Our Computer Overlord Speaks

Longer time controls? That means the humans will cheat. Oh, wait a second... that was the wrong thing to say.

45' 5'' is a rather fast time control. Frankly, I'm not surprised that humans are of little assistance to the rybkas at that speed. Certainly I wouldn't jump to any doomsday conclusions.

So chess has been "solved" by computers. How swell. Maybe it's time for people to actually take up a real game like GO for which computers and their programmers haven't completely figured out.

Even the original modern-day thinkers like Shirov, Morozevich, and Nakamura can't beat computers. Screw chess!

I've been thinking for a while about the man/machine situation and want to float a few comments - hopefully for reaction or rebuttal.

It seems to me that machine/man computer matches are inherently unfair to the human; computers have programmed in VAST opening books and furthermore they have, in effect, the ability to 'move pieces' in that they are not hampered by mental-visualisation skills in analyzing positions in their 'trees'.

I believe that future matches should allow the human full access to an opening library; furthermore, I believe that humans should be allowed to 'move pieces' ie. set up potential positions on boards, including electronic boards. In a match like this, it comes down ONLY to a contest on analytic and positional skills. I still believe that top grandmasters would have an edge.

The recent free-for-all tournament is a step in the right direction, but the human opponents were hampered by time controls that prevented deep analysis. That is the point - as computer processors get faster and faster it is clear that human players will NOT be able to keep up - but computer algorithms are STILL limited by making decisions in unclear positions where their trees cannot see all outcomes - I still expect that in this sort of analysis humans would prove to have the edge.

Unfortunately we will never know if nobody arranges matches of this sort. Maybe I'm wrong; I'm open to reactions.
Dan

doomsday for chess my fanny. Computer chess may be entertaining to many, but it's not a substitute for real human vs. human chess any more than cheesy programmed drums and electronic sampling can replace music made by humans. Chess with a computer is much like sex with an inflatable doll I presume. Of course, some prefer their silent plastic lovers I imagine, but there are an awful lot of diehards who need human contact for maximum enjoyment.

i have a blow up Catherine Zeta Jones and I'm always telling her she's better then the real thing.

Dan G.'s points harken back to a comment I made on another thread, about how someone a few years ago had explained convincingly why a strong human can be expected to beat any computer at a correspondence-chess time control (i.e. 3 days per move, rather than an average 1 minute, as in the just-concluded Freestyle tourney, or the average 3 minutes per move of classical OTB chess).

On the earlier thread I'd forgotten who the "someone" was who made this case. In fact it must have been Arno Nickel, the correspondence GM who went on to crush Hydra in a (computer-assisted) match, but came in behind Hydra in the finals of this year's Freestyle event.

Nickel's argument in the article I recalled reading was in essence an elaboration of what Dan G. just wrote: since chess is far from being "solved" from a mathematical standpoint (the number of possible positions and games still being many, many orders of magnitude beyond the capabilities of even the next 5 generations of computers beyond today's), there will always (within our lifetimes at least) be "unclear positions where their trees cannot see all the outcomes."

Longer analysis times allow the computer to extend the search tree a bit. But the underlying math is such that, adding more time brings diminishing returns. At the point where the beast is already looking 20 plies deep, an additional hour of search time might extend the search by 1 ply; but to get another ply beyond that would take TWO FURTHER hours; adding a third ply would demand FOUR more hours, etc.

In contrast, the human uses judgment to hone in on specific paths for in-depth analysis. Thus adding more search time yields progressively ACCELERATING, rather than diminishing, benefits.

It is reasonable to believe that a 1500 + 2300 < 2300, when that 1500 has final say in what move is to be played. Pretty soon it will be embarrassing for humans to play in any event with computers (either by assisting them, or competing directly against them). I hope that doesn't mean human chess is over; I plan on entering a tournament later this year, after a 10 year hiatus. But enforcement against computer cheating in human tournaments will become harder and harder, as the technology becomes smaller and less detectable.

Logical flaw in your reasoning, Maury: The 1500 human can't be shown all the lines and ideas that the 2300 human saw when deciding which move he preferred. But in Centaur (aka Freeystyle) chess, the human sees all the important calculations and sub-variations utilized by the engine he is cooperating with. So it's reasonable to expect the human can only add, rather than subtract, value, to a computer's output, notwithstanting that the human would lose to the computer in an unassisted, head-to-head match.

Thanks for your comments, Jon (I missed the earlier post). It sound like we agree, along with the postal players.

So why is this not happening? Matches where top players use their analytical skills to defeat computer algorithms - where the human in not hampered by limited time and limited book knowledge can only be GOOD for Chess; the games will be better and techniques will advance! Is it simply that they are less interesting for the paying spectators, who seem to be there just waiting to see the human foul up and lose to the mechanical monster?

Remember that game that Kasparov resigned against Deep Blue, assuming that the computer had checked for a perpetual? Do you think that he would have resigned if he had more time AND the ability to move around pieces on a chessboard?

The old-fashioned man-machine matches no longer are interesting for me; even if Kramnik manages to survive his Fritz match, I'd attribute that to Fritz not being that strong (compared to Hydra, Zappa, and the 'new' breed) and also to the fact that Kramnik will be spending months practicing lines against his home copy of Fritz. One might almost say that this is unfair to the 'poor' computer!

Games in which a human beats a computer based on positional play - where tactical inaccuracies are virtually non-existent and where opening knowledge is equal are, and will be more entertaining to me, as a spectator, and I believe they will benefit Chess play as a whole.
Dan

Dan,

Why aren't man-machine matches with much longer time controls (more in line with correspondence chess) and/or altered rules being planned? (such as, letting the human move pieces on a physical board, or on an electronic board, presumably with no engine running) Probably more inertia than anything else. I suspect it won't be long before what you're talking about pops up on the scene.

Over the past 25 years, as computer software and hardware steadily advanced from the level of club players, to low-level professional (~2400 by the late 1980s -- but only for dedicated machines running custom-built software), then by the mid-1990s to be competitive with the strongest GMs, first in speed chess and finally under classical time controls -- throughout that lengthy period, what could have been more natural than pitting these silicon upstarts against top humans head-to-head under the sort of playing conditions everyone has been long familiar with?

Even after Deep Blue beat Kasparov, the question of parity between man and machine was hardly settled. So it still made perfect sense (in terms of sporting interest) to have computers face GMs without seriously tinkering with match rules.

It's only since Hydra-Adams, really, that's forced us to consider whether machines really have eclipsed us to the point where the traditional kind of matchups no longer make any sense (like a man running a footrace against a car, as someone said).

So, give it a little time; it's been less than a year since that monumental barrier was crossed for the first time. Yes, alternative formats (especially, going meaningfully beyond classical time controls) do pose logistical and/or commercial problems; but I imagine they can be solved through creative thinking.

One interesting alternative someone proposed on another thread here: Instead of trying to level the playing field by letting the human match the computer's calculating power (i.e., Freestyle chess and the like), how about attempting the converse? Create a match where the engine's search depth for generalized (i.e. all-possibilities) searches would be restricted by agreement to some pre-set level, corresponding to the calculating capacity of an unassisted human? If a strong GM, that probably would lie somewhere between 6 and 10 plies. The engine presumably would be allowed to extend the search beyond the standard 6- or 10-ply threshold, like strong humans do -- but such extended searches would also be restricted to a pre-set number or proportion of lines from any given position (again, designed to mimic human limitations).

The point of this exercise would be to see whether engines really "understand" chess as well as GMs do, as opposed to understanding nothing more than brute force. (That was a subject of intense debate on another thread.)

A couple of quibbles: Regarding opening preparation, this actually favors the human over the computer -- at least if the human is a world-class GM. Humans can innovate, while unassisted computers can't. (Yes, Hydra came up with at least one legitimate opening improvement during its match with Adams. But it was allowed to make that move only because it arose in pre-match prepration, and Hydra's human associates decided they liked the move enough to add it to the computer's "book" and make it the first choice in the given position. Had Hydra relied solely on pre-existing opening "book" knowledge, it could not have played or even looked for the improvement it found.)

Also, while I've been the most vocal proponent of the idea that there is such a thing as "positional play" that engines thus far fail to "understand" beyond the level of most 1800-players, the wording of your final paragraph makes even me feel a bit queasy. Even I cannot deny that chess is 98% tactics. So, strictly speaking, there is no such thing as a purely "positional" game "where tactical inaccuracies are virtually non-existent." Scratch beneath the surface, look into the paths that weren't taken in the actual game, and you will ALWAYS find tactics (and therefore, potential tactical inaccuracies). You don't see them played out on the board, but, like germs, or jet propulsion, or blackmail, they are there below the firmament, subtly driving and directing the things you do see.

Hi Jon,

I'd rather see conditions that improve the human's play vs. conditions that decrease the level of the computer's play. Also, the positional vs. tactical debate has raged on for years and is not really what I'm trying to address, but what I mean is that without the occurrance of tactical innacuracies (sometimes called 'cheap shots') then winning a game positionally refers to maneuvering to create a superior and winnable endgame - kind of like Capablanca's style; it is this type of game where I'd be gratified at seeing computers proved inferior (as their search trees surely don't extend for the 20 - 30 moves between the middlegame and the ending). Naturally I agree that top level Chess games have many tactical lines that only 'occur in the notes'.

Regarding 'book' knowledge it seems fair that since computers have pre-programmed opening books the size of ECO then humans should be allowed a library as well, though I concur that humans should be superior at generating new variations in opening lines. Again, one might propose restricting a computer's 'book' as with restricting it's ply, but I'd rather see better Chess by both parties.

Back in the old David Levy/Gary Kasparov days top players used to look forward to man/machine matches as a way to prove human (and they're own) superiority; once computers became too strong, most top players use the disclaimer in interviews that 'these matches aren't of interest, and computers are only useful for training'. I don't think they'd be making disclaimers if computers could still be beaten (under old playing conditions); give them the chance to improve the quality of their games to the point that there is a reasonably good expectation at victory and I think the matches will start again.

As for logistics from a spectator and money making standpoint, well obviously postal time controls are out of the question; but by extending the times by a few hours and, for the spectators, having several matches played simultaneously, these games could prove to be popular exhibitions.

I guess I'm being a bit selfish here in saying what "I'd like to see", but what can I say - as a lifelong class B player and infrequent competitor I find that I'm more of a spectator than anything else!
Dan

I for one welcome our Computer Overlords...

If computers were to obtain ratings, they should do it like other players, through tournaments and matches like 'Chinook' did for Checkers. If their rating is sufficient, only then should they have the privilege of playing against GMs.

Mig wrote: "GM + computer < computer"
volte-face
The evidence has been there for some time.

It always seemed to me that Man vs Machine matches are characterized by unfair terms for the human. So were i a top GM i would not accepted to play with such conditions. It is a matter of dignity to not accept to be humiliated for a couple of bucks, as Adams did..

