Jen Shahade: Fritz Control

I'll exercise privilege to make first comment on this 600th DD entry before it goes online.

People kibitizing online with computer help has long been one of my own pet peeves, especially if I'm doing commentary on an event. True, I'll usually have an engine on myself in those cases so I don't suggest a line that turns out to be a selfmate in six, but having dozens, if not hundreds, of players, most of them humbly rated, shouting computer evals and recommendations all game long is both obnoxious and worthless.

Worthless, that is, if you actually want to learn something from watching the game. Instead of using your own brain you use Fritz. Using an analysis engine to check your work after annotating a game is one thing, having it there all the time is a powerful crutch that cripples. Sure, it's interesting to know what Fritz thinks, but just a main line and eval every once in a while is more than enough.

And for spectating, it's a disaster. Most amateur players - and this isn't their fault - can't tell the difference between human chess and computer chess. They shout out suggestions that are comically silicon, impractical combinations in winning positions, etc, that a GM would never play. Worst of all, they seem to think that +1 or -1 means something in human chess. To hear the world's top players insulted because they made a move that lowered Fritz's eval by half a pawn is sickening. I remember a guy at Playchess who would dare to add question marks to GM games during the broadcast based on small computer eval changes. Unless the change is +3 or more, it's best to ignore it unless both players are computers too.

It simply dehumanizes the game. Having a strong player give commentary is useful even if some of the suggested lines aren't computer-best because we arrive to the conclusion in a logical, practical way. Every once in a while that conclusion might be refuted, but that doesn't mean the logic behind it isn't helpful or that it won't serve you well nine times out of ten. So long live Fritz, but turn it off when you watch online!

Mrs. Shahade tolerates weak players' annoying behaviour. Weak players tolerate Mrs. Shahade less-than-stellar analysis.

I sure find such attacks on players like Ms. Shahade to be more than annoying. Don't people realize that there are many different levels of players?? In fact the vast majority of players out there are of low enough level to find Ms. Shahade's analysis to be quite helpful. I am tired of people acting like on the top GMs are worthy of analysing anything. I would suggest that it is actually more helpful to lower level players for someone only slightly above their level to analyse for them than it is for a top GM to do so.

Obviously there are arguments for both sides here. A lot of people who watch live chess, myself included, watch it primarily for the sport. We may analyze the games later to improve our own play, but when we're viewing them live, our main concern is what the result is going to be. And computers are simply more likely to accurately deduce this than the brain is.

I don't own Fritz, but I appreciate when someone who does posts analysis. The trick is not to get carried away. If you don't want any help looking over the game, just don't read the kibitzing.

And we tolerate Bacan being an insulting jerk in every thread. It's the cycle of life.

Back to the topic, it's about defining what "help" means. Fritz certainly isn't any better at predicting the result of a game between humans than a human master or GM. Quite the opposite in most cases. Comps have no clue about chances and initiative, only the eval. That's just an opinion of what would happen with best play by both sides.

Ms. Shahade is a star player and her point is a fine one: a game between titled players is the product of highly-skilled professionals doing their human best under all kinds of pressure and it should not be derided by spectators whose only claim to express an opinion is the fact that they are armed with computer programs.
Also, her live commentary is always interesting and L Bacan's remark sounds like poor-taste advocacy.

it's an annoying occurrence on playchess for the simple reason that anyone can open an engine window of the current game. but, i like tactfully spaced computer evaluations while watching icc relays i'm just too lazy to follow the game in a separate comp window.

i too watch for the sport of it, and while watching super gm games i require an engine evaluation to dependably assess positions that are often over my head. gm coverage on icc can sometimes supplement this, but playchess is terrible about having live gm analysis in chat. it's usually fredric making some colorful remark, and nigel showing up for 3.25 seconds to tell a dirty joke. unless they're trying to do video or audio broadcasts, in which case the chat is usually no more interesting than 'is it working?' 'do you hear anything yet', to which everyone replies 'no' and the loop repeats. then, if we're lucky, the full broadcast becomes available later for the fine price of 9,000,000 ducats.

i only listen to chess.fm if i'm lonely and desperately need to hear voices to stave off insanity.

the lot of my opinion, though, is that it's an easy thing to ignore and commentary on it isn't worth the column space.

Personally, if Jen gets that upset that easily, then she might not want to ever go on the ICC again. There are morons everywhere and you aren't going to get away from them, especially there.

Let's face it, most players are very weak, so for we stronger players to criticize them for using Fritz just isn't fair. Many, if not most, no matter how hard they try will never become masters either, so they use engines to help. Even if you think it is overdone, that is not really for you to decide.

There are those who go overboard for sure, and I too, HATE the idiots who append question marks for small fluctuations in evaluation, but that is just because they don't know any better. When that happens, I suggest the liberal use of the censor command...I do. Moron added. :)

That's not the point. It's not about Jen, or me, or anyone being annoyed, at least not primarily. It's advice about why this negative for the chess development of those who do it and those who are bombarded by a constant stream of evals and computer moves instead of thoughtful, human, evaluations of the game and its progress.

It's more of a philosophical discussion. Several people have said they like to watch Fritz because they want to know who is winning, which only rarely makes any sense, unfortunately. But of course everyone can do as they like. It is further computerizing a human game, and in a way that is wholly unnecessary. It becomes an addiction and a substitute for thinking about the game in a useful way. In chess, "why" teaches while "what" does not. In post-game analysis you can use the engine to answer some of your "why" questions, but during a game it's rarely anything like that.

Jen:

What is your solution? There was a discussion about foreign languages and we thought it might be good to have separate observation boards for this. It does not bother me but I know a lot of people are annoyed by it. I am wondering if it would be worth it to have a board for the engines to give analysis. Just a thought.

There's no solution, just encouragement. As long as open chat is allowed, and it should be, there will be people shouting computer evals and suggestions.

Mig: Looking over my original post, I believe it sounded too much like I didn't agree with you about position eval. Certainly there are humans who are better than Fritz at telling a crowd of kibitzers why a given position is advantageous. There will always be players who take Fritz eval for more than what it's worth. But at the same time there are occasions where electronic aid comes in handy: for example when a player takes that extra-long think after the time control, the computers are usually the first to find the winning line (depending on which/how many titled players you have kibitzing). This is what I meant by "deduce the result."

