers using some novelty? As if Kramnik and his trainers, using Fritz before his match, couldn't have figured this out? The fact is, he lost with adequate preparation. Yes, the computer made some dubious moves but it made brilliant ones as well. What human wouldn't have crumbled after the knight sacrifice that Kramnik unleashed after 43 minutes of thinking?
Sneaky, good comments, but I don't agree with you that computers are "fish" -- although, to be honest, I'm not even sure what that means. Kramnik had more knowledge of openings sitting at the board than most of us mortals would have possibly gained poring over books for several days at home. And you say their novelties stink, but I didn't see the 2nd best chess player in the world doing much about it. I'd give him the benefit of the doubt about whether Fritz's play stunk or not.
Also, during the Fritz-Kramnik games, many were surprised that Fritz wasn't so greedy. Computers have started to understand that not every free pawn needs to be grabbed; we saw that change of heart in this match. I can't remember which game, but there was a time where Fritz moved a pawn in front of the king -- I was surprised, because computers hate doing this because of some programmed "king safety" idea. This pawn move was a subtle positional play that was much more human than stereotypically computer-like.
You may be right, sneaky, that you could tell if you were playing a computer. But probably you could also tell if you were playing Kasparov as well. In both cases, I think any non-GM would lose easily.
♡ 12 ( +1 | -1 ) I agreeI don't think any human player here at gameknot would have any change against one of these strong computerprograms
♡ 12 ( +1 | -1 ) ok what about thiswhat if there was an id on gameknot that u knew was a computer. if u knew before hand would u play against it?
♡ 94 ( +1 | -1 ) Sarah with all my respect to tulkos, that there's no match between his CM8000 and cairo allowed to think a week per move. Cairo could beat CM8000 even in a 2 hours game! As I wrote before, in the slow games the human player is advantaged, especially if he's a 2400 ICCF player :) The only way to have an interesting game with cairo is to play it as fast as possible (3 days/move), to use the best software/hardware possible combination (Deep Fritz on a dual Pentium 1600, or Fritz 8 on a Pentium 2000, at least), allowing the computer to analyze for hours (at least 4, 5 per move). Even so, the human player keeps an advantage. Otherwise there is no match at all :)
Alex I think Brunetti is saying the same thing sneaky is say only more forcefully. If you are going to argue this point of view, your going to find many experts raining down on you and very few (if any) supporting you.
♡ 352 ( +1 | -1 ) CertainlyComputers are not fish. As we've seen, computers like Deep Fritz on superpowerful computers can match even the World Champion at chess. Some opening novelty or playing for a "positional" game will not always work, for example, while Kramnik got the better of it in the second game after Kf1, Fritz was decisively better after Qf4! in game 3 (despite the fact that it still lost).
However, computers tend to go way, way down in strength when analyzing positions at something like 2 weeks/move. The amount of time allotted appears to be directly proportional to a computer's strength, even for the strongest programs, as evidenced by blitz matches, programs tested on ICC and other servers, etc. Whatever Chessbase has done to improve Fritz, it still does not grasp basic positional concepts (as evidenced by games 1-3 of the match). Kramnik would be much better poised to exploit its weaknesses with a game at 2 weeks/move instead of a few minutes/move. Conversely, Fritz would have no sense of how to "exploit" a positional advantage against Kramnik--without cognition, computers have little chance of ever truly matching humans, even if they can simulate human thinking.
So, one of the factors that gives a human a big edge is a psychological factor, which plays more of a role in correspondence than in OTB chess (since more positions are evaluated). But the main factor is how humans and computers evaluate a position. Computers have to search every possible move. A computer engine will NEVER eliminate a move, even if the value of it is "minus 20 pawns". The only time a computer will stop searching a move tree is when it results in mate. Humans, on the other hand, understand that moving your queen right next to the opponent's king will just get it caputred and lose straight out. So they eliminate 95% of the possible search tree, looking at just a few moves each time. While a computer's evaluation algorithm is purely exponential (O^N), humans tend to be almost linear (O). This doesn't matter so much in short games, where computers can tear a human apart because they can "see" 8 moves ahead. But when N becomes large, that is, when the time allotted becomes 2 weeks/move instead of 2 minutes, computers markedly decline in strength.
There are things you can do to offset this, like create a "correspondence analysis" feature, as kingofpawns pointed out in another thread, but these have all had limited success at best, because computers simply will not eliminate moves from their search tree.