Does anyone here knows about game theory ? If yes, please correct me: i think that chess belongs to a class of games that theoritically can be solved. Of course that is difficult, nevertheless matter of time. So the question is when this happen would chess keep his appeal? I am not sure. I speculate that chess will face a slow death because elite chess would have nothing interesting to post (everything would be written down to a database). So chess would be another game like backgammon or monopoly is.

Another hot issue that will arise soon is electronic dopping. If it is the case that anyone can have Fritz in his mobile phone , the consequences would start to appear. Because in an elite tournament maybe it is probable to apply some control over players. But in tournaments like us, amateur players participate, credibility of results will be a utopia.

Sorry for the pessimistic eschatology but that are my opinions . Hope to prove wrong!

It always seemed to me that Man vs Machine matches are characterized by unfair terms for the human. So were i a top GM i would not accepted to play with such conditions. It is a matter of dignity to not accept to be humiliated for a couple of bucks, as Adams did..

Does anyone here knows about game theory ? If yes, please correct me: i think that chess belongs to a class of games that theoritically can be solved. Of course that is difficult, nevertheless matter of time. So the question is when this happen would chess keep his appeal? I am not sure. I speculate that chess will face a slow death because elite chess would have nothing interesting to post (everything would be written down to a database). So chess would be another game like backgammon or monopoly is.

Another hot issue that will arise soon is electronic dopping. If it is the case that anyone can have Fritz in his mobile phone , the consequences would start to appear. Because in an elite tournament maybe it is probable to apply some control over players. But in tournaments like us, amateur players participate, credibility of results will be a utopia.

Sorry for the pessimistic eschatology but that are my opinions . Hope to prove wrong!

Christos may be right, but he is ahead of his time. The present problem is that the ( presumed few ) amateur cheaters do get caught, but there is no agreement on how, or even whether, they should be punished.

Nick,

I disagree about the "presumed few" amateur cheaters. I'd like to go into more detail, because of what happened at a recent tournament in Reno that I participated in, but this thread may take off in a direction not intended. When the right time presents itself, I'll unload.

We know the names of the computer portions of 7 of the 8 participants. Does anybody know which computer program ciron (ICCF GM Arno Nickel) used?

If it is a secret, I understand!

chesstraveler,
i would like to hear your story.
Anyway the cheaters may be not so few and may be not amateur. Does anyone know if Naiditisch was caught cheating in chessbase? (i think that i heard something like that...)

"One interesting alternative someone proposed on another thread here: Instead of trying to level the playing field by letting the human match the computer's calculating power (i.e., Freestyle chess and the like), how about attempting the converse? Create a match where the engine's search depth for generalized (i.e. all-possibilities) searches would be restricted by agreement to some pre-set level, corresponding to the calculating capacity of an unassisted human? If a strong GM, that probably would lie somewhere between 6 and 10 plies. The engine presumably would be allowed to extend the search beyond the standard 6- or 10-ply threshold, like strong humans do -- but such extended searches would also be restricted to a pre-set number or proportion of lines from any given position (again, designed to mimic human limitations)."

IMO, this is a sad proposition - nothing more than a way to cripple the strength of a computer to give humans better chances. Would you feel good about beating a stronger human opponent who was saddled with similar restrictions solely for the purpose of bringing him down to your level?

Yes, computers have significant advantages in processing speed and large databases which can be recalled without error. Does this constitute an unfair advantage? If you answer yes then the human vs computer game becomes completely meaningless cause you can then allow yourself any measures you like to "level the playing field". Whether such measures are designed to increase the strength of the human or decrease the strength of the computer is irrelevant. BTW, judging by what many people on this site seem to think, the resulting "level" playing field would invariably consist of the top GMs having a slight edge.

Ah, comments from 'the cynic'!

We agree that dumbing-down the computer would not be satisfying.

I believe that you over-generalize my suggestion of allowing humans to move pieces and have databases - since these are my ONLY suggestions and I never threatened 'any measures ... to level the playing field'; just these two, because I believe that they DO constitute an unfair advantage for the machine. The fact is that computers DO have ECO-sized opening books programmed in (and ending tablebases) and when computers 'think' ahead, each 'node' is an error-free representation of a potential position; something humans have to visualize with fallable accuracy.

I wouldn't be in favor of giving humans an 'unfair advantage', either.

Nor do I define a 'level' field as one giving top grandmasters a slight advantage; HOWEVER, it is my belief that those two adjustments would in fact, have that result. I am very curious to see whether I am right and this is why I'd love to see matches of this sort.

Should humans start scoring regular victories in these conditions, then besides proving a point I believe that it would advance Chess theory for all of us in that we'd gain a greater mastery and understanding of the type of strategical play that can defeat brute-force approaches.

Should computers STILL continue to dominate then I would not be advocating further 'adjustments' in the playing conditions; I think then that it would simply go towards proving the 'point' discussed above - that Chess play can be entirely defined by tactical calculations and that we will be closer to seeing the game mathematically 'solved'.

And for a fan like myself I would hope the result would be better Chess games to enjoy. I see it as a 'win - win' scenario.
Dan

Ah, comments from 'the cynic'!

We agree that dumbing-down the computer would not be satisfying.

I believe that you over-generalize my suggestion of allowing humans to move pieces and have databases - since these are my ONLY suggestions and I never threatened 'any measures ... to level the playing field'; just these two, because I believe that they DO constitute an unfair advantage for the machine. The fact is that computers DO have ECO-sized opening books programmed in (and ending tablebases) and when computers 'think' ahead, each 'node' is an error-free representation of a potential position; something humans have to visualize with fallable accuracy.

I wouldn't be in favor of giving humans an 'unfair advantage', either.

Nor do I define a 'level' field as one giving top grandmasters a slight advantage; HOWEVER, it is my belief that those two adjustments would in fact, have that result. I am very curious to see whether I am right and this is why I'd love to see matches of this sort.

Should humans start scoring regular victories in these conditions, then besides proving a point I believe that it would advance Chess theory for all of us in that we'd gain a greater mastery and understanding of the type of strategical play that can defeat brute-force approaches.

Should computers STILL continue to dominate then I would not be advocating further 'adjustments' in the playing conditions; I think then that it would simply go towards proving the 'point' discussed above - that Chess play can be entirely defined by tactical calculations and that we will be closer to seeing the game mathematically 'solved'.

And for a fan like myself I would hope the result would be better Chess games to enjoy. I see it as a 'win - win' scenario.

I love the ignorance on 'chess is solvable'... It's a P-NP hard problem. Please learn some basic computer science and math before making absurd 'chess is solvable' statements.

Maybe chess will lead to some interesting new theorems on being nearly solveable.

regarding Mike P's snide comment -

well, this sort of unnecessary derogatory attack from ignorant individuals is one of the reasons I rarely spend any time posting at blogs. It's a free country and I respect that you can say what you like, but I simply have better and more valuable uses of my time than responding to moronic statements; I will this once, however -

Mike, I've been a Math teacher at the college level for 20 years and before that I was a professional computer programmer; my suggestion to YOU regarding proper behavior is to respond to topics of conversation and specific commentaries rather than attacking individuals; you exhibit the mentality of an 8th grader, and I know.

For those interested in the TOPIC under discussion, you might want to read at this link:

http://en.wikipedia.org/wiki/Solved_board_games

Dan,

As a computer scientist i was just found it amusing that people actually believe chess is a solvable problem. I gave all the information they needed to google on, P=NP... Your credentials as a maths instructor have nothing to do with the correctness of my statement, as evidenced by your lack of disagreement.

As for me having an 8th graders mentality, that's great. Let's review your own statements:

"this sort of unnecessary derogatory attack from ignorant individuals is one of the reasons I rarely spend any time posting at blogs."

"you exhibit the mentality of an 8th grader, and I know."

I think this highlights the truth in the cliche, "If you're not part of the solution, you're part of the problem."

Cynical (and Dan),

Your criticism seems to assume that the only motivation for altering match conditions between humans and computers, is sporting (meaning competitive -- i.e., to see which type of entity is "stronger" in head-to-head competition on a "level playing field").

However, there are other possible reasons to consider the experiment I suggested.

If engines were subjected to human limitations, then programmers who hoped to prevail under such restrictions would have to creating chess-playing engines that "think" more like humans. And that effort would ultimately create objects whose instructional value (for humans) would far exceed any engine whose playing strength is based almost solely on brute force (and the latter is true even for today's most strategically sophisticated programs, such as Hydra).

As you may know, early chess programmers did seek to mimic human chess thought processes. This approach continued into the 1970s, but petered out once the rapid growth of processing speed made the brute-force approach more effective in practical competition (both machine-vs-machine and man-vs-machine).

Now that we've reached the point where computers can convincingly beat the best humans, perhaps it is time for programmers to adopt a new goal: maximizing engines' ability to TEACH humans to improve their knowledge and understanding of the game, rather than maximizing their ability to DEFEAT (other engines and) humans in competition.

Of course, although I call using computers to improve human understanding a "new" goal, for a certain subset of the chess-and-computers crowd, that was really the goal all along. Mig said something to this effect in another thread.

Dan, that's why I think that subjecting computers to human computational limits (thereby pushing engine-makers to work harder at incorporating the "understanding" dimension of human play), would be at least as productive a means to "advance Chess theory for all of us," as would your preferred approach of removing the "blindfolds" imposed on humans by the prohibition against moving pieces as part of their decision process.

Mike, this is a chess blog not a scientific journal. So even us, poor non-experts are allowed to express an opinion! Enough with your garbage-comment.

Wikipedia has ,as usual, some interesting information, i recommend to take a look. When i spoke about the solvability of chess, i hadnt in mind computer science, but game theory. Game theory is a branch of mathematics which aims to formalize, explain and predict the strategic interactions of one or more players who participate in a "game" (very loosely defined). Games are classified with respect their characteristics (for example sequential games, one-shot games, games with complete information etc). And there exist theorems which can prove if in a game which belongs to a certain class of games EXISTS a solution, notwithstanding the solution is not specified at all.

In fact Nash took the Nobel prize exactly because he proved the existence of a solution in non-cooperative games. So what is interesting is to answer:

1. Can be proved the EXISTENCE of a solution of chess?

2. And if it can be proved, how this will affect the playability and appeal of our game?

Michael,

Chess is a finite game. Thus complexity does not enter.

http://www.ics.uci.edu/~eppstein/cgt/hard.html#chess

That there are fewer atoms in the universe than possible chess positions, is a different matter.

Jan,

Whether or not chess is a finite game is irrelevent to it's solvability. What is important is the time it takes to solve such a problem. A great resource for those interested in the problem is:

http://www.claymath.org/Popular_Lectures/Minesweeper/

Christos,

Application of Nash's theories to chess are misplaced. Nash was looking at decisions to maximize benefit. Chess is an objective game, it can be measured in terms of absolutes. In chess the intentions of the players are not in play (in a perfect world. This is what we all try to master, not letting emotions or psychology to get in the way)... In short, chess has more in common with combinatorics than game theory.

Also Jan, your site proves my point. P=NP determines whether or not problems can be solved in polynomial time, according to the paper cited it takes exponential time to solve... ie chess is NP Hard problem.