I understand that there are some people who want to search for those winning lines themselves, but I'm interested most in finding out whether there's a win or not, without having to wait twenty or thirty minutes for the next move. Is Fritz overused in some situations? Absolutely. But it can be a help to those who know how and when to use it.

Well, this time i agree with Jen on the use of engines...it's much more interesting to have a GM commenting on the angle that humans are looking at the games from.
i watch games on playchess and i'm happy when some even quite strong players give their evaluations and self-figured out lines...even though i'm a patzer, i must say even GM analysis usually doesn't go over my head but i can grasp it when it's well explained.
Of course, some commentators are better than others, but what really i find annoying is when a commentator starts telling with the self-assurance of his belief in himself on what for example Kasparov is going to do next, usually they get it wrong. Weaker analysis is understandable on the level of super-tournaments, but they should not pretend they know for certain what's going on in the game when they don't.
It's better when the commentator honestly just says what he thinks of the position and what possibilities he sees in it, without pretension of knowing what super-GMs see. Ironically enough, it's usually these commentators who get it right anyway (i like Larry Christiansen and Danny Kopec a lot on chess.fm, for example, and Yasser is great on playchess, also Nigel usually knows what he's talking about if he bothers to talk about the game :) )...it's the weaker ones who are way too certain of themselves.
In the absence of commentators i've put on Fritz a few times to verify my own analysis, but i really prefer a human angle.

I agree wholeheartedly with Jen and Mig. And also, ICC kibitzers are, in general, all-around jerks and not just for shouting comp evals. I'd take 3 comments from Talion (Kamsky) over 99% of the crap ICC kibitzers come out with, everyday. I log on as a guest just to watch the broadcasts, and now I "set kib 2" (only see kibitzes from titled players) as fast as I can.

Once again, the blogmeister resorts to personal insults when confronted with the truth. He calls me a jerk just because I said what everyone knows: Mrs. Shahade is a 2300 player, who is given the opprtunity to do chess commentary ONLY because she is a woman. My poit is that with that especial privilege, she should be tolerant enough to accept that some weaker players will use Fritz hoping to anticipate the result.

Also, contrary to what the blogmeister claims, Fritz is quite accurate, but he will probably deny what everybody knows: the worst Fritz analysis is still 300 points better than Mrs Shahade's bset. So, why not invoke it?

Instead of feeling hurt by the facts, Mr. Blogmeister should avocate intellectual honesty.

The truth is you are unable to post without insulting someone and it happens too often to be an accident. But I'm going to be reasonable.

The entire point of Jen's posts and my comments is that a computer suggestion, no matter how accurate, teaches you very little. Many people can't see the difference, but it exists in a big way. Fritz can suggest move after move of accurate but completely illogical moves and without any way to explain them other than one being .24 better than the other. A move suggested by Jen or any other master has a lifetime of experience and logic behind it. This logic can be communicated and understood.

Even computer chess programmers will tell you that a program like Fritz has the knowledge of maybe a 1600-player at best. Combined with database openings and endgames plus 3000-level tactics, they play like 2700 players most of the time. That's great, but that doesn't mean their middlegame recommendations are useful beyond tactics.

And speaking as someone who has done live commentary on my own and alongside of GMs and others for years, any 2300 has a better chance of predicting a move in a GM game than any computer engine. Computers don't play chess, as GM van der Weil said long ago, they only play a game that looks a lot like it. Playing good moves for humans is about logic and process and computers can't help too much with that.

Amateur players put way too much emphasis on "yeah, but Fritz can kick your butt" and ratings because they are objective factors easy to understand. But they are not useful. The comment "300 points better" is spectacularly ignorant when comparing a human to a computer. It's more like the 2300 being a thousand points better at commentary, especially if he (or she) has an engine on the side to check the tactics, where they are obviously great.

Chess is a human game and when it's being played between humans, human commentary is going to be far more useful and accurate.

Saying that Jen does commentary only because she is a woman is revolting. There are plenty of male players of her level or weaker writing analysis and commentary all over the place and being paid to do so. (Myself included.) The ability to teach and explain coherently and put yourself in the shoes of your audience are all far more important than a few hundred Elo points.

Not only can't anyone rated under 1800 tell the difference between commentary from a 2600 and a 2300, but he may actually be better served by the comments from someone closer to the same universe. Few 2600 players have any idea what it's like to be a 1500 since they passed that mark over a weekend when they were nine years old. But someone who works with players at that level all the time and can explain a GM game to them in a fun way is worth her weight in gold. I've hired and fired players from 1600 to 2700 and the ability to communicate has always trumped rating. Big numbers and big names can be valuable for selling product, but you need people who can really reach your audience in the long run.

Or maybe you're one of those people who believe that all great teachers must be wild successes at their chosen topic. I've had a lot of good ones and I don't recall any of my best English teachers being famous novelists or business teachers being billionaires.

As a tiny minority, women chessplayers draw a lot of interest, we know this. This is neither a bad thing nor their fault. If you choose to focus only on their gender that's your problem.

I think the essential point of Ms. Shahade's post is very important, particularly since most of the players kibitzing computer analysis on ICC seem to be weaker players. One reason weaker players ARE weaker players is because they tend to take the easy way out in terms of chess improvement, and aren't willing to put in the hard work necessary to improve. They use gimmicky, easy to learn openings; they read chess books without trying to figure things out for themselves before looking at the analysis; they buy books with titles such as "...Made Simple" or "The Easy Guide...," hoping to discover a magic formula.

The problem is, chess is neither simple or easy. There is no magic formula for chess improvement and the only way to really improve is to put in some hard work. Watching a game with Fritz may be fun, because it can give you a false feeling of superiority over really good players. But you don't learn to think for yourself; all you learn is that computers are very, very strong these days and that humans, even the best humans, make mistakes. Humans don't play chess like computers (not even Hikaru Nakamura can do that!), and having a computer spoon-feed you answers does relatively little for your chess. It's different listening to a strong human analyze during a game: there at least, you can pick up useful hints about how a strong human player thinks about the game and try to emulate her/him.

If you have no ambition to improve at chess, and you just like watching chess with Fritz running, there is nothing wrong with that. But I agree with Ms. Shahade and Mig that such people should keep their computer-generated comments to themselves. Probably 95% of the people watching games on ICC have access to strong chess computers anyway, so what is the need to shout out computer-generated analysis? You ruin the experience for the very people who are trying to get the most out of it; the people who watch without computers but who are interested in the commentary of their fellow humans.