This is not to say that computers simply cannot win against a human (they're not "fish"). The fact is that even top-of-the-line computersare still only around 2500 or so in strength in correspondence matches. Fritz 7 (or CM8000), run on a mediocre PC, is significantly worse, and IMO, brunetti would have no trouble beating it, even with new features like correspondence analysis. However, computer programs are RAPIDLY gaining strength. While Fritz 7 may not be able to win a correspondence match against even middling players, who knows what Fritz 10 will bring? It could be that soon computers will simply overpower humans in sheer calculation ability.
♡ 68 ( +1 | -1 ) atrifix wrote: "A computer engine will NEVER eliminate a move, even if the value of it is "minus 20 pawns". The only time a computer will stop searching a move tree is when it results in mate. "
Well, you just exposed your lack of knowledge about computer chess engines.... Because this is not at all how computers work. They DO eliminate moves with value of "minus 20 pawns" and they DO stop searching a move tree if it doesn't result in mate.
In fact, one of the most sophisticated part of the chess-engine algorithm is the decision "when to stop searching a given line" - also known as "pruning"
♡ 50 ( +1 | -1 ) PruningComputers do prune, but it's not at all similar to the way a human does it--they can't eliminate moves without searching a specified depth down them (or else you end up with situations where computers will completely overlook sacrifices and the like). And most of the legal moves are still within a given search tree--"minus 20 pawns" was a bit of an exaggeration, but computers don't prune the large majority of the search tree, e.g., continuations that don't lose a large amount of material but are strategically egregious lines.
♡ 98 ( +1 | -1 ) The point isWhen given more time, computer cant improve its play as much as humans can. For example there are many correspondence chess IMs (GMs too?) who are in 1800-1900 range OTB (also before the times of computers). Of course its different if we are talking about computer cheaters - then I believe it depends on how strong the human player is. If a strong human player uses lots of time to combine his chess knowledge and calculation power of Fritz, Im sure he will be almost impossible to beat.
Personally I cant understand why people admire chess progs so much nowadays, I mean will people cheer for Fritz next time it plays Kramnik? The only thing I admire is that in more or less mathematical game, progs still arent clearly superior compared to best human players! Of course one day they will be, but who cares? My Volvo is faster than I am, but it doesnt make walking and running pointless.
♡ 196 ( +1 | -1 ) Computers really are fishyatrifix and vietnam_girl come to the defense of computers, saying "Hey, if they can beat Kramnik, Kasparov, etc., how can you say they are fish?"
Consider: Deep Blue's opening book was programmed with lines from ECO and from Batsford Encyclopaedia of Chess Openings. Well guess who's contributing the most to the Batsford book? That's right, Garry Kasparov! So his own analysis, the fruits of his fertile mind, were programmed into this machine so it could spew the moves back to him at the right time.
Now I can play Fischerrandom chess and make pretty good opening moves. I can find plausible pawn sacrifices in opening positions I've never seen before, and I'm not even a master-strength player. Computers on the other hand will NEVER gambit a pawn in an opening unless they've been specifically programmed to do so! (I'm speaking of true gambits, not tactical shots that recover the pawn later.)
So there is definitely some bridge of creativity between computers and people. Also, as correctly pointed out, computers do not benefit from extra time. Fritz perhaps goes 14 moves deep after so many hours, right? Leave it on for a week straight, and now it goes 15 moves deep. Big whoop-dee-do. That's not analysis, that's just spinning its wheels.
Compare that to me. I am ruminating over a position one night and suddenly I see something that I never saw before: A PLAN! I could sacrifice a piece in the center of the board in an effort to lock his defenses from his k-side. Follow it up with a swift attack on his king position and he'll be forced to play ...g6 or ...h6 to defend. Then comes the pawn storm against his weakened kingside and soon he'll be forced to give the piece back, with interest. Now whether or not my plan is sound is another matter, but the point is, computers don't think like this, and human creativity will defeat them in correspondence chess for a long time.
♡ 18 ( +1 | -1 ) SimpleI believe the stories about articifial intelligence of chess progs etc when they manage to develop a prog that calculates 60 moves max per minute and plays on GM level :-)
♡ 138 ( +1 | -1 ) Sneaky.When you write: "I am ruminating over a position one night and suddenly I see something that I never saw before: A PLAN! I could sacrifice a piece in the center of the board in an effort to lock his defenses from his k-side. Follow it up with a swift attack on his king position and he'll be forced to play ...g6 or ...h6 to defend. Then comes the pawn storm against his weakened kingside and soon he'll be forced to give the piece back, with interest", you make it sound as though you are shooting in the dark. But the real crux of the matter is that you have real (calculated) hopes of realizing tangible results from your gambit (tangibles which you outline, and which can also be programmed into chess software.) Moreover, at a high level, such a plan is only really feasible on OTB play where time-trouble and psychological aspects might play a more important role. In correspondence play, you might not find it such a good idea. I am sure that Capablanca (the chess machine) would never (perhaps almost never, as I cannot claim to know all of his games), indulge in such tactics as you describe without having very high hopes of what he could gain from such play and how he would exploit it when the time came. You might say his algorithms are firing on all cylinders. Gramario
♡ 144 ( +1 | -1 ) atrifix..actually, computer chess engines DO prune the vast majority of search trees - an increasing percentage with search depth. Only lines which fall between the "alpha-beta" bounds are searched fully, which ( with today's algorithms ) is a relative rarity. With a typical search to a depth of, say 10 ply, well over 99% of possible positions are pruned.