I dont agree. In game theory, the benefit of the players is very loosely defined. Benefit can be anything you want from earning money to maximizing your pleasure. There is no restriction on what is contained in the playoff/utility functions of the players. So i believe that game theoretic methodologies are completely applicable in chess.
Also the underlying rationale of game theory is that every human action from flirting to playing a board game can be understood in a objective manner.

How does this apply to the impersonal objective view that a computer takes when providing move trees? Chess is a game of accurate moves, and correct asessment of positions. What you are talking about is the condition of humans playing chess. A totally different, and interesting topic, though one that relates in no way to the paradigm of computer analysis of chess positions.

OK, Jon - now I see the 'value' of your match experiment.

You make a good point in that by limiting the search depth it may inspire designers towards new methods of play. I hadn't considered that. It's just that it won't be as satisfying - at least to me; I think, also, to the programmers.

What I mean is that limiting the search depth to prove that computer's 'positional' understanding is faulty would be great if you'd already convinced the programmers that THERE WAS A PROBLEM with the computers 'understanding'. As it stands, with computers winning regularly against all competitors, then one has not demonstrated any need for improvement ie. 'why should I revise my algorithms? it wins just fine! of course, if you turn it's brain off it won't win, but why should I care?'

Having human players defeat the best programs when they are operating at full strenth (my proposal) would be strong motivation for those programmers to 'care'. Your idea has merit too, it's just that I think there'd be less motivation for anybody to try it.

An interesting solution to the problem is to limit the use of databases and opening books to a certain number of moves. Unless you're kasparov, you probably only memorize 8-12 moves of opening theory that you're playing (the average player i'm speaking of), meanwhile some computers have opening books that go out 20 or more moves deep... and if they're hooked up to a database they can conceivably play an entire game without using it's engine to generate the moves played... It would be interesting to see how Fritz/Rybka/Hydra/Whatever would fare if they were limited to 8-10 moves in their opening books, and had 0 access to databases.

As a fellow mathematician – applied industry statistics rather than teaching -- I agree with Dan’s comments here. Recently, I got interested in the applicability of artificial intelligence programs to chess, since many of the issues are analogous to problems that we face in econometrics or time series analysis (see some earlier posts on this issue). Perhaps it is not surprising that all this grew out of a debate among some statisticians as to how to combine methods from operations research (analogous to chess tactics) and econometrics (analogous to strategy) in server networks.

It is clearly premature to say that computers have “solved” chess any more than they have rendered the human capacity for strategizing obsolete. One problem that we are facing in system design is the tendency of many mathematical algorithms to find a dominant short-term signal in the data (again analogous to tactics) but fail to pick up lower-frequency signals (again more analogous to strategy). The solution that we are adopting is to include separate equations for the lower-frequency components (analogous perhaps to having two algorithms for strategy and tactics, running by parallel processing).

This is extremely similar to the problem of designing chess algorithms that play effective positional games. A few nights ago, we set Fritz 8 and some other chess programs to work on different types of positions. The results were impressive, but hardly unexpected.

The computer program loved open tactical positions with obvious lines for attack – it played brilliant sacrificial attacks in positions that arose out of the Najdorf Sicilian. It also did quite well in positions from the Modern Benoni, finding some deep combinations. On the other hand, the computer did not do particularly well on the black side of the Tarrasch French. Once it was out of the established opening lines, and in a passive defensive position, the program’s recommendations were uninspired. This was one of the positions where the human analysts were regularly able to find better moves than the computer. On the black side of a Reti, the computer immediately tried to open the position. It chose lines that led to roughly equal, but double-edged games. Few human players would choose these lines, since they are psychologically difficult: players who favor the Reti often prefer positional chess, and dislike being forced into tactical lines early.

As an example, you might test your programs on the following sequence: 1. Nf3 d5. 2. c4 dxc4 3. Na3 Nc6 4. Nxc4 f6 (intending e5). Fritz 8 played several gambits in this line, for instance 5. d4 e5. This was a style of chess that was common in the nineteenth century.

In sum, the main impact of computer programs will probably be to enhance the ability of human players, by identifying combinational sequences that would be difficult to find with limited time. In this respect, their contribution to the game has been positive.

A while back when I first heard of Aleksandr Sashin's theory of chess,( there is a calculation that one makes to determine in what style one should be thinking, like Tal or Capablanca). I was wondering whether this has been applied to any programs. Namely when the position has Sashin coefficient >1 engine Capablanca plays. If coefficient <1 then engine Tal plays. As Sashin notes, the problem would be at the bifurcation points. Maybe there could also be an engine Spassky which played only when .8<C<1.2? Does something like this already happen(probably not using Sashin's theory, but are there specialized routines that run depending on the character of the position)?

"Now that we've reached the point where computers can convincingly beat the best humans, perhaps it is time for programmers to adopt a new goal: maximizing engines' ability to TEACH humans to improve their knowledge and understanding of the game, rather than maximizing their ability to DEFEAT (other engines and) humans in competition."

Yes, not unreasonable to ask for something back from computers since humans were responsible for their creation and advancement, not to mention the invention of chess.

However, if the main focus is on strength, this seems to me a bit of a low aspiration. What we'd really need is for the computer to teach itself how to play.

I won't try to reinstate our earlier debate about the how computers do and should play, their "chess IQ", learning machines, etc. But if the restrictions Jon suggests would in fact motivate programmers to incorporate human-like efficiency, decision making, and learning into their programs, this could only be beneficial for our understanding of chess. Only when this happens will we discover whether what we perceive as chess intelligence holds any significant advantage over the brute force approach.

I recently saw an estimate of the number of possible chess positions (from someone credible) of between 10^46 and 10^47. This equals the number of particle in a smallish moon, so storing a solution for each possible position of chess, a 32 piece tablebase, should be possible, but not really feasible, with the resources of the solar system. Wikipedia also puts the number of chess positions somewhere around the same number:
http://en.wikipedia.org/wiki/Shannon_number
(I believe these all assume a modification of the 50 move rule to require, rather than just allow, a draw). I'm not sure exactly how quickly the calculations could be done to find out what to store in such a tablebase, but back of the envelope calculations suggest it is probably doable before the sun turns into a red giant.

If chess is as drawish as we think, it is probably much easier to show that each side has a draw than it is to solve every position. I much doubt that the necessary calculations for even this will ever be done, but I'm not quite certain.

I don't understand why we believe chess to be drawish. (It seems good players do: e.g. the recent comment of GM Short that "best" or "logical" play of both players should result in a draw.) But I see no evidence or even hint. All I see is that at a given playing level it is easier to draw than to win. The best exact evidence we have - tablebases - show that some positions previously thought drawish are actually wins. Darn difficult wins. Actually, the highest ELO guys (i.e. computers) don't play more, but less draws. (Yes, the guys who don't make obvious mistakes tend to play more decisive games.) Opening theory has shown that crazy sharp openings thought impossible by the "drawish" tendencies of the past are actually playable. So, if there's any trend, it points not towards balance/drawishness, but imbalance in chess.

Cynical,

Regarding computers using "artificial intelligence" in chess (although you didn't use this phrase, it is what you are getting at): I actually was NOT thinking about AI when I suggested that programmers should resume the long-abandoned approach of trying to model human decision-making processes in chess programs. But I agree completely with what you have said about AI and its potential significance for all the issues that have been hotly debated here regarding human vs computer chess playing skill and methods.

I remember you jumped in at the end of a lengthy debate I was having with someone else on another thread, and you said that if computers could learn on their own (and beat humans by doing so), then there would be no question that they "understood" the game better than the top humans. I posted then that I agree with your proposition. But it remains hypothetical, because any programmer will tell you that no successful chess program thus far has ever incorporated anything remotely describable as AI.

Cynical,

Regarding computers using "artificial intelligence" in chess (although you didn't use this phrase, it is what you are getting at): I actually was NOT thinking about AI when I suggested that programmers should resume the long-abandoned approach of trying to model human decision-making processes in chess programs. But I agree completely with what you have said about AI and its potential significance for all the issues that have been hotly debated here regarding human vs computer chess playing skill and methods.

I remember you jumped in at the end of a lengthy debate I was having with someone else on another thread, and you said that if computers could learn on their own (and beat humans by doing so), then there would be no question that they "understood" the game better than the top humans. I posted then that I agree with your proposition. But it remains hypothetical, because any programmer will tell you that no successful chess program thus far has ever incorporated anything remotely describable as AI.

Michael Parons has stated several times that chess is NP complete, therefore not solvable in practice. This statement has a couple of fundamental problems.

First, complexity hierarchies refer only to problems of variable size, and describe the growth in number of operations required solve a problem in the family as a function of its size. The whole theory says nothing whatsoever about the solvability of a particular instance of the problem family in any specified amount of time. Thus saying chess is NP complete is simply a false statement, because chess is a specific problem of specific size. What you can say is that some particular way of generalizing chess to NxN boards is NP complete. But this statement only guarantees (well, assuming P <> NP) unsolvability in practice for sufficiently large N. The difficulty for some particular N is completely outside of the complexity hierarchy theory.

The second fundamental flaw is that deterministic games have been solved in the sense of which player wins or can force a draw without any substantial compution at all. There is no known proof or argument that it is impossible to prove e.g. black can force a draw by some fairly cheap method. Of course exhibiting a proof that black can draw would not help a computer or human play. However, this is considered one type of 'solving' a game.

The real discussion about the difficulty of brute forcing chess, as someone else points out, is determined by the number of chess positions. The size of 32 piece table base would be a small multiple of this number of bytes. The cost of computing it would also be roughly proportional to this. This is MUCH smaller than computing or storing a complete game tree, but still too large for the forseeable future. However, one can avoid the storage by storing a sparse subset of the complete tablebase such that necessary missing portions can be filled in on the fly in time acceptable during play (in effect, store enough so brute force search can always reach a solved position within e.g. 20 ply, for Hydra). However, this does nothing for the compute time, which remains at least proportional to the number of chess positions. So it will take continuation of Moore's law for a long time, or practical quantum computation, to plausibly solve chess. (A quick calculation indicates that if Moore's law were continue without limit - a doubtful assumption - then in 140 years a 32 piece tablebase could be computed in one day).

vxqtl simply has no idea what he is talking about, the very definition of P=NP problems is that a problem with a finite set of elements is unsolvable in polynomial time. This occurs all the time with pairings of elements for example. Let's say you have 400 people, and they're going to stay in 100 rooms in a hotel, and have a list of 50 couples that dont want to be in a room with eachother... It is very hard to find an algorithm that solves this collection in finite time, and therefore we say p=np. Chess is the same way, this is a matter of combinatorics.

vxqtl simply has no idea what he is talking about, the very definition of P=NP problems is that a problem with a finite set of elements is unsolvable in polynomial time. This occurs all the time with pairings of elements for example. Let's say you have 400 people, and they're going to stay in 100 rooms in a hotel, and have a list of 50 couples that dont want to be in a room with eachother... It is very hard to find an algorithm that solves this collection in finite time, and therefore we say p=np. Chess is the same way, this is a matter of combinatorics.