I am fairly certain that virtually any player below, say, 2200 elo will learn more about chess from listening to an hour of Ms. Shahade's 2300 elo commentary than they would from a week of watching the numbers flicker on Mr. Fritz.

Just my opinion.

Interesting topic! I must admit to following online games with Fritz and paying more attention to the evaluation changes after each move than to trying to understand the game. Now, I realise that from a learning point of view this is counterproductive, but I have NEVER watched games online to learn (although perhaps I should) - I study my own games/read chess books for that.

I watch online games because I'm interested in who is going to WIN. Computer foibles notwithstanding, I have usually found Fritz to be a good predictor of game results.

I agree that kibitzers quoting Fritz evals at every opportunity are irritating, and I'll add my voice to the chorus asking them to stop, but I think we all know that's not going to happen.

Human commentary is much better (and infinitely more entertaining) than watching computer analysis, but good commentary isn't easy to come by and I'm sure it's a lot harder than most people seem to think. Commenting on any game is difficult, but I imagine chess is harder than most.

Nevertheless, we should aim high. Four things I want out of live chess commentary:
1. An interested and entertaining host who is not a master, but is knowledgeable enough to ask intelligent questions.
2. Two chess pundits who ARE masters to talk to the listeners and each other about the plans for each player in the position and concrete lines that might occur.
3. A professionally produced broadcast without the technical problems that seem so prevalent.
4. Above all enthusiasm! The ability to communicate effectively with the audience is much more important than whether they are a 2300 or 2600.

If all live chess commentary had these characteristics then I think I really would be able to learn something from following online games.

Several errors with blogmeister's "logic":

>>
Fritz can suggest move after move of accurate but completely illogical moves and without any way to explain them other than one being .24 better than the other.
>>>

The moves selected by Fritz are NOT illogical. How can the be when they actually work? How "illogical" can they be if they are accurate???? That computers see far more possibilities than humans doesn't make their moves "illogical". That's easy to see.

>>>>
A move suggested by Jen or any other master has a lifetime of experience and logic behind it. This logic can be communicated and understood.
>>>>

I'm pretty sure Mrs. Shahade has a lot to offer, but it pales in comparison with Fritz when it comes to chess analysis and the quality of that analysis. Mrs. Shahade COMMENTARY is far more entertaining that Fritz's dry analysis. Fritz's commentary is weak, just like Mrs. Shahade's analysis is weak (or strong, depending on the audience's own chess strenght).

>>>>
Saying that Jen does commentary only because she is a woman is revolting. There are plenty of male players of her level or weaker writing analysis and commentary all over the place and being paid to do so. (Myself included.)>>>>
There's nothing revolting about it. It's a fact. Can you produce a list of 2300 players getting Mrs. Shahade's exposure? I personally have nothing whatsoever against her getting these opportunities to further her chess career, but the truth can't be ignored. As simple as that.

>>>
Even computer chess programmers will tell you that a program like Fritz has the knowledge of maybe a 1600-player at best. Combined with database openings and endgames plus 3000-level tactics, they play like 2700 players most of the time. That's great, but that doesn't mean their middlegame recommendations are useful beyond tactics.
>>>

That's a childish argument borne out of pure denial. Its a dumb as someone saying: "Well, if you take away Kasparov's opening preparation, incredible memory - he claimed to have over 10,000 games in his "storage bin"! - and endgame knowledge, he is just a 2300 player".

The facts is simple: run-of-the-mill chess software consistently plays at the 2500+/GM level. Fritz is not some "enhanced" 1600 player, and trying to imply that Frit's strenght is all "about tactics" only exposes your ignorance: chess IS about tactics. Short-term, long-term, simple, complex,. But tactics, nevertheless.

>>>
Or maybe you're one of those people who believe that all great teachers must be wild successes at their chosen topic. >>>

This where your dishonesty really shines. I never said anything remotely close to that (my post is still there, and I hope you are honest enough to not erase or edit it). I only said that Mrs. Shahade should not be so tough on her audience, because she's not devoid of weaknesses herself. I repeat: she has a less-than-perfect audience; her audience has a less-than-perfect commentator/analyst.

If this is perceived as a personal attack on you or Mrs. Shahade, then that's your problem. I have only stated facts as know by anyone who will read this posts.

Another important point lost in the blogmeister's diatribe: most people connect to ICC to watch games and follow commentary in a very informal way; they switch between games, kibitz, joke, ask questions, play other people while waiting for the GM's to move, etc. Let them enjoy Fritz's analysis. There's no harm in it, other than perhaps exposing Mrs. Shahade's occassional 1600 move.

Not to point too fine a point on it, but there is this thing called 'censor' on the ICC. By simply censoring anyone who is 'shouting' computer analysis, there you go. Not to mention you can also set it to filter out all comments by the unwashed and untitled:)

Peace...

Well, LBacon, you have made an argument that is ridiculously easy to blow a hole through. In stating that there are not many players of 2300 strength who get to analyze games and present them to weaker players, you conveniently overlook perhaps the most booming elemebt of US chess today, that being scholastic competition. In the scholastic ranks, weaker players are constantly learning from players rated 2000-2500 or so, and they do just fine. How many players have benefitted from working with Pandolfini or other such people? Hell, I am only 2100, and I have a student who recently broke this same rating level after I worked with him since he began playing chess, followed by another who recently won 2nd prize in World Open U1800 section. If the people who are analyzing are GMs, then of course Jen's analysis may pale in comparison to theirs (although she has outcalculated GMs over-the-board to win games, so this is not even a given), but if most players are so weak that they don't understand anything about what is going on (some suggest moves that just clearly lose pieces in two moves), then how is their strength different from that of our scholastic players? Jen has helped to coach national championship teams, and they have not suffered for her not being a GM. MOST players do not need a GM analyzing for them. If you speak to the glaring statistics, which indicate that there is not even 1% of chess players in the US who reach Jen's level, then her analysis surely must serve some value to weaker players, who would likely get swept in a match against her or at the very least lose in some absurdly lopsided score. She is teaching them to think as a 2350 player (her usual range) rather than as a Class C or what-have-you, and you criticise this? Obviously, you are not familiar with how people actually learn chess; maybe you should refer to Jen, who knows, rather than providing your weak analysis of the learning process. Your attitude and elitism is both unwarranted and repulsive, and I don't see you on a FIDE list anywhere, so sit back and learn a thing or two. You can't call Jen weak unless you can beat her over the board in a match, and my money is on the lady, so STFU about her already.