sneaky, you state that computers will NEVER gambit a pawn, but that simply isn't the case. A program can easily be given an evaluation function which will give certain positional considerations more than a +1 pawn positive score. It all depends on how the evaluation function was written... however, your point isn't lost.. research is being done on "plan construction" but the human mind is still far far superior to anything in this area.
peppe_l, I can tell what I admire about chess programs.. the skill that went into programming them. I have worked on chess programming myself, and I can tell you that while the chess side of me appreciates the sheer talent of someone like Kasparov, the programmer side of me appreciates the programming prowess, blood, sweat and tears that went into a program like Fritz. By the manipulation of 1's and 0's, some resourceful people have managed to create a program that can beat the most talented chess players in the world.
♡ 40 ( +1 | -1 ) WAKE UP.....ALL OF YOUusing computers? what's the bloody point? yeah - my computer (except i haven't got one, unless you count my old commodore 64...) beats your computer...wheeeee!!! great!! get my rating up!!! yeeee-haaaaa!!! something to make my decisions for me!! why don't you just forget gk and stay in and play with your little computer on your own... wake up all you monkeys using computers.... phew.....that feels better..
♡ 12 ( +1 | -1 ) Wake up, etcOK, Roblear, I think we get your point. But don't you find it ironic that you are using a computer to post your messages?
♡ 24 ( +1 | -1 ) hehe----A question on the computers that played kasparov and kramnick, did they have large databases of GM games available? and by the way, QG accepted is a interesting and usable opening, and theirs even a line that b4 (protecting it's "ill gotten gains") is played in.
♡ 48 ( +1 | -1 ) my opponent........whose rating is above 1800, thought that i was using computer aide when i made a good move. he even said to me that "u don't understand sh*t". why is it like that?! are we, 1600 rated players, can't sometimes think of a brilliant move?! i told him that if he feels he has been cheated, he can ask the webmaster to cancel our game (unfortunately [or fortunately] he can't, can he?)
Me, I don't care whether my opponent is using computer or not. I just want to play and want to improve my game. Period.
The anonymous nature of the Internet doesn't allow you to determine with certainty wether your opponent: - Is young or old - Is a guy of a girl - Is Swedish or Finnish (of course you can read those things in his profile, but there is no confidence they are true)
So the statement "I know nothing about my opponent" is an axiom inherent to any internet chess play.
All my opponents are anonymous chess-move-producing-entities to me. If any of them use a chess program, I will not be cheated, because I never assumed they aren't using a chess program! (of course I also never assumed they are using a chess program too... I never assumed anything about them...)
So, let's just play chess! And may the best chess-move-producing-entity win!
♡ 61 ( +1 | -1 ) agreeI think it is not much fun or gratifying to have your play dictate by a progam. The use of aids is a graduate one not absolute. Databases- computerprograms used as blundercheck f.i. Still there is a borderline, but crossing that gets the fun out of the game especially for yourself. What is the use of tormenting yourself with the question if in your opinion someone crosses that line?Just play chess and have fun in the game. If you can't stand that and want to be sure that a person goes only on his own strenght,that is he does not use the thinking efford of a better player(database) or a program, go to a chess club and play OTB chess!
♡ 21 ( +1 | -1 ) ..."The anonymous nature of the Internet doesn't allow you to determine with certainty wether your opponent: - Is young or old - Is a guy of a girl - Is Swedish or Finnish"
Hey, you can always spot the difference between Finns and Swedes! :-)
♡ 66 ( +1 | -1 ) Just Play!What is the use of tormenting yourself with the question if in your opinion someone crosses that line?Just play chess and have fun in the game. If you can't stand that and want to be sure that a person goes only on his own strenght,that is he does not use the thinking efford of a better player(database) or a program, go to a chess club and play OTB chess! -------------------------------------------
My sentiments exactly, Lex! There is no way one can tell if a person is using an engine, or is consulting a chessmaster friend, or is white, or is black, or is red, or is yellow, etc...etc...etc...