MP says vxqtl has no idea what he's talking about.

Well, it seems to me like both MP and vxqtl have and "idea" of what they are each talking about, despite the fact that they don't agree. That's ok, isn't it? ;-)

Fascinating discussion, in any case.

phil

MP says vxqtl has no idea what he's talking about.

Well, it seems to me like both MP and vxqtl have and "idea" of what they are each talking about, despite the fact that they don't agree. That's ok, isn't it? ;-)

Fascinating discussion, in any case.

phil

I think Michael Parsons is mistaken again. Two simple points.

First, what 'almost all' computer scientists believe is that P<>NP. Saying P=NP implies that NP complete problems are all actually solvable in polynomial time.

Secondly, the cost of solving an individual problem IS ENTIRELY irrelevant to the classification of a problem family. Suppose some bizarre traveling saleman algrithm solved a 10 city problem in 10**1000 operations. This would tell you nothing about the classification of the traveling salesman problem. However, if it took only 4*(10**1000) to solve a 20 city problem and 16*(10**1000) to solve a 40 city problem, and this pattern was general, then it would imply P=NP, and all NP problems are really 'easy' in a theoretical sense. It would not matter that if this was the 'best' P time algorithm for traveling salesmen problem it would be worse in for any practical problem than a simple exponential time brute force algorithm.

This sort of thing happens for real in algorithm theory. There are are assymptotically 'best' algorithms for matrix multiplication that have such high fixed costs that they don't overtake the simplest n**2 algorithm until you have a million x million matrix.

Note: I agree chess is unlikely to be solved in any sense in the foreseeable future. However, whether some generalization of chess to NxN boards is NP complete (actually it is EXPTIME complete, a much stronger statement), is completely irrelevant to this question.

I like to watch paint dry...

Just to clarify the example in my prior comment, I should contintue it further to be unambiguous. Thus, if a 50 city problem could be solved in 25*(10**100)), and 100 city problem could be solved in 100*(10**1000), leading to a pattern of ((N/10)**2)*(10**1000), then we would have P=NP. Because however insanely large this cost is, it grows only as O(N**2), which is polynomial in N (the problem size).

Why do Mathematicians think they need to discuss this way in a public forum.

Probably a good thing no one is still reading this far down the thread.

Hey - I'm a Mathematician, and I gave up some 16 posts back! However, I'm happy to claim to have been a part of some 'spirited' discussion!

I say - blame it all on MIG. It's his fault. He hasn't started a new thread here in days!!

Agreed... the funny thing with anal-retentive pseudo-intellectuals like vxqtl is that every argument becomes a discussion about who they are. I bet that ultra-abstract nick "vxqtl" is specially designed to hide some obscure racial origin. It is easy to see that the rest of his posts are mere elaborations on that identity defining term, the erratic blabbering of someone who hides within lack of meaning.

OMG the real intellectuals are now let loose.

Mig better start a new discussion or I might join this one. I simply can not think of anything intelligent enought to say to these guys.

I think I will do a Wikipedia search on Einstein and maybe copy some good stuff from there. that should impress someone.

Thank you vxqtl, I gave up on him. And for the record, I'm a mathematician too.

For those of you living in New York city, please start calling the various police precinct's to try and find which one Mig is incarcerated in. Then we can start a fund-raiser for bail and attorney fees.

I like vxqtl. I think he has a point there :-) I'm not matematician, but software engineer.

It looks like we can solve chess if we can solve the storage space problem assosicated with the problem solving. That's an enormous problem of course. It's not possible using technology available today.

Instead of speaking out of their theoretical behinds, it is much simpler to take a look at games comparable to chess and see how progress was made there.

Three games:

Checkers: Many starting positions solved, and entire game expected to be solved within a reasonable time (10-20-30 years, whatever).

Othello: Basically as with checkers.

5-in-a-row: Raw version solved as a win.

In all three instances, the basic method is/was to combine (guided) search from starting position with endgame tablebases.

In particular, 5-in-a-row was solved by applying a search algorithm that limited itself to look only for set-ups threatening in certain ways.

So, in essence, vxqtl is perfectly right.

In general, people should be cautious about thinking they can take some hard-learned theories (that are usually prefaced by dozens of conditionals) and use them to decisively settle problems that are mainly practical by nature.

So, N/NP crowd, please take your pop-statements to slasdot.com ;-)

guys i think that Mig and chessbase stuff are abducted by aliens. Blog and chessbase news are standstill for days...!

I've already told everyone in a previous post that he's in jail, I just don't know what precinct? Besides, I doubt seriously that he would be abducted by aliens when he's one himself. I mean look at him, he's not even trying to fool anyone.

HEEEEEEEEEEEEEEEEEEEELPPPPPPP.

MigInJail

I think computers are reaching the point where it is time to take away their opening books and make them play from scratch.

I would like to see chess programs developed that would customize themselves to the player who is using to a huge degree. You cold enter all your games into it, and you could play many games against it and then it would analyze everything for weaknesses and then coach you on your specific weaknesses. Do you repeat certain types of mistakes without even knowing it? Well, the computer would see this as it gets to 'know' you and let you know about it, and it would even give you drills to help overcome the specific weakness.

Tim Krabbe opines that Rybka is so strong because its programmer is 2300 elo, whereas other programmers are 2000 elo. If we could only deflect Kasparov to a programmer career ...

Can I tell polynomial time by looking at my watch?

My theory is that Mig is trying to force us to use the message boards instead of Daily Dirt threads.

Ha! Fat chance of that happening.

I recently heard that Mig was being incarcerated at the 62nd precinct in Brooklyn. Perhaps Kamsky should show up and represent Mig just in case he actually is in trouble.

Earth to Mig, earth to Mig, come in Mig, over...

Spock, what's the latest on the whereabouts of Mig? I was only gone for four days of R & R in the hologram chamber, fully expecting to catch up on the Dirt this morning! Were you able to verify reports that he is back on Earth? What about the rumor he was abducted by overzealous law enforcement officials?

Scotty, remian on standby and be ready to transport our resident Chess Ninja expert back to the Enterprise.

Uhura, notify Starfleet at once!

Bones, you're an idiot!

Kirk out.

I think Mig is holed up with an incredible woman and he has finally fallen in love. He may never return to chess.

We will have to wait for them to have a lovers spat.

The only thing I have to say about computers is that it spoils almost all the pleasure from online chess. There are two types of online chess players, those who use fritz (or rybka, for the best ones), and those who expect to play fast enough to avoid the first ones, who therefore only play the clock and not chess any more.

That's why you meet a vast majority of cheaters and idiots on playchess and icc.

Another remark : with fritz, the tactical level of humans is much higher today than it was some years before. I mean, a lot of young players are educated with computers, and they all use Fritz as long as it comes to tactical play (when not so long ago, long analyses where required to understand a position).

But on the whole, I have to say that computers, databases, opening science becoming as fat as the average US burger fan, with all this I feel less and less pleasure playing chess : even at my poor 2200 level, I meet players who prepare against me, who study openings ...

Chess is too much about tactics and openings. On the whole, it's a game where huge memory and huge work can bring you to the very best level (who said Kasparov ?). I think that there are a lot of games (go, wargames) where a better balance between strategy and tactics exists, and where 2 games will never follow the same path.

"I think Mig is holed up with an incredible woman and he has finally fallen in love. He may never return to chess."

Replace "woman" with "cat" and I think you're on to something.

BTW I hope Mig is not in jail. That would be an ugly rumor to spread without basis.

I heard a good one recently: "Why is it that people with misdemeanor minds are always looking for felonies?"

Heading: Mig found, see below.

I like computers / I think they're neat /
But one at chess / I'll never beat /

In reply to Ruslan a few posts up--oh incidentally I found Mig; he has been playing in an STC Swiss tournament in Pueblo Colorado. We met in the seventh round, Mig had white, here is the game:

1. e4 e5 2. Nf3 Nc6 3. Bc4 Bc5 4. b4 Bxb4 5. c3 Ba5 6. d4 exd4 7. O-O d3?! 8. Qb3!? Qf6 9. e5 Qg6 10. Re1! Nge7 11. Ba3 b5?! 12. Qxb5 Rb8 13. Qa4 Bb6 14. Nbd2 Bb7 15. Ne4 Qf5? 16. Bxd3 Qh5 17. Nf6+!? gxf6 18. exf6 Rg8 19. Rad1! Qxf3 20. Rxe7+! Nxe7? 21. Qxd7+!! Kxd7 22. Bf5+ Ke8 23. Bd7+ Kf8 24. Bxe7# 1-0

Anyway I had a very fine time reading through some Chernev annotations of Capablanca games/endings the past couple nights... so yesterday at lunch I intentionally traded queens into a "my king in the center" endgame, thinking I would play like Capablanca and therefore win like Capablanca.

First point: No computers were involved in this very entertaining bit of my chess life.

Second point: If the game has lost its pleasures for you then by all means switch to Go or Axis and Allies--there are lifetimes of fun there--however I am profoundly glad that at my low level chess continues to serve tactics and strategy over memorization and computer borgmanship.

Hey I coined a term! Feel free to use it.

Nice to see that Mig has been keeping up with chess history. If he's planning to reprise Marshall's Shower of Gold Coins the next time he picks up the pieces, I'd recommend he consider Fox-Bauer instead -- it's a better use of the same theme (Q-sac on KN6, in conjunction with mate on h-file).

Nice try Rob. You could have picked a lesser known brilliancy, though. Unless it really did happen, and you wanted to let Mig have his day as Anderssen. I believe everyone should be given this privelege at least once.

Axis and Allies was cool in the 80's, along with the Avalon Hill games, Shogun and the long-extinct and hilarious Fortress America.

Still, it's kinda bittersweet to hear that Mig isn't skinhead chattel in a Turkish prison after all.

Yes, Rob. The problem is that your rating is 1650. So it's true that as long as you play at a level where people give up pieces in one or two moves, tactics will be clearly more important in the game than any memory effort you could produce.

At a better level, things are different. I guess you all have already played in tournaments. Haven't you ever noticed that on the first tables you meet most of the times people who talk an obscure language (russian in most cases), have psychotic behavior, strange dresses, smell funny, wear glasses, have overinflated egos and ... who obviously never f--k? They are what we, common people, call the GMs.

And that's when you play against them that the magic quits chess. All of them are living books. They know variations till the 20th move, and have clear ideas about the following plans. They have studied all the endgames (and believe me, you HAVE to be nuts, you NEED to feel very, very lonely, to study rook endgames).

So, of course, after years of practise and monomaniac study, they master the art of chess and can crush you in quite a bunch of variations. And that's precisely why I'm not interested to play any more in tournaments : I'm fed up to learn useless variations. Chess is too much about memory, and too little about strategy. There's too much technique in chess.