Hotep,

Maliq

Can you dispense with the massive quoting? My post is right there. Breaking things down out of context is just a distraction.

You prove my point nicely. To you, any move that is accurate is logical. That's the way Fritz plays. Humans can't do that, they have to start from somewhere to get somewhere. So you are a slave to Fritz, like a deaf and blind person having something pushed into his hand. There are unlimited examples of good moves that aren't logical at all, and computers specialize in them for the simple reason they don't employ (much) logic to find their moves, they employ counting.

Chess is tactics, eh? Profound. And since computers are good at tactics, they must be perfect teachers for humans! Great! I hope you eventually get strong enough to realize what a load of garbage that is. Human chess is NOT tactics. It's a complicated blend of planning and patterns that are reinforced by tactics, which are called upon when opportunity strikes. It's like saying your calculator is a great mathematics professor because it can do sums faster than any human. Your calling me ignorant about computer chess is amusing and desperate.

Computers reduce everything to tactics because they can and because they must. They don't have enough knowledge to do anything else and they don't really need it because they are now so fast at the tactics. The 1600 knowledge level comes from a director at ChessBase, by the way, but I'm sure he's dumb too, right? Only you really understand chess and computers, just like every other under-1600 who worships the machines because they beat humans. This is precisely the mentality we are trying to illustrate as lunacy, and damaging to development.

The computer is an oracle, not a teacher. Knowing something works is useless other than from a totally passive spectator position unless you are strong enough to figure it out. Knowing WHY it works is how you improve. Understanding WHY white is better is what is important, not reading +0.93 and saying "WHITE IS WINNING!!!" because the oracle told you so. You're a caveman looking at the tree burning where the lightning hit it and saying that will teach you how to make fire.

My example about teachers was just that, an example. I didn't accuse you of saying it so drop the drama. You imply rating is everything, I pointed out that it isn't. You seem to believe that because a computer can play at a 2600 level that it understands something about chess. You are wrong, tragically wrong, and I can only hope some people here will actually follow Jen's advice and learn something instead of huddling around the oracle so they can get a false sense of empowerment. If you can't understand why computers play at a 2600 level you aren't qualified for this discussion.

Just to stop at one example, I have more "exposure" than Jennifer without being nearly as accomplished a player. It's for the same reason. What we're most accomplished at isn't playing, certainly not on the international GM scale. Being a good commentator does not require being a 2600 player.

I am a Corporate Training Specialist. I teach many facets of our business, including computer training classes.

I mention my background to bring up an important point. Computers alone are not great teachers. The #1 thing I hear from all my participants is their preference for human feedback.

Though they can learn computers on line they do not like it. When we offered computer training recently, they jumped at the chance and the feedback has been great! Note: the company has offered on line training for years, but nobody uses it.

Fritz is a computer program. It can crunch numbers, and throw numbers at us, but it cannot replace human interpretation. Besides, if Fritz is that good then why are students of the game still looking for, and getting human teachers to help their game when they all ready own Fritz?

I agree with Mig and Jen, if you want to better understand the game, then sit back and listen to the commentary.

It's sad when an interesting topic for discussion gets lost under a flurry of impolite remarks and self justification.

L Bacan - your first post on this thread made it clear that you think that Jen Shahade's commentary isn't very good. OK, everyone is entitled to an opinion, but reading the post back it sounds like an insulting remark rather than a considered comment.

I'd like to hear comments about how others watch games online and whether they use Fritz and/or listen to commentary and which they find most useful (or perhaps both together is best!).

Are we all sometimes guilty of relying on computer evals too much? Can we learn more from comments from human players, even if those comments can be flawed? Or can computers really comment on chess games better than humans? I'm with the humans...

Dear Blogmeister,

your ranting doesn't validate your absurd position. No matter how bad it makes you feel, computers consistently perform around 2600, which is 300+ above the best Mrs Shahade can do and 500+ above you. That's a fact.

And I'm still waiting for examples of 2300 male players getting the same exposure Mrs Shahade gets.

I find preety abusive that you or Mrs. Shahade would have the gall to tell players they shouldn't use computers and computer analysis while watching chess on ICC (or any other server). As others have pointed out, those lowly players (a few of them the same lowly players finding value in your paid "subscription") can be easily censored.

No need to become so agitated, blogmeister.

BTW, my point of view has nothing to do with my own strenght as a chess player. I am a master, but my opinion would be equally valid coming from a 2600 or a 1600.

i think the point is that computers don't tell anything about chess, they merely crunch numbers. They make no plans and have no strategy, and thus a human player who wants to improve on chess learns nothing from the computer because humans make plans and have strategy. Human players don't simply crunch numbers.
Computers can show a winning line, perhaps, not even that always, but never will they tell you anything about the game as it is played between humans and what is there for a human to understand.

blogmeister and others claim that:

1. Chess is NOT about tactics.
2. Computers only do tactics.
2. Computers exhibit the chess understanding of a 1600.

Question pops to mind:

How, then, can we explain that computers play at the 2600+ level?
How can we explain the recent Hydra-Adams result?
How can we explain Kasparov loss to a computer?

The answer is quite evident: if computers have shown anything, it is that tactics are the game's foundation. Strategy is being exposed by computers as long-term tactics (a view held by some from the days before computers).

A more balanced, honest appraisal of computers and chess is necessary. I personally don't find computers a pleasant or meaningful opponent. I don't play against computers, because I don't see the point. To me (your mileage might vary) the game is just a source of pleasure and winning or losing against a computer is not pleasnat or interesting. But that doesn't mean I'm going to ignore the facts: computers play chess at a level that only a handful of humans can match. That's not magic. That's simple and pure good old-fashioned chess strenght. We may not like it, but the results are there to remind us that it is very real.

L. Bacan wrote:
"I asked for a list of 2300 MALE players who get the same exposure Mss Shahde gets."

Dennis Monokroussos 2352 is one example
http://www.uschess.org/msa/MbrDtlMain.php?11502229

Peace...