The best thing to do is to not worry about whether someone is cheating, and simply proceed to play the position on the board to the best of your ability.
♡ 74 ( +1 | -1 ) Just Play.Message: I think that no/one would disagree with the sentiments, "Just Play". However, this is somewhat getting away from the part of the thread which interested me. I believe that there was supposed to be a challenge between the Computer and Cairo and/or Brunetti. Regardless, of whether it is cheating or not, I want to see how it turns out, and I have already posted my prognosis, to wit: "Although Brunetti is probably stronger ratings-wise (just how strong are these computers?), I am sticking my head out and stating that (if the guy/girl running the computer knows what he/she is doing), in a three-game match Brunetti will get his tail whipped." :) Are they going to play or is someone backing out? Do I hear the sounds of Cluck, Cluck? Gramario
♡ 266 ( +1 | -1 ) Shooting in the dark?!Gramario apparently finds my hypothetical example of a human strategy a bit far fetched. To recap: I describe a plan (but offer no position) to sacrifice a piece in the center of the board in an effort to lock his defenses from his k-side, followed by a swift attack on his king position and he'll be forced to play ...g6 or ...h6 to defend, then comes the pawn storm.
"you make it sound as though you are shooting in the dark. But the real crux of the matter is that you have real (calculated) hopes of realizing tangible results from your gambit (tangibles which you outline, and which can also be programmed into chess software.)"
Well, I suppose it COULD. But nobody has! This kind of thinking is not what chess software does. Keep in mind that a true sacrifice cannot be analyzed to the point where you prove that you regain the material. I'm speaking of a positional sacrifice, where the benefit of the sacrificed piece is an interference in communication.
"Moreover, at a high level, such a plan is only really feasible on OTB play where time-trouble and psychological aspects might play a more important role. In correspondence play, you might not find it such a good idea."
That may be true. I guess it depends upon the position, you seem to be very confident about your appraisal even though we haven't seen a single diagram.
"I am sure that Capablanca (the chess machine) would never (perhaps almost never, as I cannot claim to know all of his games), indulge in such tactics as you describe without having very high hopes of what he could gain from such play and how he would exploit it when the time came. You might say his algorithms are firing on all cylinders."
The magician of Riga, Mikhail Tal, certainly would indulge in such tactics. In fact when I described this plan I was describing the famous Tal vs Larsen game
Take a look at this game:
I really don't understand what you find so preposterous about this strategy. Is it the idea of sacrificing the piece in the center, in exchange for disrupted communications? It's hard to put a price on communications, my friend. If I can give up a knight so that your pieces are locked from the defense of your king, I'll do it. If you take out the part of sacrificing a piece, you're looking at the way hundreds of thousands of games are won! Threats against the K-side lure the opponent into moving a pawn, which of course always constitutes a weakness. (The Yugoslav Attack, or the Keres Attack, for instance.)
♡ 40 ( +1 | -1 ) programsPerhaps gameknot should allow five or six strong programs to become members and play the ratings game like everyone else. No aliases, they will be called by their names fritz6, fritz7 for instance. Will they reach the top spots? Will they get into the top ten? Will they find this type of chess (days to move) to tough and opt out?
♡ 136 ( +1 | -1 ) BrunettiFirst: "Moreover, at a high level, such a plan is only really feasible on OTB play where time-trouble and psychological aspects might play a more important role. In correspondence play, you might not find it such a good idea."
Not necessarily true. Some of the most interesting and crazy-looking sacrifices come from correspondence play, usually when they've been accurately calculated to the point of regaining the material.
As for gramario's comments: "'Although Brunetti is probably stronger ratings-wise (just how strong are these computers?), I am sticking my head out and stating that (if the guy/girl running the computer knows what he/she is doing), in a three-game match Brunetti will get his tail whipped.' :) "
Of course a computer program's analysis, compounded by the operator's analysis, is stronger than the computer alone. I think brunetti has excellent chances.
"Are they going to play or is someone backing out? Do I hear the sounds of Cluck, Cluck? "
Between organizing the olympics, TTs, playing all his 200some games, etc., brunetti seems to be pretty busy these days. Give him a break :)
♡ 28 ( +1 | -1 ) AtrifixCertainly, I'll give him or anyone a break. I don't want to hassle him or anyone. Further, Brunetti's life is his own affair and I can't be expected to know how many committments he has or doesn't have. I was just responding to a couple of posts he made in which it seemed that a match was on. If it isn't, so be it. No sweat at all. Gramario
♡ 58 ( +1 | -1 ) Program usersIf somebody use opening databases its ok we can say he wants to learn opening and he want go to midle game with equal position and than he play alone to see how strong is his strategy and tactic are.But when somebody "ask" his software for good move in hard position i think thats cheating.In some games what i played i see my oponnent is weaker player than me he lost one or two pawn or he is in wery bad position and what hepening?Suddenly he start play fast whit thinking time few second and he in that few second find best moves i think thats cheating.