Can anybody explain me how two 1600 players could, last year, win a centaur tournament against a centaur team with GMs? The explanation is simple : with a computer on, any human can afford immediately the same technique as a very strong GM. Opening repertoire, endgame perfect skills, and blunder-free games. Therefore, what remains is only strategic talents. And last year tournament just proved that a 1600 team could be strategically more talented than a 2600 GM team. This shows how much strategy is useless at chess.

Yes, Rob. The problem is that your rating is 1650. So it's true that as long as you play at a level where people give up pieces in one or two moves, tactics will be clearly more important in the game than any memory effort you could produce.

At a better level, things are different. I guess you all have already played in tournaments. Haven't you ever noticed that on the first tables you meet most of the times people who talk an obscure language (russian in most cases), have psychotic behavior, strange dresses, smell funny, wear glasses, have overinflated egos and ... who obviously never f--k? They are what we, common people, call the GMs.

And that's when you play against them that the magic quits chess. All of them are living books. They know variations till the 20th move, and have clear ideas about the following plans. They have studied all the endgames (and believe me, you HAVE to be nuts, you NEED to feel very, very lonely, to study rook endgames).

So, of course, after years of practise and monomaniac study, they master the art of chess and can crush you in quite a bunch of variations. And that's precisely why I'm not interested to play any more in tournaments : I'm fed up to learn useless variations. Chess is too much about memory, and too little about strategy. There's too much technique in chess.

Can anybody explain me how two 1600 players could, last year, win a centaur tournament against a centaur team with GMs? The explanation is simple : with a computer on, any human can afford immediately the same technique as a very strong GM. Opening repertoire, endgame perfect skills, and blunder-free games. Therefore, what remains is only strategic talents. And last year tournament just proved that a 1600 team could be strategically more talented than a 2600 GM team. This shows how much strategy is useless at chess.

A mutual friend tells me Mig is hard at work on another project, and should resurface next week. :)

--duif

I can easily explain it Ruslan, the GM started to make 'better moves' than what the computer was suggesting. The 1600's simply were using multiple engines and choosing the most common moves/variations.

Michael Parsons,

I hope you were speaking tongue in cheek. Since you spoke so contemptuously of others in the porton of this thread that you turned into the 'Math Table', I find it ironic that you would so willingly expose yourself to the same sort of ridicule now, by

1) pretending to expertise that anyone reading this can see that you lack (i.e., you glibly second-guessed GMs' judgment in using chess computers -- an activity at the heart of what they do to earn their living), and

2) making a proclamation about how the ZackS team (which won last year's Freestyle) used their engines, without bothering to read any of the published interviews with them.

Those interviews reveal that -- rather than being mere "postmen" as you confidently proclaim on no evidence beyond your personal imaginings -- Zachary and his buddy actively guided the computers' searches and sifted the results, just as GMs do. What's more, they did intensive preparation, which meant making independent judgments about the value of what the engines had come up with in various opening lines.

For instance, on a Dragon bulletin board they claimed to have found a bust to the Soltis Variation, which remains is still a theoretical main line today. Because no one tested them on this by playing a Dragon, they still haven't published their analysis. But since they beat Hydra along with numerous other powerful unassisted engines and engine-plus-GM teams last year, their claim must be considered credible.

Thank you Jon for your continued clarity and lucidity.

Thank you Duif for getting off your duff to let us know Mig's status. "Another project" is much more credible than the jail theory, however juicy it might have been.

Peace to all chess lovers (off the board of course...there is no place for peace on the board!),

T-Rex

Jon Jacobs,

My other posts in this thread have been ravaged by the same people whom I claim are the most ignorant on the subject. Whatever ridicule they dealt, was taken with the grain of salt that other people should take their conjecture on theoretical computer science, of which I am a practioner.

As far as team zachs, you say

"achary and his buddy actively guided the computers' searches and sifted the results"

In other words, took an analysis from fritz and inputted it in junior (or other engines) for verification.

"What's more, they did intensive preparation, which meant making independent judgments about the value of what the engines had come up with in various opening lines."

Yes, one way (the way they did it) was compare the analysis of multiple engines and merely played that move.

At a way to colorfully agree with me. Are you a democrat?

Hmmm, A little late on this post but as Mig seems busy...Let me just preface this by saying that I have never played in a chess tournament. I have played about 400 games on Playchess.com, most of them long games with 15-30 minute time controls. I have never cheated, though several times I have been accused of cheating. I have never had a problem finding like rated people to play an enjoyable game of chess with. Heck, Mig has seen fit to publish one my patzer games in the white belt newsletter. Some of the posts here (not hard to figure out which ones) smack of jealousy. Sure enough I will never be able to beat a GM in a game of chess, no matter how much I memorize or prepare for the game. So what? They are professionals and I am not. I see no reason to denegrate them personally simply because they have more talent and have put much more time and effort into chess than I ever will. To sum this up, if perhaps one day in the future a computer starts a game with 1.e4 and announces mate in 45 moves, it will not in the least bit diminish my enjoyment of this great game.

As the 4th post in this thread (go back and check), I can claim the 'honor' of starting most of this discourse. During that time I was, naturally, savaged by Mike Parsons - who seems to post exclusively to feed his own ego - Mike, this is a CHESS bulletin board, and not one for Mathematics or Computer Science. Your credential of being a 'theoretical computer scientist' impresses no one - including myself.

Perhaps you should learn two things -

1. everyone should know their limitations (quote from Dirty Harry - The Enforcer);

2. this is an OPEN DISCUSSION FORUM in a FREE COUNTRY; people really don't appreciate being called 'idiots' and 'morons' simply because YOU happen to disagree with them. This does not bode well for your ability to make friends or keep them.

I've stated my Math, Computer Science, and Chess background, but NONE of this is really relevant to my original post. That was said simply as a FAN of high level Chess and as a SPECTATOR; I stated simply what I'D LIKE TO SEE:

And that was a match between top programs and top grandmasters where the playing field was equalized by providing the humans opening books and allowing them to move pieces.

The only other thing that occurs to me would be to allow grandmasters to TEAM UP against the best computers and discuss their analysis.

My only motivation is to see GREAT CHESS GAMES played, which I can enjoy! I have no agenda and am not trying to prove any point.

HOWEVER

I've been analyzing Math problems in polynomial time vs. exponential time since high school (30 years ago). I never wanted to get into this subject, since it actually becomes theological - and this is not a religious website. Some individuals are so emotionally overwhelmed by large numbers that they form the opinion that they are unreachable by mankind - only deities. In fact, certain individuals use Mathematics as a 'proof' for the existence of G*d - namely that the laws of probability preclude the likelihood of humanity forming via random chance, without divine intervention. (My reaction is that these individuals have no concept of infinty - in that there existed infinte time before 'creation' where those probabilities approach CERTAINTY).

Numbers do not impress me. Considering the advances in computer processors in the past 20 years - also exponential - it does not really matter to me whether a computer can generate all possible chess positions (as the earlier post: 1. e4 and mate in 45 so wittily suggested).

MY QUESTION is really to explore whether brute force analysis can be proved superior to strategical analysis in fair conditions, and that's what I've been proposing.

The regulars here hardly see me post - even though I've been around for years. I'm not always in the mood for gabbing, and I'm really not that mean an individual - I don't MEAN to offend (even though I know that I said some mean things about one individual - the tone was selected to make a point, not to attack - SORRY). I'm STILL waiting for MIG to chime in with a pronoucement - IT'S HIS WEBSITE!

Until then, it's either type this or pay full attention to 'American Idol'. Which would YOU choose???

FYI, the immediate preceding post with my name on it ("Jon Jacobs, My other posts in this thread...") is from Mike Parsons, not me (I haven't a clue why he signed my name to it).

As for content: anyone able to read or hear anything beyond the sound of their own typing, can see that my preceding comment was NOT in agreement with yours, Mike.

I said that the ZachS players merged their own insights and judgments with the results of computer analysis, and made independent decisions -- something the GM teams also did. That was meant to correct your brief comment aimed at Ruslan, where you stated your belief that the GM teams lost because all the pros were too vain to just sit back and trust the computer(s) like ZachS did. Only, it ISN'T what ZachS did. Go read the interviews; they show no reason to think ZackS's approach was any less human-interventionist than that of the GM teams (and was by definition MORE human-interventionist than those entrants, including Hydra, that consisted of pure engines with no human input).

I'd also call your attention to the fact that Kasparov loudly accused the IBM team of cheating by giving human assistance to Deep Blue in one game of the match he lost in the 1990s. (He was evidently wrong about that.) That is prima facie proof that Kasparov, for one, believes that human+comp can at times generate a better move than unaided comp.

Of course, from the tone you exhibited in your earlier comments to (people who claimed to be) members of your own profession (mathematicians), I would not be at all surprised to see you respond now with something along the lines of, "Kasparov, what does he know?"

Ruslan,

If you are frustrated with your chess knowledge, rating or whatever please keep your frustration to yourself. You don't need to denigrate those who have reached master level in chess because of it. Strong chess players are not alone in their obsessive dedication to the game: professional tennis players, swimmers, guitar players, pianists and others have to put long hours too in order to become good in what they do. Furthermore most strong chess players (IMs and GMs) that I know of (and I know many) in no way meet your stereotyped description of the fellows who inhabit the first tables (a place where apparently you will seldom be found given your attitude).

Yeah, Ruslan, us Dirt readers can be certain that at least one GM f--k's ... if a certain chess widow named "Amber" who once posted here is to be believed (and I must say her catfighting rant struck me as eminently believable, before Mig snipped it).

Darn it Jon,

Why did Mig snip such an interesting post. I would have enjoyed reading her rant. After all I read hundreds of rants from others on this web site and other web sites.

Ruslan,

May I suggest that you take some time off from chess. I have done that several times over the years. Each time I come back with a good attitude and new enjoyment of the game.

What I have done in the past is to get involved into new hobbies. I find a new hobby to be most interesting. Because I know so little about it, it always provides plenty of time for new learning in totally new areas of knowledge. of course it causes my book collection to increase. but it seems good for my mind and for my life.

then after a while, I come back to chess. fresh and ready to learn again.

Since I am a life member of uscf and a life member of my state chess federation, I continue to receive the magazines. but I dont study hard and stop playing actively. I get involved in a different life.

This spring I have returned to one of my fun activities. Growing Hot Pepper Chilies. Of course I am trying some new varieties I have never grown before. and I have found hot pepper chilie forums on the internet so it will bring new fun to an old hobby.

although the people on the gardening forums do not fight with each other like on chess forums. I wonder why they are so nice to each other.

Tommy,

Very interesting observations about gardening vis--a-vis chess. Although I usually have little patience for negative public stereotypes about chess people, in this case it strikes me as entirely natural that people on gardening forums would be much nicer to one another than people on chess forums. Is it a case of "laid-back gardeners" vs. "up-tight" competitive intellectuals? Beats me.

The benefits of taking time off from chess and then coming back to it (and finding renewed vigor and interest upon doing so), have been extolled by no less than GM Maurice Ashley. He put it in a way I found especially eloquent, not to mention appealing. Ashley wrote that, when he sits down at the board for the first time in a few months, he experiences a certain electricity, a special thrill, "as if I was about to flirt with a beautiful stranger."