Here are a few players who are Jen's strength or below and are yet pretty well known in chess circles, to say the least: Dennis Monokroussos, Sunil Weeramantry, and Bruce Pandolfini. I am sure that there are more if we care to dig deeper, and this is excluding young players who technically would satisfy the inquiry because they are, indeed, male. Jen is a strong player who has had some of the best results for a non-GM in recent US Championships, making two IM norms. Don't speak as though she is some Class C player analyzing the games, and PLEASE do not presume to speak as though you can't learn a thing or two from her, because you obviously can. When you beat Ehlvest, let me know. Until then, keep it shut regarding her chess strength and how it helps weaker players. Obviously, players below GM strength have been benefitting weaker players for a long time.

Hotep,

Maliq

Hotep,

Maliq

You seem to be missing part of the point, L. Bacan, maybe unintentionally. The point is not whether Fritz-generated analysis is stronger than Ms. Shahade-generated analysis; of course it is. The same could be said in comparing Fritz-generated analysis to that of other human 2300-2600 elo players, regardless of gender.

The point is that the 2600-strength analysis generated by Fritz is simply not that helpful to you if you want to improve your own play. Like Mig says, Fritz is an oracle, not a teacher. It will tell you the truth, but it won't teach you how to find the truth yourself. So using Fritz for critical analysis (say, opening preparation of a very sharp position) is certainly justified, because for such purposes it is more important to know the truth that to be able to derive it. But for most players (particularly players below 2300 elo), chess knowledge is not as important as chess understanding. And computers cannot help you with understanding, only with knowledge.

No human will ever be able to emulate Fritz, we do not have the computing power. What we can do is try to emulate stronger human players, who have found approaches and strategies that make them stronger players. That is what you get when you listen to a stronger human analyze: insight into their approach to thinking about chess.

Like I said before, there is nothing wrong with running Fritz while you are watching online games as long as you are aware that the process is not helping your chess much. If you enjoy that, fine, more power to you. What IS wrong is imposing your analysis on other observers, many of whom are interested in hearing human ideas ("How does that FM approach this position? Why?") but not interested in having the computer-derived "truth" handed to them on a plate. In any event, what is the point of sharing computer-analysis? For the few observers that don't have computers of their own, on ICC they can go to a channel where strong computers are providing analysis anyway. So the only point of kibitzing computer-analysis seems to be ego-driven: a kind of "I know the truth as stated by Fritz; Shirov doesn't." But so what? It takes no skill to read evals off Fritz; even a novice player can do that.

I don't want to knock Fritz or people who use computers to analyze; I do it myself when I am looking at specific positions with lots of tactics. But as many people have said here and elsewhere, the best time to use Fritz is after you have tried to figure things out on your own. Not before. So please don't force your "Fritz-first" ideology on us "Fritz-lasters" by kibitzing.

Dan Heisman has done game commentary on chessfm. He is also a superb teacher who is only a 2200 or so.

Elliot Winslow is also a frequent commentator and I don't think he even plays chess any longer. I am sure Jennifer Shahade is easily in his class as a player if not stronger at this point.

I do think Jennifer's gender was a consideration in putting her on. However, while I am normally against political correctness in all its forms, in this instance I am not. I am not even sure I would call it that. If J.S. were an A player or an expert and still got on, I would have a problem with it. She is a master and certainly worthy of doing game commentary. Are Winslow, Kopec, Paschall that much stronger then her that they should be on and she shouldn't? I think not.

And if her being on the air can in any (small) way attract women to chess, great. Chess needs all the help it can get.

Btw, the makers/marketers of Fritz are lucky I am not master of all I survey because if I were there would be men with billy clubs knocking on doors asking (politely, at first) people to turn over their copies of Fritz. These engines are a blight on the chess landscape. Their primary legacy will be the perception (if not the actuality) that chess is solved or played out. Good riddance to them.

Bacan, I'm not buying your premiss that the strongest players are the best teachers. What's the basis for this premiss?

This is a phenomena I've noticed on rgcp as well as here. Someone who is losing an argument on the merits pulls the "rating card" and starts attacking the other side's skill at chess rather than their position in the argument. It's the kind of ad hominem argument that is frowned on in civil discourse, but I suppose civil discourse is getting rarer these days.

You seem to be trying to make a point, L Bacan, but you convince noone when you state your conclusions without giving persuasive reasons, and when you don't respond to other people's arguments. And starting with the insults just makes your position seem even weaker.

You describe yourself as a master, but although you don't give your rating I suspect without knowing that Ms. Shahade is a stronger player than you or you would have tried the same argument with her you did with Mig. I suspect without knowing that I am also a stronger player than you. So why should anyone listen to you when Ms. Shahade and I disagree with you? Perhaps you can understand that such arguments have little merit when you are the target of them.

Just my opinion.

As far as planning in a human way, of course computers are not the same as we are. That said, you often see humans play gambits to increase mobility or get a lead in development as the whole plan, with the idea of out-calculating you. I would say that even in the absense of a gambit, that a computer is, in effect, always playing using this same plan. They just try to increase their mobility every move, and frankly, I have tried to incorporate that into my thinking as well as the old human way of thinking in certain cases. I don't think that it is fair or correct to say that a computer plans like a 1600 if you consider the increasing mobility a valid plan in and of itself. If it really planned that poorly, regardless of the tactics, it would lose. It is just a matter of what you consider a plan.

I also liked the comment about watching the relay games to see who wins. No one ever said that anyone watching was looking for an education. They are there for fun from what I have seen, so let the comp users kib what they want. Let Zek handle the muzzles (people who go overboard with comp analysis are usually asked to take it to the comp analysis channel), and you use your censor command if you don't like what they have to say.

Who made the top players the chess police?

I'm deviating from the crux of Jen's post significantly. But as the precedent for this has already been set in this thread, I'll feel free.

From the Popularization angle. Jen mentioned the Drama in a game. Until some acumen (techniques?) is developed to reveal this in a game, the game is doomed to a small niche market in my opinion....even among chess players...let alone the wider public.

But what to do? Posted it previously. I was watching the Nakamura/ Beliavsky game in minnesota. Rooting for Nakamura... knowing the importance of the game for him... Nak's got white....the stage is already set at move one. So I settle in and enjoy the game with my C level understandings. Gets to a point where each have a Q, R, K and Nakamura's a pawn down. I'm looking at the board and am marvelling how these guys hold these sharp positions together. I guess we're 2-3 hours into it and i'm excited...looks like Nakamura's gonna get the pawn back and open up the black king. Bing, he takes the g pawn....gets up and walks off the stage... which isn't particularly unusual for him ... but it seemed just a tad different than usual... a little more urgent.... i remember thinking it meant he just made strong, game tipping move....Beliavsky pushes a pawn. Nakamura comes back on the stage, looks at the move and resigns. I sat there stupidly wondering what the hell just happened. The climax in the Drama was lost on me.