♡ 33 ( +1 | -1 ) GramarioIf you're looking for the match thread, it can be found at gameknot.com/fmsg/chess/1153.shtml. However, brunetti hasn't moved in months--he probably forgot about it.
♡ 3 ( +1 | -1 ) GramarioOK. Thanks I'll have a look at it.
♡ 316 ( +1 | -1 ) Going back to one of the discussions on this thread. The point I was making is that I believe some people seriously underrate computers' playing strength. I'm not happy about this and always fervently wish that the human prevails in any man vs. machine contest. I remember some time ago in this forum and forums at other chess sites that people predicted Kramnik would have no problem whatsoever with Fritz; but as much as I hoped that wouldn't be the case, I had great doubt. And it turned out that even with all his time using the program beforehand, Kramnik still only managed a draw for the match. During the games, Fritz showed remarkable improvements (for a computer) in the way it played positions. I'm sure that many of the views people have of the weaknesses of programs are 2 years or so out of date (a lifetime, of course, in computer technology).
Fritz still made some mistakes that even an intermediate player like myself knew were weak, but it made up for it with moves that were genuinely brilliant.
Someone said here that in correspondence play even "middling" players could beat computers. Not true, I think. What that means is that middling players can make moves of GM level strength as long as they have the time to push around the pieces. I've never known that to be the case. I believe it was sneaky who said this. But I'm not sure he would really suggest that if an 1800 player had a week per move that they'd play at master level strength. Looking at a board for any length of time isn't going to somehow give you the insight, the intuition and knowledge through experience to play at 2200.
Also someone said: "with all my respect to tulkos, that there's no match between his CM8000 and cairo allowed to think a week per move. Cairo could beat CM8000 even in a 2 hours game!"
Let me just say that I have a GREAT deal of respect for Cairo as a player. He's truly remarkable and I've looked over many of his games. But in beating CM8000 in a 2 hour match easily, as was suggested, he would be doing something that GM Larry Christiansen was unable to do during a four game match of 2 hour games with Chessmaster (CM won +2=1-1). To beat CM in a 2 hour game, Cairo would have to play about as well as, probably better than, the 3 time US chess champion.
Anyway, I don't want to be in the position of being a defender of computers. My only interest is that people shouldn't invest any real emotional interest in man vs. machine contests, because we're sadly going to be disappointed again and again. I think people assume that there's some kind of honor at stake as GM after GM confronts computers (usually losing), but the future of chess lies elsewhere. And I believe the game is only harmed as the only public media coverage chess gets nowadays are these computer matches.
♡ 87 ( +1 | -1 ) Sarah"But I'm not sure he would really suggest that if an 1800 player had a week per move that they'd play at master level strength. Looking at a board for any length of time isn't going to somehow give you the insight, the intuition and knowledge through experience to play at 2200."
Lets say you organize a match between 1800 and 2200. 2200 has 90-150 mins per game + usual OTB rules. 1800 has week per move + usual correspondence chess rules (permission to use opening books + databases etc...). Are you sure 2200 will win? :-)
"I think people assume that there's some kind of honor at stake as GM after GM confronts computers (usually losing), but the future of chess lies elsewhere. And I believe the game is only harmed as the only public media coverage chess gets nowadays are these computer matches."
♡ 380 ( +1 | -1 ) Sarah"Going back to one of the discussions on this thread. The point I was making is that I believe some people seriously underrate computers' playing strength."
"I'm not happy about this and always fervently wish that the human prevails in any man vs. machine contest. I remember some time ago in this forum and forums at other chess sites that people predicted Kramnik would have no problem whatsoever with Fritz; but as much as I hoped that wouldn't be the case, I had great doubt. And it turned out that even with all his time using the program beforehand, Kramnik still only managed a draw for the match. During the games, Fritz showed remarkable improvements (for a computer) in the way it played positions. I'm sure that many of the views people have of the weaknesses of programs are 2 years or so out of date (a lifetime, of course, in computer technology).