Tommy, Jon, Gmfan, I really love your comments about my remarks. Do you guys understand what's a joke? Or shall I put some XML tags like this one ???

Well ... let me (try to, but I guess it may prove hard) explain you : for instance when I say "who speak in an obscure language", I am russian myself, and fluent in russian. Ruslan is the most russian first name you may find in Russia. Understand the point?

Of course, now you'll say it's not funny. Which is normal, since it should not take half a day to undesrtand a joke. And, as said a french humorist, you can make jokes on any subject, but not with everybody.

Ruslan,

years ago when my daughter was about 10 or 12 she used to say. "Teasing is not nice. It hurts people."

You must curtail your form of joking. When you are ready to post your type of humor, before posting read over what you have written and ask yourself if everyone will understand what you have said.

Especially look to see if you are joking or teasing. if you are, then make sure everyone will know it is a joke or else do not post it.

Ask yourself. are you trying to be controversial. do you need attention. do you try to say outlandish and rediculous statements so that you will get attention. and then hide behind the words. "well I was only joking"

Ruslan such joking hurts other people. if you love them you will not hurt them.

You might stop and ask yourself. "is everything I have written here an act of love" if some things are not an act of love then do not post them.

and try to remember that when you are making jokes on this website, you are making jokes with everybody. so be careful.

Let me point out one consequence of you joking. from what you are saying I now get the feeeling you are asking eveyone to never believe anything you have ever said on this forum because you might be joking. How are we to know when you are telling the truth and when you are telling us a non-truth that you want to call a joke.

Even your above explanation, I dont know what you claim is a joke ( a falsehood ) and what you claim is your true belief. You can not have it both ways Ruslan. you must show us that either you tell jokes ( falsehoods ) or you tell the truth. It is not up to us to guess what is going on inside your mind. that is impossible.

For me for now, I am going to assume that everything you say is a falsehood. let us see how that goes. If everyone always assumes everything you say is a falsehood. You are not going to like that very much. I assure you.

So you really want to slip in many falsehoods, and you want us all to believe them. you play mind games. then you want to get out of your responsibility for what you said by claiming you were joking. I say that you are abusing us Ruslan. I dont want to be abused by your mind game jokes.

Tommy,

Humor is a funny thing.

My friends and I routinely call each other the foulest names imaginable. It's our way of saying, "no matter what you do or say to me, nothing can damage our friendship." It's our way of showing we love each other. I'm sure you can't imagine it. Just as I can't imagine living in your world.

Mig's overall approach to running his blog is commendable: He censors profanity, slanderous personal attacks, references to his 1836 chess rating, and little else.

I found Ruslan's last post amusing. I find many of your posts pompous, silly, and tedious. But there's room for all of us. Let's show a little tolerance, shall we?

there is a form of "joking" which is really psychological abuse.

Are we going to take up a collection to spring Mig from jail?

My sources tell me he has been put behind bars for habitual, egregious punning, a serious crime to be sure but it pales by comparison to cloying love spam and mind-numbing disquisitions on tree-pruning strategies. (Just take out the damned shears and do it!)

Free Mig Greengard!

Are we going to take up a collection to spring Mig from jail?

My sources tell me he has been put behind bars for habitual, egregious punning, a serious crime to be sure but it pales by comparison to cloying love spam and mind-numbing disquisitions on tree-pruning strategies. (Just take out the damned shears and do it!)

Free Mig Greengard!

Are we going to take up a collection to spring Mig from jail?

My sources tell me he has been put behind bars for habitual, egregious punning, a serious crime to be sure but it pales by comparison to cloying love spam and mind-numbing disquisitions on tree-pruning strategies. (Just take out the damned shears and do it!)

Free Mig Greengard!

There is a risk in trying to be humorous in writing in a international, intercultural forum like this one.

You do not have the body language that would help your interlocutor know inequivocally that you are joking.

Also, there are situations that seem "funny" to some people, but could be considered, immoral, obscene, embarrasing, plain stupid, etc, to others, particularly if those other are from a completely different cultural background.

For instance, Ruslan, How Am I supposed to know that Ruslan is a very common name in Russia...???

Because Ruslan is a common name in Russia.

greg

you said:

"My friends and I routinely call each other the foulest names imaginable. It's our way of saying, "no matter what you do or say to me, nothing can damage our friendship." It's our way of showing we love each other. I'm sure you can't imagine it. Just as I can't imagine living in your world. "

Well I live in an adult world of Love and Dignity and Respect for my brothers. I am sorry to hear you can not imagine living this way. If you truely would like to transform your life, please contact me outside this forum and I will be happy to point out guidance that will help you to transform your life.

You will know when you are ready to change when your deepest inner self says: "There must be a better way."

There is.

Ruslan,

It's all about the presentation. I enjoy satire as much as the next person but your post came across as mean spirited instead of satirical. If you present a joke and your audience does not "get it" then perhaps it is your delivery of the joke that is bad and not the "humorless clods"(my words) that are at fault.

wait!...
shhhhhh!!...

did you guys just hear something?...

Yeah Greg,

Why don't you give Tommy an opportunity to transform your life? I mean heck look what its done for him. Although too many harsh laxatives can take its toll. As far as humor is concerned, either a person gets it or they don't, and I'm not going to baby sit the ones that don't. Also, if a person does get it and still can't accept it...too damn bad. I'm sure not going to concern myself with trying to understand "international or intercultural" mores when posting hear, and the pc crowd can osculate my posterior hindquarters.

First point : I do live in France, and am originating from Russia. Therefore, french is my first language, and russian is my mother tongue. I'm fluent in both languages.

Since english is just some language I did learn at school (like spanish and german), I'm terribly sorry that some native english speakers may not get my international or intercultural or intergalactic jokes (HUMOR).

I'm also terribly annoyed to read posts displaying so much energy, so much hate, written by people who talk about love, about the love I should put in my messages towards them.

Again, some other explanations about humor : when I say "you NEED to feel very, very lonely to study rook endings" (HUMOR), that probably because I've done that myself.

I've studied Keres endgames, Averbakh endgames, Shereshevsky's "strategy in the endgames" (which is a fantastic book), and yes, yes, yes, I did all this when I was single. Right now, being married, I can tell you - very honnestly - that there are chances that I may find something more interesting to do during my nights (HUMOR).

When I say that GMs don't f__k, there is both a part of realism and of humor : humor is generaly based on an exageration of reality. Some GMs have beautiful wifes, of course (why f__k your wife if she's not beautiful ? Hmm, again, TAKE CARE Tommy it's HUMOR !!!). And a lot of them are just maried with chess (like some pianists, writers, scientists, and like any guy or woman who felt too deeply in love with his/her job).

I don't think that a GM reading those jokes would feel hurt. I think that most of them would just laugh. And I don't think that GMs need to be defended by you by the way.

Oh ... a few years ago, the wall was falling. And russian GMS were starting to play tournaments in western europe. It was the 8th round of a tournament near Paris, and I was playing against Neil Mac Donald, close by a russian GM (can't remember his name) who was playing against Philip Morris. He looked at me and said "nu vot, my igraem protiv kapitalisma" (which means "so ... we're playing against capitalism") (HUMOR). So you see, Tommy, GMs have humor too. And you should try to buy some (NOT HUMOR for you, but HUMOR for other people Tommy).

Mendrys,

Yes, humor is not universal. Some people may laugh at some jokes, some other won't. That's because we are all different. That was PRECISELY why I already wrote "you can do jokes on every subject, but not with everybody", and seemingly this sentence is out of reach for your mind.

Basically, when a guy tells a joke, some people may laugh, some other won't. And some other (not the funniest ones) will just say "it's not funny". Which is 100% useless, since a joke is a FREE attempt to make some people laugh (free means that you don't expect anything in return).

If I managed to amuse only a few guys (and reading the thread I've got the feeling that I did so), then your words are totally useless. In french there is a typical word for that, which is "rabat-joie", meaning a person who never laughs and does everything possible to deprive other people from this elementary right...

Actually, I never find anything Ruslan posts to be in the least funny (NOT HUMOR). But Tommy makes me laugh out loud quite often (FUNNY GUY).

Ruslan, keep it up and you'll end up getting a reputation for "humor" comparable to that of a decidedly non-Russian GM -- Nigel Short. (Anyone confused by this reference can Google Short AND slavery.)

I guess, Jon Jacobs, that what you said is well worth one godwin point.

By the way, about Short article (slavery in ancient Egypt), I'm just sorry to say that it shows his ignorance. There were just no slaves in ancient Egypt. The idea was imported later on, but was certainly not part of this culture.

Ruslan,
I don't think you get it. Lighten up and learn how to take some contructive criticism. My only suggestion was that the delivery of your joke could have been better. It did not come across as a joke to most of us here. I am not a nitwit and already understood that not all jokes will be funny to everyone. That is a given. It was not "out of reach of my mind." However, if most people do not get your joke than perhaps you need to polish your delivery instead of calling them "rabat-joie". I have this image of a stand up comedian bombing on the stage. No one is laughing at his jokes and a few are booing his performance. Instead of critiqueing himself and improving his act he instead attacks his audience as humorless nitwits. Rabat-joie indeed! Cheers!

Actually, Mendrys, Ruslan might go over real well as a stand-up comic. I say that because his (former?) obsession with ethnic backgrounds (which, to his credit, he has restrained lately), has always been and remains a staple of live comic performances (in the New York area, at least). In fact live comedy clubs probably are the last remaining places on earth where jokes based on ethnic stereotypes are still considered acceptable for public consumption.

Also, I have to thank Ruslan for raising a term I never heard before (godwin point) -- thereby prompting me to educate myself about it, and I am very glad I did.

To wit:
"Godwin's Law: prov.
[Usenet] “As a Usenet discussion grows longer, the probability of a comparison involving Nazis or Hitler approaches one.” There is a tradition in many groups that, once this occurs, that thread is over, and whoever mentioned the Nazis has automatically lost whatever argument was in progress. Godwin's Law thus practically guarantees the existence of an upper bound on thread length in those groups. However there is also a widely- recognized codicil that any intentional triggering of Godwin's Law in order to invoke its thread-ending effects will be unsuccessful. Godwin himself has discussed the subject. See also Formosa's Law."

I think it's a great rule -- not least because its application would have made me a quick winner of the one debate conducted here and on other blogs that was most important to me. That's because my critics out there, such as jegutman and various knee-jerk America-haters (you know who you are) explicitly called me a Nazi almost from the get-go, over my proposal that TDs figure out ways to watch all players at big-money OTB events (including class sections -- ESPECIALLY class sections) to prevent potential computer- and/or cell-phone-aided cheating.

Sadly, Ruslan seems to have mis-applied the "godwin point" definition in the present argument -- at least according to the definition presented above.