If a person or an algorithm could have prepared me for the implications of Nakamura's move Before he made it, the experience would have been a great deal sweeter. How to transfer the Drama that you know the two players are experiencing, to us, is the key. Frankly, don't have an opinion whether it's a chip, a GM, a Master or a 1400 that achieves this.

I've tried interpreting the algorithm outputs in terms of it showing the sharpness of a position. This is often a feature of a game that i can't see.... particularly early in the game. I'll look at the early output of the 4 lines the algorithm is evaluating and look for wild swings in the statistic. But when you look at the lines, they're often not even 'natural' looking to the E player. Few minutes goes by and the statistics converge as the ridiculous liines are pruned. The apparent inability of the algorithm to immediately start working on moves that a human player would find logical minimizes it's value to me.

That said, strong player human analysis runs into this problem for a C player. I think it was pearl that suggested that different levels of player/ commentor be used in presentations. Certainly you'd need some pretty thick-skinned, gregarious 1500, 1800, 2100 analysts... and strong analysts not too full of their own powers. Suppose you'd need to have a flow from low to high in the commentary... i'm guessing that it would be quite an art to get the most of such a setup... but worth the trouble.... from both the popularization and learning angle.

Parting jab...not sure people yelling out fritz evaluations is less or more annoying than making the analysis of a chess game sound like a boxing match (eg. the 3D kasparov match)... neither is effective for me.

I should know better than to feed the trolls, but I think that L Bacan's point about "Chess is merely Tactics" can be answered directly.

Mig's observation was a good one to start with: Fritz is to Chess as a scientific calculator is to Math. There is NO understanding in either of them, just the logic their programmers put in there. Most of this logic is tactical. Someone with enough programming experience can easily verify this; examining any top-level open-source chess program (such as Crafty) will suffice.

Using this to conclude that "Chess is fundamentally tactics" (as Bacan does) is a classic mistake. It's like concluding that all of Art, Literature, and Math are fundamentally physics and biology, because that's what we (as humans) all made of. Bacan is misusing Occam's razor and has begun cutting away necessary pieces of the puzzle.

With that in mind, his questions are fairly easy to answer and should sound familiar. Humans have other qualities that figure in: aspirations, fear, the ability to withstand psychological pressure, among many others. You can't conclude that "Chess is just tactics" because Kasparov or Adams or even every player over 2300 fails to win a computer. There are too many other factors in play. Ignoring them makes your argument sound like a classic case of "false dilemma" -- a logical mistake, pure and simple.

His most recent argument that Jen or Mig don't have the authority to answer what's best for weaker players is hardly worth the effort to answer. I have personally gained more from reading their analysis and commentary than most books authored by players with ratings higher than theirs, not to mention the algebraic notation spit out by a chess engine. As I've already mentioned, their strategic and tactical understanding is only one factor in winning rated games. It's also only one factor in their value as teachers -- their ability to communicate being equally as valuable. I may find some tactical ideas when looking at Fritz while replaying my games. On the other hand, Jen's explanations of potential strategic plans during Chess FM live game coverage has shown me insights that I have used to win over-the-board games recently. In terms of my own personal chess improvement, knowing the reason behind "the solution" will stick in my mind to the next tournament, and perhaps spark other ideas as well.

Gil Epis:
Myself, I don't like to hear any commentry during the game. Tell me, if you get to know the truth much ahead, then there wont be any excitement. Its same like reading a Sherlock Holmes story. The truth must be hidden till the end to really enjoy the game.

L Bacan:
You made some good points about Computer Chess. Some people may learn well with the computer help, but others can feel closer following master's commentry than with the computer.

It is also conceivable that one could follow an in-depth computer analysis of a line and fail to derive any useful conclusion from it because of the essential difference between simple brute force (that is what the chess computer algorithm used to be called) calculation, and the different schemes, pointers, general principles, hunches, and plain guessing that humans use. That is the value of opinion makers, they may not always be right but they contribute importantly to educating the public, and in this chess is not that different from politics. From a different perspective, as great a figure as Einstein remarked something along the lines of "I cannot calculate but I can think" because he had to hire mathematicians as helpers.

Peace...

Damn it, LBacan, can you address the MAIN point of Jen's post?! She DIRECTLY spoke to why it is inappropriate for weaker players to throw out Fritz analysis on ICC when others are observing the game and trying to figure out things on their own, and you idiotically continue harping on "But see here, they have the right to use Fritz instead of trying to figure it out for themselves!" That was NEVER the primary point of the post, and you damn well know it, which is why you keep avoiding it. She said that it is wrong for the weaker players, having used Fritz to learn the "reality" of the game, to then throw it out, in essence spoiling the mystery. If you were to pay to go to a movie theater, intent on seeing some suspenseful movie, and then the person seated in front of you was to remark "I have seen this movie before on bootleg CD, and the demon is ultimately a hoax," would you be pleased with this. Yes, they have the right to know what they know without having to guess at that precise moment, but is it appropriate for them to take away the chance of others to study the situation for themselves? Address THIS point. This is the main point of the post, and to ignore it is to concede that you are wasting everybody's time.

Hotep,

Maliq

Peace...

I should also point out that Fritz has its moments of allowing people to make themselves zombies despite the obvious being right before their eyes. I recall a recent game with Nakamura against an IM in which Hikaru reached an ending which was not winnable, even up a piece for a pawn or some such. People kept insisting that the ending HAD to be a win, because computers still had him winning the position. One look at the board could tell even the weakest player that stalemate tricks, etc. stopped this from being winnable, but yet it was insisted upon that there MUST be some win there. People submit to the idea that computers MUST see more than we do, and this is not always true. I once, for fun, entered an opening trap into Fritz and played it through. Fritz insisted that one side was winning for the longest while, because it couldn't find the finishing line. Meanwhile, the game was over by force; the solution was just beyond its horizon. Slaves to computers will run headlong into drawn endgames thinking that they are winning and losing endgames thinking that they are drawing. Thus, it is sometimes the case that people who are shouting that Fritz has found a win for one side are actually sharing bogus solutions and thus even detracting from the evaluation process by leading people, under guise of authority, down the clearly wrong path.