Fritz still made some mistakes that even an intermediate player like myself knew were weak, but it made up for it with moves that were genuinely brilliant. "
Not entirely true. The computer played a few very good moves, but, well, when a supposedly 2700-2800 strength computer can lose 2 drawn endings in a row in a couple of moves something is wrong. Fritz showed a generally weak understanding of a positional strategy--certainly superior to that of an 1800 player, but nowhere near Kramnik's ability. Kramnik was not on the top of his game in the match, and the last two games were particularly weak. Fritz's saving grace was, similar to Deep Blue, its sheer calculative abilities (Game 6) and its intimidation factor (Game 5). It is difficult to say whether Fritz or Kramnik will win more games on average, but it's obvious that Kramnik's best game is clearly superior to Fritz's best game and that Kramnik has a far deeper understanding of chess.
"Someone said here that in correspondence play even "middling" players could beat computers. "
My interpretation of "middling" players would be people like GM Larry Christiansen, GM Larry Evans (as compared to Svidler and Kramnik), or people like brunetti, cairo, etc.
"Not true, I think. What that means is that middling players can make moves of GM level strength as long as they have the time to push around the pieces. I've never known that to be the case. I believe it was sneaky who said this. But I'm not sure he would really suggest that if an 1800 player had a week per move that they'd play at master level strength. Looking at a board for any length of time isn't going to somehow give you the insight, the intuition and knowledge through experience to play at 2200. "
Simply not true. Correspondence ability varies from person to person. An 1800 OTB might play at a 1400 level in correspondence or a 2400 level--both have been known to happen. brunetti is only a candidate master OTB. Computers, however, do not show any significantly marked improvement in correspondence play. The exponential nature of the search algorithm leads to drastic improvements at searches of 1 or 2 ply, but barely any at 20-21 ply.
"Also someone said: "with all my respect to tulkos, that there's no match between his CM8000 and cairo allowed to think a week per move. Cairo could beat CM8000 even in a 2 hours game!"
Agreed. With all due respect to cairo, he probably has little chance in a 2 hour match :)
♡ 54 ( +1 | -1 ) sarahIt was Brunetti who said Cairo would beat CM 8000 in a two hour time limit. You'll have to admit, when it comes to programs, Brunetti has been around the block a time or two. I thought it was common knowledge that when an otb player, in the 1800 range, goes into cc he gains on the average of 300 pts. Just the relaxed nature of playing in my home, in my underwear, smoking, eating or listening to my favorite music, running to the toilet whenever I please---taking a nap will add a hundred rating pts. and I haven't even begun to elaborate.
♡ 52 ( +1 | -1 ) correctionChristiansen played CM 9000 (not 8000) and there definitely was an advantage to CM 9000 in the playing conditions. Four games in two days! Two on Sat and two on Sunday. International chess tournaments are one game per day because human beings need rest---Machines don't.
If Cairo is playing 8000 at one game per day, his chances improve considerably. Can you imagine playing a really tough game in the morning and then having to play another tough game the same day. That would be tough on a young master and Larry is no spring chicken.
♡ 120 ( +1 | -1 ) I think this is a good...discussion. However, Sarah is absolutely correct in everything she has said. I find it amazing all the claims made about humans in correspondence games with computers. If a human player is given two weeks to move, how is that an advantage that makes it easy (or at least probable) that the player will be Fritz 7 (or Fritz N)? The assumption is that the human correspondence player has the theoretical knowledge of a GM and the extra time allows for his or her calculational ability to match a GM. I simply don't buy it.
Sarah is also correct in her view that chess programs will inevitably become invincible with the only hope being a draw.
Finally, there isn't a thing we can do about it. Not all people are honest and as Fritz becomes stronger and stronger we will see the effects at GK. I actually think that it can lead to the development of a more social chess community at GK. People don't like to play machines, so the best way around it is to get to know a community of people you feel comfortable with playing.
I know of one case here where some of the top players here believe another top player is using Fritz 7 and most are losing or at best drawing against Fritz. So, Fritz 7 is good in correspondence chess.
♡ 190 ( +1 | -1 ) KOP"The assumption is that the human correspondence player has the theoretical knowledge of a GM and the extra time allows for his or her calculational ability to match a GM. I simply don't buy it."
Not really my claim, although such a situation is possible (brunetti, etc.) But programs still show no dramatic improvement in correspondence play, whereas most humans do. This is not to say that if you take an 1800 correspondence player and have him play against Fritz 7 he'll somehow make GM strength moves instead. But if you take strong correspondence players like cairo, they should still be able to beat machines like Fritz 7 and CM8000. Eventually this will probably change, but machines in correspondence play are still not approaching the highest level of playing.