Perhaps being implicitly linked with an apologist for slavery (which I did to Ruslan; I never mentioned Nazis or Hitler) merits making the accused the instant winner of any such debate. But the whole justification for Godwin's Law starts to fall apart once you expand the scope of rhetorical offenses that invoke it. For instance, while it's extremely rare to anyone accused of being an apologist for slavery, people posting to online boards get accused of racism all the time -- sometimes with no justification at all, and sometimes with plenty of justification. Would anyone propose that any person who accuses someone else of racism based on their posted comment, should automatically lose the debate? I didn't think so.

Two godwin points. Second one without any doubt.

By the way, those who are most likely to score Godwin points are the Trolls.

The slavery comments in the Short article:

http://www.thechessdrum.net/newsbriefs/2003/NB_Slavery.html

were a poor attempt at humor on his part. We colonials are still a little sensitive on the subject so my first reaction was OMG, can he really get away with saying this? Then realized that he must be joking. It was almost as funny as jokes about rape and molesting children.

The Godwin law is just the main idea: any hainous ideology works as well. Saying that I'm an esclavagist because I don't think like you or have a different sense of humor is clearly worth a full Godwin point.

Do you have any reason to call me an esclavagist? Where's the point? Why? Because, at some point, the thread becomes too emotional for you. And that's when hate becomes dominant. That's the main idea of the Godwin law, Mr Jon Shrekobs.

By the way, did anybody notice that my post that has been so much debated was originally about strategy at chess, computers + 1600 vs computers + 2600 ? This post main idea was fully part of the thread. And its main idea (which was the tiny importance of strategy in chess ... yes yes yes, it's just my opinion I know, please don't feed the troll) hasn't been much debated.

I think that you can become a 2600+ player without any strategic idea. Mastering the technique, the tactics, the endgames, learning 20 opening moves in any given opening, looking for the last novelties, all this will give you enough to be a good GM.

By the way, looking at top GM games with Rybka (or just Fritz) on shows you that even for 2800+ there is a lot of room for tactical improvement. And any soviet trainer will tell you that before you start talking about strategy (and even openings), first you have to study endgames and tactics (Maxim Blokh books are excellent) to reach an IM level.

(about so many <2000 players reading opening books, I can only ask : why are you spending such a time to get += positions with white and = positions with black when you are fully able to loose one piece in 2 or 3 move combinations?

Ruslan, We can put the joke debate behind us. I admit to being guilty of being off topic. That being said are there english translations of Maxim Blokh's books? My endgame technique is somewhat lacking.

There should be distinction between <2000 players who study openings above all else and those who study the principles of opening strategy. There is nothing wrong with lesser players studying basic opening principles without trying to memorize endless variations to the 20th move. I read a beginning book by Fine (I think it was him, it was several years ago) and feel it helped my understanding of chess immensely.

The great bulk of my "studying" has been replaying master level games and reading a few books on middle game tactics. To this end I found Ludek Pachman's book on the middle game to be especially helpful. I have not studied endgames and as a result I suck at endgames.

I have a few thoughts about why 1600 Centaurs are able to beat 2600 Centaurs. Maybe the 2600's see a position and analysis from the computers and choose a variation that is not on the top of the list because it looks more "strategically sound" and then lose to unforseen tactics while the 1600's who don't have the strategic sense are more trusting of the computer's analysis.

Mendrys,

Your final paragraph repeated the initial comment within this thread that started the conversation down this path. That comment, made by Michael Parsons and addressed to Ruslan, read: "I can easily explain it, Ruslan, the GM started to make 'better moves' than what the computer was suggesting. The 1600s were simply using multiple engines and choosing the most common moves/variations."

In my responses to Parsons' comment, I went on to refute that simplistic explanation. In a nutshell, the 1600-rated human members of the winning Centaur team (in last year's PAL Freestyle championship) are on record as actively intervening to guide their programs and sift and choose among the output, presumably to no less an extent than the GMs did with their respective engines.

The other big problem with such armchair theories is, they are testable, but the people who voice them evidently are too lazy to test them against the available evidence. It proves my point about blogs: it's always easier to type than to think.

To those who enjoy making pronouncements off the top of your head about how GMs or 1600 Centaurs do or don't play, I say this: why not take a look at the actual games (all are publicly available, for free I presume, via PGN download from PAL and/or Chessbase) and see if they tend to support your argument?

In real life, evidence usually is preferred over hot air; but who said a blog has anything to do with real life?

Jon Jacobs,

This is a BLOG not a rigorous technical forum on computer chess. You should expect fairly simplistic answers to some of the questions posed here. I was not trying to suggest that the players on the ZackS team were merely parroting the moves of the computers and were not actively involved in choosing the moves. I was suggesting that the GM's were perhaps overconfident and distrusting of some of the computer's suggestions and guided them to positions that they wrongly thought were favorable. I admit that I have not seen much commentary from the GM's involved but I think this quote is from Steve Crampton:

"Candidate moves were usually chosen based on our own experience in certain types of positions or by the consensus of the computer engine programs. Once we established our possible candidate moves (usually three or less, but sometimes more) we began to investigate the lines extensively."

This and a few other comments lead me to believe that the ZackS team were simply better prepared to play with computer assistance than the GM's who probably did not prepare much and may have had their judgment "clouded" by the superiour knowledge.

Again, as this is a BLOG, I will NOT apologize if my any of my above comments are not rigourous or technical enough for you.

Cheers!

Mendrys, thanks for posting that Crampton quote, which proves my comment (much earlier in this thread) that the ZackS players took a very active role in choosing moves -- even "based on our own experience in certain types of positions" (note this is coming from a guy with a 1600 rating!!) -- as opposed to dutifully following the computer's suggestons, which is what Mike Parsons had said earlier and what you seemed to be saying in your previous comment.

I agree with your conclusion in the followup comment that "the ZackS team were simply better prepared to play with computer assistance than the GM's..."

On the other hand, I've seen no evidence that the GMs had their judgment "clouded by the superior knowledge," as you put it.

In fact, although I myself haven't played through the PGN games, I did play through the single game from last year's event that was annotated by Chessbase. It was ZackS vs VDobrov, I think (Mr. Dobrov, a GM, was also using an engine or engines.)

The thing that struck me about that game was that the notes stated that at a crucial moment, all the fans and commentators using their own engines to follow the game were all expecting and recommending a certain move for ZackS -- but the team wound up playing a different move that turned out to be deeper, and better. The point is, ZackS improved over the move recommended by most or all engines being used by non-participants. And this cannot be explained by the quality of ZackS's hardware or software: they used 3 plain-vanilla desktop PCs running ordinary commercial versions of Fritz, Shredder, and one or two other widely available PC chess programs.

And still more: Come to think of it, the error made by Dobrov (the GM team) in that game -- far from being the sort of mistake that might be made by a GM overriding an engine's tactics-driven choice on the basis of his own presumably superior strategic "knowledge" -- actually smells like the exact opposite. It smells like the sort of mistake that COMPUTERS typically make, while strong humans are more likely to avoid it because their "Sense of Danger" alerts them that a sac might be in the air that a computer won't consider because the ultimate justification for the sac lies too far ahead, so it's beyond the computer's "horizon."

The attack launched by ZackS began with the typical "Greek-gift" sacrifice (Bxh7+) in a French-defense type of structure. The specific position had enough atypical features that the sac was not a slam-dunk; the sac required a much more extended follow-through than the usual quick mating set-up via Ng5+, Qh5, etc. White eventually won by bringing his QN to c7, then e8, then sacking it for a Black pawn on g7, and using his queen and both rooks to ensnare Black's fatally exposed King.

So, it's understandable that a computer playing Black would overlook the sac possibility. Had a human GM been playing Black UNASSISTED against a GM-level human or against a comp, the unassisted GM quite likely WOULD have seriously considered the sac possibility, simply because he would know that Bxh7+ is a thematic peril in such kinds of positions.

So, again, in contrast to other commenters' uninformed speculation that GM teams lost to ZackS because the GMs through misplaced pride overrode the computers' choices, on the basis of this (admittedly limited) example, it is more logical to speculate that the exact opposite occurred: It appears the the GM was too trusting of the software's "judgment", while the 1600-rated humans on the ZackS team dared to choose candidate moves "based on our own experience in certain types of positions", and to second-guess their engines in other ways.

Jon, Mendrys, you're both trolling again.

Explanation : a troll is a discussion where everybody stands on his position, whith one side having typical arguments like "Windows sucks, you can't trust it", and the other side saying that "Linux is pure crap since you can't find everything on Linux".

Since both sides have good reasons to believe that they are both right, the discussion may go on for years (till the point when insults will inevitably start).

And if you ask my opinion, I'll say that despite of the fact that you're saying opposite things, you are both right.

And if anybody says that it's unpossible for me to say that 2 persons saying opposite things can be both right, I'll tell this guy that he's right as well.

Linux is the best OS for a web server. Windows is the best OS for newbies (and there are more newbies than IT technicians). Both systems have their pluses and minuses.

Now Jon Shrekobs and Mendschrek, you can start trolling again, but you should be aware of the fact that apart from Mig (who is forced to do so), nobody reads your posts. And I believe that even you both aren't reading your "opponent" posts till their respective ends...

A few more posts and I think I'll have good chances to achieve your forum education.

I love all those Trolls and Godwin concepts. When people discover them, and see how predictable their behaviors are, they suddenly start acting better.

And Mendrys,

1) You don't need a translation of Blokh's books. They speak the same language as informant (their language is even more basic). They are the best books I've seen till now, contain more than one thousand tactical problems, progressive difficulty and theme by theme study. The informant combination books are also good, but not as good as Blokh's ones (less organized).

2) To improve your endgame, first start with Paul Keres endgames. Ok, it should take more than a year to assimilate this book, but after that, the bible are Averbakh's endgames, and some general ande excellent endgame principles can be found in Shereshevsky's "strategy in the endgame".

On the whole, I'm certain that reading and understanding three of those books (One Blokh's combination, Keres endgames and Shereshevsky's "strategy in the endgame") should bring any 1800 player to a 2100-2200 level.

Of course, it will take a lot of time, but no more than learning the main lines of the scheveninguen for instance... and most importantly, better than using your memory, you'll learn to think and calculate (which is not usefull from move 2 to move 15 as long as a GM hasn't crushed your variation, but will prove usefull all - chess - life long).

Anyway, learning chess should start with endgames. How can you hope to master the game one day if you're not able to deal correctly with a king+pawn vs king+2 pawns position? Once you master all the basic stuff, you can start studying basic pawn structures. Then you add pieces one by one. Then it becomes a middlegame, and you're already at least an strong IM.

And this is how a master thinks (at least, against those who survive the opening and the middlegame phases) : the titled player always think about the pawn structure and kind of endgame he can get out of the middlegame. He can feel or calculate which one has winning chances, and which one will be dead draw.

What's more, a pro player won't trust the theory. Even if it's dead draw, he'll play against you till the very end. Yes, he plays like if he was playing against an idiot. Which proves right most of the times. A russian chess proverb is "let him play". Let you deteriorate your position. To become a master, you also have to learn to focus on the game. First, a few minutes, then an hour, then two, then 4 and then 7 hours.