Hotep,

Maliq

I'm a Mac user..and haven't figured out how to get chess.fm to work (help somebody??). I treat internet games as a quiet learning excercise (I'm a puny 1936 USCF). I try to select a plan or a move and then pay close attention to possibilities in the position seen by the players that I missed. I never kibitz on ICC. Boring, eh? I bet there are a lot of players like me following the games..we're just a silent segment of the audience. I'm not saying there's anything "wrong" with people who are less into learning and more into socializing and following the competition like one would a football game. I don't think anyone has suggested that even though a lot of the chat seems moronic to some of us. To me, if you want to learn there's no better way than to buckle down and think out the position BEFORE checking Fritz.

This is the first blog thread I have followed on Migs site. It is always good to see reasonable people set the record straight. Briefly, regarding Mr. Facts' comment. Sports and media coverage is no different than chess coverage. Personalities of varying talent and experience share insights that make it more enjoyable. How many of ESPN's casters are their sporting equivalent to a GM? Perhaps the two best sportscaster personalities (Bob Costas and Al Michaels) are not "GM's" in any sport. Maybe John Madden is Fritz on the side? Ms. Shahade
has earned every opportunity she has. Chess needs people like Mig, both Shahade's, even a Waitzkin to increase its visibility. If chess only had people like our master level Mr. Facts chess would be even a smaller blip on the American radar screen.

One thing I am certain of----Mr. Facts likely lacks the people and communication skills to promote chess in any venue.

Personally I (in general) enjoy the IM's (or nearly IM's) I've heard more than the GM's I've heard. Probably because the lower rated commentators have a talent for making it interesting and that is in part why they got the job.

What's this attack on Mig? Give me a break. Get off his site and enjoy sites of people who do have enough status.

Finally--Are you jealous of Ms. Shahade's overall chess accomplishments? Are you compensating for your own insecurity and frustration.

This is my post. I hope that I am not too argumentative. I am tire of chess players shooting chess in the foot by their narrowness and jealousies.

Alaskanbishop

Maliq, who the hell is Jen or anyone else to tell them what to do when they aren't even asking? People pay money to be on ICC to do what they want within the ICC "atmosphere" policy. You have the option to censor anyone you don't want to listen to.

If people want commentary like hers, they can tune into ChessFM, which personally makes me want to barf due to Tony Rook's comments since he has no business commenting about chess as he knows less than nothing about chess.

This is going a bit off topic but I too can barely take Tony "Rook." He deserves our thanks for starting chess fm but it is true he knows next to nothing about chess. And from what I can tell, really has no desire to learn about the game's history/culture. In fact, I don't think he even bothers to play the game any more.

For any event that I run, I will do everything possible to disallow computer evaluations from being used during kibitzes/whispers. There is nothing that makes a game less interesting to me (and to many others) than having such evaluations read, and we made it our habit to ban people from watching the games if they didn't follow these rules.

For the most part I feel that it made the NY Masters event one of the most entertaining to watch, as you didn't always feel like the result was preordained due to a quick Fritz analysis.

Also this is much more relevant for computer chess, because if Fritz finds a brilliant combination, you can bet that Hydra is going to find it as well, whereas if there are humans involved there is some doubt as to whether or not they will even see the combination.

I wish ICC would disallow computer kibitzes for ALL matches. There is little so annoying as the Chinese Water-Torture trickle of junk like the following scrolling by:

------------
somenitwit kibitzes: what do comps think?
------------

As someone stated earlier with uncanny accuracy, there are no shortage of morons and other defectives on ICC. No way to stop 'em all. However, liquidating the stream of computer evaluations would be a hopeful start.

One idea that Zek had used with some success was to confine computer evaluations to ch 227. That way, if someone WANTED to hear all that noise, they could simply type +227 and leave the other kibitzers blissfully unaware.

As it is, ICC remains unresponsive to the majority on this issue. So the rest of us have to endure unsolicited computer spam along with the normal stream of server gibberish (shouts, etc).

L Bacan: "And I'm still waiting for examples of 2300 male players getting the same exposure Mrs Shahade gets."

Eric Schiller - he apparently has publishing contracts, and I've seen his books in Barnes And Noble and Books-a-Million. You would think that he is the voice of American chess. I won't comment on the quality of the books, as that would be impolite, but that is the example you asked for.

btw, I think Larry Christiansen and the Winslow guy are funny, and entertaining to listen to.

i agree that Tony Rook is horrible. He's simply ignorant, vulgar ("let's go to the squares" is extremely vulgar), and even rude enough to disrupt the commentators. Sometimes i get close to being allergic to his voice and yapping.

Let's not forget another male 2300 player who has certainly received his share of exposure: John Watson. It'll be a long time before a computer can duplicate HIS abilities.

Peace...

Funny, anonymous coward, I actually DID avoid including Schiller because I did not want to comment on the quality of his books. Actually, when perusing it work, it might seem that some of his books appear to be assembled using ICC kibs! Nevertheless, he is quite well-known and is yet another example of a 2300 player who does not suffer for being male.

Hotep,

Maliq

A very interesting topic indeed. Saludos and one question: if engines like Fritz are so packed with knowledge and insight, why do the people playing correspondence games have such different results, even if both of them are playing with progs aid? (Don't tell me they don't if they surpass a certain rating level)

Correspondence champions know *when* to tell an engine *what* to calculate and to *educate* the comps memory by guiding the engines through the engines' own suggestions until they realize for their own that this line decreases in quality behind their horizon.

By the way: Hydra lost two correspondence games. How could it be? The guy who won is a correspondence grandmaster, but his engines can't possibly match Hydras 200 kilo knotts per seconds.

------

Topic: I have been one of this kibitzers posting moves from comp evals by myself from time to time, and I stopped to do it only three of four months ago. I had the impression it had something to do with feeling on the right side of something. And in the end I think I might have done it to boost my ego as well. Like rasing the hand in the classroom.

But now I tend to feel bothered if someone else does it. But they have to go through this as I did probably. Now I like kibitzing with others and discuss in aficionado style: "Ah, the 15.c3 line in the Breyer. Wonder what Leko would have played against Short this year in Wijk if Nigel had went for 15...d5!? and such things. Great that Graf is going for that."

---

Finally I sympathize with Gil Epis:

"From the Popularization angle. Jen mentioned the Drama in a game. Until some acumen (techniques?) is developed to reveal this in a game, the game is doomed to a small niche market in my opinion....even among chess players...let alone the wider public."