"Sarah is also correct in her view that chess programs will inevitably become invincible with the only hope being a draw. "
Unlikely in the near future, but possibly inevitable. If computers ever come remotely close to solving chess, of course, they'll become invincible, but then why will anyone still play? I do agree that since the strength of programs in increasing far more rapidly than that of humans, computers will eventually overtake humans, possibly quite soon.
"I know of one case here where some of the top players here believe another top player is using Fritz 7 and most are losing or at best drawing against Fritz. So, Fritz 7 is good in correspondence chess. "
This is an entirely fallacious conclusion :) First of all, you have no proof that the player is using Fritz 7. Secondly, if he is, he may be combining his analysis with Fritz's. When you combine your own analysis with a program's you get, well, an uberprogram.
♡ 148 ( +1 | -1 ) atrifixI certainly agree with your last point. But, this is also how computer programs will improve beyond human capabilities. Rules of thumb and the knowledge greater than any one GM can be incorrportated into a chess program. Computational speed and seach is only one aspect. The aspect that will allow computer to greatly go beyond humans is there ability to integrate all knowledge and available games played. Programs reflect the creativity of their programmers, so we will also see more and more "human" creativity in programs.
I was looking through players greater 2100 and I found one player who clearly could not add to a program when he switched to using one. He played here for half a year at 1100 to 1300 level (fluctuating up and down). Then he suddenly improved to the 2000 level, but he did not do so well against some of the top players here, though he beat one of the best. Probably, he used a program with no input from him, but we don't know what program he uses, so we can't draw conclusions yet about the best programs.
As someone suggested, it would be nice if we could have volunteers runing say Fritz 7 under the name Fritz7, and then each could judge for themselves how strong Fritz 7 is in correspondence.
I don't have a PC so I don't have Fritz, but I would love to know what it is like to play Fritz 7 in correspondence. Is there anyone willing to play an email correspondence game where they use Fritz 7?
♡ 548 ( +1 | -1 ) On computers in correspondence chessBack in 2000, Steve Ham, a US Master and correspondence player, challenged Fritz 6 and Nimzo 7.32 to two games apiece. He scored two draws against Nimzo, and one draw/one loss against Nimzo. Now, Steve Ham classifies himself as primarily a technical player, and the games were rather active, tactical, and dynamic, so this may have had something to do with his performance. Still, the computers played okay. See www.correspondencechess.com/campbell/ham/ham.htm
kingofpawns, regarding your comment about being able to program rules-of-thumb and general knowledge into a computer, in theory this can be done. But in practice, this is difficult. Some of the earliest chess engines tried to do just this; they employed heuristic algorithms to mimic (on some level) how a human understands and learns about chess. They didn't do so well, and that approach was abandoned in favor of the current "evaluation function+brute-force search" methods.
In my mind, the weakness of a chess engine is precisely that it doesn't use heuristics. It doesn't learn, it doesn't understand anything. It blindly follows its programming to the letter and never deviates. A bishop is always worth 3 (3 what, I have no idea) or whatever. A fianchettoed bishop that eyes the entire diagonal is worth more, 3.1, 3.2, whatever. Always, in every position, every single time. But that's not how chess works; the evaluations may be useful as average evaluations, but they don't apply in every position. Similarly with an isolated d-pawn, which might be evaluated as -something. Or a strong knight on the 6th rank that gets assigned a +something score. In a way, modern computer engines play like super-Tarraschs or super-Nimzos; they have a (albeit complex) set of rules that they use to evaluation a position, and stick to them dogmatically. But even Tarrasch and Nimzo understood that general rules didn't apply to every position, and they deviated from their "programming" when necessary. Well, sometimes, anyway.
To use a couple of well-worn examples, try tossing the Najdorf Poisoned Pawn into a chess engine, and most will come back with the evaluation "Black is better." Analyze 17. Rxb7 and 18. b4 in Kasparov-Shirov, Horgen 1994. "Bad moves, Black is winning." And so on. The computer simply doesn't understand that its "rule-of-thumb" evaluation is not applicable in the particular position at hand.
"All" that's needed to beat a computer chess engine is to determine what values it assigns to various positional factors and then play into positions where these evaluations are not applicable. No mean feat, to be sure; programmers of today's top engines are not just going to release the source code for us to look at. But figure out how to beat it once, find just a few cases where the programmed evaluation function is way off the mark, and you can just keep playing the same thing over and over, beating it every time. Because computer chess engines don't learn. They don't understand why they win or lose. And they never correct their flaws; they have to wait for human programmers to reprogram them with a new "average evaluation" which will not necessarily be any better; just applicable in some situations and not in others, as the previous evaluation function was. Although, hopefully, the new evaluation will be applicable in a wider range of cases.