The ability to stay focused on a chessboard for hours, to eat when it's appropriate, all this is fully part of the IM level and of the pro attitude.

After all this, you'll probably ask me why I'm not a master or GM myself. My answer will be simple : it's a job, and it's not mine. I own my company, have 15-17 employees, and make more cash than most players in the top 10 with much less talent ... and ... I am rated 2200, I am a patzer and will remain a patzer.

... and again, from my description on how one should improve his game from average level till master level, you won't see a word on strategy. That's again what I was saying : you don't need any strategical ideas to become an IM. You just need a strong technique.

I'm almost certain that you don't need any strategy to reach a 2500-2600 level. I don't have the feeling, for instance, that Aronian, who's rated 2750, has a deep strategic understanding. I just have the feeling that he calculates as fast as Fritz.

A few years ago, I played a game against GM Mikhail Ivanov, that I know quite well. His strategical play was totally unimpressive. At the 34th move, I saw a combination, calculated it till almost 9th move, and saw it wasn't working.

Played 30 more moves, lost the game (although I managed to have at some point a better position and a pawn up), came back home, started Fritz (always analyze your games just after you play them), and ... surprise, a few more moves, and after 15-16 moves it's fully won, Fritz gives total advantage for me. I come back the day after, tell him about the combination, and he answers immediately "no guy, because in the final position I play this, this this this ... and the endgame is winning" and finally he shows me on the board what was a huge trap in 22 moves, fully effective, and tells me with a grin that he expected me to see a few more moves than what I did...

This guy, who's never been rated much higher than 2500, is able to calculate better than Fritz. Oh yes, it's clear, his strategical concepts are equal to zero and he's quite clear about that. But he's able to defeat Fritz in a 4 pawn king's indian on both sides of the board !!!

So take this guy, give him Rybka. Take some player with deep understanding of chess (which does not necessarily requires to have a high rating, some chess correspondance players don't have a high rating although their chess understanding is above most GMs...) : he'll just have no chance, because his only strength will be balanced by the opponent's computer, while his ideas will be non-existent.

And again, here's my point : you DON'T NEED a deep chess understanding to become a strong chess player. What you need is proper technique, combativity, energy, tactical skills ... and that's why Magnus Carlsen's rating is above Korchnoi's : I guess anybody will admit that Korchnoi's chess ideas are amongst the deepest you may find. But finding 15 deep ideas in a chess game is generaly erased by a slight tactical mistake.

A chess game is more often lost by the guy who makes the last mistake, than it will be won by the guy who produces 20 bright ideas in a row to outplay another guy playing almost perfectly.

Mendrys

If you consider yourself weaker in the endgame, you may want to consider Essential Chess Endings by GM James Howell. One of the few endgame books that really talks you through each example. My favorite engame book by far. Another good book is Fundamental Chess Endings. If you want an endame book with one variation upon another then probably Dvoretsky's. Personally, I'm too lazy to tackle the latter. At best, it would probably give me a headache.

I'm probably about 1600 strength in the endgame, but seeing the reference to a "headache" endgame book, I can't help adding my recommendation in that category: "Endgame Preparation," by GM Jon Speelman. An extraordinarily difficult book, crammed into just 200 pages or so. (Like we used to say during my short-lived bid to be a math major: if a math textbook is such that you can get through more than 5 pages in an entire day, then it's too easy.)

I know the standard pedagogical advice is (or at least used to be) "study endgames first." Yet, I have to wonder: isn't it true that the majority of amateur vs amateur games -- let alone amateur vs pro games -- are over way before an endgame is reached? That certainly has always been true of my games...which I feel justifies my personal decision to divide my study time about 50-50 between opening and middlegame books. Maybe I'm just too lazy, and am rationalizing?

Jon,

I'm clearly an amateur (USCF low 1700s). Have always loved endgames. And many of my games against other 1500 and 1600 players go to endgames. Probably because I miss so much stuff in the middlegame! :)

(GM Roman Dzindzichashvili once said to me, "You should study a book on the middlegame." I said, "OK, which one?" He said, "What difference does it make? You obviously haven't read any of them." :) )

duif

I am a 1300 in strategy, a 1500 in endgames, an 1800 in openings and a 2300 in tactics. My rating is 2280.

By the way I disagree that you can become a 2600 (FIDE) player without having a clue about how to play real chess. I know several guys that I play around 2400 are not so great strategically and are fuzzy calculators. I have never played anyone 2500 who I felt was very weak though. Perhaps if you have extraordinary talent you can become even a GM without working alot and learning chess properly. Any contact, I have had with 2600+ players and their games suggest that they are on a different level, but that they feel pretty much in awe of the ideas and concepts that guys like Anand play.

DP,

I've been around long enough to know that with a rating of 2280, your a lot stronger in strategy, endgames and openings than you indicate. In my younger days, there was this old timer at the Mechanic's Institute that would tell me time and time again..."You can't fool me, it's impossible." Darn if he wasn't right.

Of course it was slightly exaggerated but it was meant to illustrate that tactics if someone wants to become reasonable (2280), all they have to do is stop giving away their stuff.

Of course if someone wants to give away their stuff, speaking for myself, I always find that quite charitable...especially in chess. :)

Anyway, I knew what you were saying, it just brought that "old timer" to mind and I thought a little levity was appropriate. The not so funny part is that it won't be that long now before I'm the "old timer."

So are we all in agreement, then, that the writings of Lasker, Capablanca, Steinitz, Botvinnik, and 90-95% of all instructional chess books ever published by all authors, are pure hogwash? Because, the explicit message of all those writings is, "Here are principles of strategy, endgames, etc. Learn to use them, and you will prevail -- or at worst, you will improve." In most cases, little is said about tactics.

Of course I'm aware that today's fashion in chess teaching has shifted a bit toward being more flexible and eclectic. Therefore, it's more respectable nowadays to talk about tactics along with principles -- even among people who are not dogmatic byte-worshippers like some who posted on these threads not long ago.

Still, I think tossing out the considered opinions of history's greatest players (along with countless not-so-great players who might nonetheless have been great or very good teachers) about the value of strategy, principles, and knowledge, seems to me a bit of a stretch. All the more so if it's being done largely on the basis of the result of one match between a computer and a strong GM.

Thanks to all for the endgame book suggestions. I think I will look into the endgame book by Howell.

Ruslan,
if the definition of a troll is "a troll is a discussion where everybody stands by his position" than I everyone here is a troll. It is not trollish behaviour to stand by your convictions. A troll is one who while having nothing cogent to add to a debate finds it necessary to post abusive, innane comments simply to inflame the rest of the board. If you find it trollish that I don't think your jokes are funny or that I might slightly disagree with Jon Jacobs about the conduct of GM's in a freestyle chess match then so be it. I think few here would consider any of my posts to be inflammatory.

Its my belief that Kramnik will not win and will be lucky to escape with a draw While chess will not be. "solved" for many, many years (i.e 1st move, white to mate in 186 hehe) , It's clear that the amount of ply/calculations per sec is just starting to exceed top grandmaster ability. Even with human's superior positional play and understanding, the brute force and depth of the machine is starting to get the edge. I would be highly surprised if ANY grandmaster out there could beat Rybka in a match

For ending books, I'd recommend van Perlo's endgame tactics. It's just won the British Chess Federation's book of the year award. A good way to learn about endings for those of us who love tactics.

I like DPs analysis of his strength: I'm 1300 in strategy, 1400 in midlegames, 1600 in openings, 1600 in positional endings, 1800 in tactical endings, 1800 in tactics in general and 2200 in clock management. My rating is 2100.

I'd back myself against someone who is 2200 in all of the above, except 1400 in clock management. I have 80% scores against a number of players who understand far more about chess than I ever will, but who can't manage their time. e.g. they spend 5 minutes on a simple queen recapture thinking about their next move, get to a position which is +/= or +/_ (or even +- according to the computer) but don't have the time (or nerves) to convert to the win, then blunder and lose.

al, you should give lessons. If clock management can be taught, then your students would go on to make a better living than most GMs...by hustling in parks.

I don't mean that as a put-down. I've spent many a frustrating and expensive day in the company of park hustlers. Setting aside the question of honesty which is an entirely separate cause of frustration (in New York at least, at least 50-75% of hustlers are crooks who run off to avoid paying if they get a few games behind; I've had some even threaten me with violence if I told their financial backers they owed me money), I often got the maddening feeling that I was consistently better at the game of chess than this guy who'd just scored 17/20 against me, but he was much better at managing the clock.

In about half of those cases I'm only talking about table tennis -- pure and simple hand speed. Being over 50 now, my dexterity is none too great, and I was never a pianist or a painter or a sushi-chef or hibachi-chef (i.e. champion hand mover) to begin with. For instance, I know many people who will always beat me if it gets down to my 15 seconds vs. their 10, or even 8, seconds. Of course in such a case it matters not at all what pieces are left on the board, as long as each side has at least one pawnn when the time scramble ends.

But the other half aren't beating me by means of athletic skills entirely apart from chess. No, they seemed to have a sustained strategy, more of an instinct really, about allocating their time. And this strategy intermingled with their move decisions, and even seemed imbued with psychological insight about which moves would be most likely to lure ME into time pressure. When a hustler who's a half-decent chess player gets into a hopeless position on the board -- rather than giving up hope, he'll unerringly find a way to make his opponent's job just a little tougher, stave off mate a lot longer...and then the clock will be on the hustler's side, even if it wasn't before his position went south.

The hustler's desperate tactics in such situations strike me as one part silicon (straining for the miracle defense that a computer would find), one part Simon Webb (who advised in lost positions to always play quickly, set traps, and/or create murky complications any way you can, even -- or especially -- if you have to sac material) ... but most of all, clever exploitation of the clock, maximizing the possibility that the opponent will get into time pressure and go astray despite (or because of?) having a clearly won position.

So, if this clock management skill could be taught, I think there'd be a sizable market for it.

Jon,

I think Dan Heisman has already found that niche at Chesscafe.com. I know it is directed at Novices, but so much applies to players up to and including my level- it is common sense.

I try to use my time well for key moments in the game - take 15-20 minutes to work out a plan and then play the next few moves at a move every minute or two and use my opponent's time for many other decisions. A lot of it is playing good enough moves (IMHO too many people spend a disproportionate amount of time between two moves a computer may consider to be + or - 0.1 apart).

A lot of perfectionists collapse mentally when they miss a shot (even if it doesn't make them worse - I tend to say what a silly bugger, never mind, I'm in this now, so how do I get myself out of this.

The old maxim is that without increment, if you can have 15 mins and your opponent 5 in an unbalanced position, you will do OK.

In another sense the book which has the biggest effect on my play was Aaagaard's Excelling at Chess (a more professional expansion of Simon Webb's great ideas)

I think there is a definite psychological aspect to hanging on in worse positions- the secret is not to go for broke for a cheap trick, but to wriggle without worsening your position too much.

As for hustling, I am a cr@p 5 minute player. Once you're down to that amount of time, all bets are off.

A few random thoughts- off for the evening

Cheers,
Al