The alternative method traditionally employed to beat computer engines was simply to make moves that had implications that extended beyond the engine's calculation threshold. If the machine could normally see 10 moves deep, and you played a move that is crushing on move 12 in the variation, then you could win. But this obviously gets harder as computing power increases and engines search more deeply.
In the end, computers engines have weaknesses just like human players. But the difference is that human players can potentially learn to correct their weaknesses while chess engines cannot. That, in my mind, it the reason why, barring some breakthrough in which someone discovers an algorithm to solve chess, chess engines will never be invincible. So long as engines work off of a rule-based evaluative function, players will always be able to find positions that break the rules, those '!' and '!!' moves, whether it be in correspondence or OTB play.
However, I don't buy that humans suddenly get an advantage (or close the gap in playing strength) over computers when playing correspondence. While additional time usually does not benefit a chess engine significantly (it usually takes progressively longer to search the next ply of moves than the previous ply), all that extra time may not benefit a human player either. Time pressure is lifted for the most part in correspondence chess, which confers a psychological advantage to the human player. But a human is limited by what he "sees" in the position. If all a human player can see is, say, all relevant 8 move variations (just as an example), and it only takes an hour to analyze all this, rehashing the same variations for hours on end isn't going to do much good, except perhaps as a way of reducing errors. If it simply doesn't cross my mind at all (for whatever reason) to perform a rook lift that wins material , no amount of staring at the board is going to help me.
♡ 27 ( +1 | -1 ) caldazarI don't know what you're smoking but pass it around---thats some good sh**!
If Kasparov had days instead of minutes per move, he would have beaten Deeper Blue 6-0. Hell, he would have beaten Deeper Blue 60-0!! Same with Kramnik!
♡ 235 ( +1 | -1 ) tonlesuPerhaps, perhaps not, although what you say is likely to be true, at least with respect to Kasparov/Deep Blue. (maybe not 60-0, but my hunch is that Kasparov would likely have a favorable score). Big deal; the better player would score more points; that's hardly shocking. Or are you claiming that here that Deep Blue demonstrated itself to be the more skillful player in those matches? And if so, on what basis? Because the machine scored one point more in a 6-game match where Kasparov deliberately avoided his strengths?
I don't see what this has to do with anything I posted, anyway. The issue I was addressing was that some seem to be claiming that correspondence chess favors humans by its very nature, that in a correspondence setting, the extra time somehow benefits the human player more than the silicon machine on the apparent basis that a human, given sufficient time, can always out-calculate a machine. That, as long as we human players have extra time on our hands, we can always extend our calculated variations one move deeper, we can always find something else relevant to discover, see, and understand, whereas all the poor computers can do is keep looking at an analysis tree that branches 20 ways at every point. This is nonsense; if that were true, we could solve chess. Analysis done by human players branches too. Or at least mine does, looking at my notes on the games I'm playing right now. Maybe not in 20 different ways each move, but certainly in a couple of ways per move (depending on how forcing the continuations are, of course). So while it takes a long time for a computer to go from 20 ply deep to 21 ply deep, it also takes humans a long time to make the same step. Nor is extra time a guarantee that humans will make fewer errors; as I posted before, if a player (or machine) fails to consider a candidate move because it was simply discarded (through alpha-beta pruning for the machine or through experience, intuition, and judgement for a human), then no amount of extra time is going to remedy that situation. So the extra time in correspondence chess does not necessarily favor the human player outright, in my opinion. It all depends on the skill of the player.
♡ 23 ( +1 | -1 ) Let me get this straight, you went through many hundreds of words to finally say "...correspondence chess does not necessarily favor the human player outright,in my opinion. It all depends on the skill of the player."
Sir, are you taking your medication?
♡ 9 ( +1 | -1 ) Tonlesu.Man, you are a card! You really made my Sunday - I laughed so much I almost choked. Gramario
♡ 120 ( +1 | -1 ) It's funny...But those who say it is so easy to beat chess programs in correspondence never back up what they say with rational argument. I found caldazar very interesting. The only thing I disagree with you is with the inclusion of heuristics. We know there is one way that will work in small domains such as chess and that is to keep adding special heuristics for different contexts. Doug Lenat has been trying to do this for the whole of human common sense with marginal success, but chess is a much smaller domain. I suspect that is what the best program developers are doing. When a problem is found for evaluation a class of positions, new evaluation heuristics are added. Learning is also theoretically possible using genetic algorithms. So, inevitably, chess programs with have chess knowledge bases greater than anyone human.
tonlesu I know you will want something strong to smoke after reading this, but I don't know where I can get something strong enough for you! :-)