There are plenty of comments I'd like to say something about, but the only one I will address is the reference to John Watson as a male 2300 who has (nevertheless) made a reputation for himself. Watson, a long-time IM, might be a 2300 now, but he has been largely inactive as a player for well over a decade and suffered a stroke about five years ago. In his playing prime back in the 1980s, already a well-respected chess author, his FIDE was over 2400 and his USCF rating over 2500.

Dennis: thanks for educating me a bit on Watson's health background. I'm a huge fan of his work and meant no disrespect in case you may have sensed that. I consider him to be a "Nimzowitsch" (as a teacher) for our time. He's living proof that it's silly to judge chess authors and commentators (and "blogmeisters" too) by their latest published rating.

I wonder if Fritz can analyze football games?

FM Alex Dunne is similar in strength to Jennifer Shahade, and is a monthly columnist in Chess Life.

This is a fascinating thread. (Imagine Spock saying "Fascinating.")

As an adult beginner (USCF 1200) and holder of an ancient computer science degree, I am quite interested in the many comments that assert that computers and humans play chess differently. Of course they do, but how does one tell? I get the impression it's just from human experience.

I'd love to see a book that annotates famous games with both human and computer commentary. And I'd need the human (programmer?) to explain the difference between the evaluations. The goal would be for me to glimpse how the human thinks vs. how the computer "thinks".

The explanations would probably look similar because there is a big difference between the processes involved in search and discovery but the context of demonstration would be the same because a human would have to complete the "explanation" by the computer to put it in readable form, unless the comp is also programmed to offer readable explanations.
Just as in mathematics textbooks the demonstrations seem difficult to understand because they focus on logical order, which is often counterintuitive and very different from the real narrative of discovery, so it happens during a human chess game. Old studies of grandmasters thought processes spoke about a confusion of partial images of pieces and positions, incomplete explorations of possible tactical schemes that are left unfinished sometimes only to come back to them later on when possible combinations come to mind and so on. Also, each player has his/her own personal arsenal of preferred tactical positions and combinations so that the narrative of discovery by humans could be disorderly specially when confronted with unexpected moves. On the other hand, the computer has no narrative but a plain regurgitation of calculi some of which were discarded because of bad evaluations by the algorithm.

Jdmarino wrote:
>>>
I am quite interested in the many comments that assert that computers and humans play chess differently. Of course they do, but how does one tell?
>>

I think (and this is, admittedly, empirical knowledge - or just simple observation) that computers, like chess players, go through a repetitive cycle of evaluating the current position (includes the assessment of its own and its opponent's strenghts and weaknesses), calculating the tactical possibilities, evaluating the candidate resulting positions and selecting what seems like the best move. The further a player/computer can go (deeper analysis for the chess player / more "ply" for the computer) in accurately evaluating the position, the greater the playing strenght. This is, roughly, the way GM Kotov explained the move selection process in his classic series of books "Think/Play like a Grandmaster".

I know my view is not very popular in chess circles, but it's about time we admit that computers can play the game at the highest level. Any reasonable person will agree that there has to be a good reason for computer's overall playing strenght, because it is not a coincidence and there's not a randomness to chess (the rating formula proves this - chess performance can be accurately predicted and tracked over time).

I think the healthy attitude towards computers is one of neutral personal preference: if a player likes to compete against computers, he should do that. If a player doesn't find playing computers interesting (my case), then he shouldn't.

There's no point in using ridiculous arguments ("computers understand the game at a 1600 level", "computers cheat because they have an opening book", "computers don't get tired", "computers can't play the ending") to pretend that computers are not what they really are: 2600+ players. They are here to stay; they consistently beat the overwhelming majority of chess PROFESSIONALS; the have reached a level of chess skill superior to that of most humans.

Is that bad? Not at all. Computers have changed the game and will continue to do so. We can only adapt, accept that they are part of the chess scene (yes, including online "computer-assisted kibitzing") and continue enjoying our beautiful game.

Elliott Winslow is an International Master - and his not playing chess much any more has more to do with having to earn a living.

Computers are playing by correspondence rules.

That is, when they want to look at a position 10 moves ahead, they simply do exactly the same thing they would do when "moving the pieces." They can never, for example, "forget" that the Bishop is no longer on c4, or skip a move.

The same is true of human correspondence players.

This is the difference between "recall" and "recognition" in terms of cognition.

For humans, chess tests the process of visualization, the ability to project the possible. That's why "touch move" makes a difference. Humans do this by learning the language of chess, by knowing what logically SHOULD occur. This is why grandmasters are much, much better (based on the van de Groot studies) at recreating a legal chess position after it is flashed on a screen for a few seconds, but no better than amateurs at recreating an illegal position (such as one with three Kings).

Computers do not have to use this skill at all. They have the equivalent of an analysis board at their elbow, where they can just move the pieces around and evaluate the game as a present fact, not a future possibility. There is no "touch move" restriction on the computer.

To truly test a computer's ability to evaluate a position vis a vis humans, we'd have to test them in a situation where both sides were playing by the same rules.

Bacan, you are a troll talking to yourself in a mirror. How many more paragraphs will you spend asserting, in the face of zero opposition, that computers are very strong players? You are lucky that the dread Mig didn't sh**can your posts. Face it: Fritz readouts are decent teaching tools, in the comfort of your own study. As participants in group discussions, online and in real time, they are basically worthless.

The way computers play chess, is that they use brute force, via the minimax algorithm (specifically the more efficient algorithm of alpha-beta), to evaluate the position one gets if the computer plays the best possible move, the opponent replies with the best possible move, the computer replies to that, etc. This evaluation of a position usually consists of the material balance plus positional modifiers, expressed by the number of pawns the position is tilted towards one of the players, hence the +0.54 valuations etc we are used to seeing computer programs present to us.

Of course, to find that best move, you have to make sure that the final position after ALL such possible combinations of moves are worse than the one you are going to play. Computer programs do this by building a tree-structure of all possible, with each new level/ply representing a new move. Since there's on average 35 possible moves in a given position, that tree branches exponentially into ~1000 positions at 2 ply, ~35000 at 3 ply, a million at 4 ply, a billion at 6 ply etc. We see that this would quickly limit the number of moves a program could look ahead, since even today's computers can only perform a few billion operations a second. The solution to this is various pruning techniques and ordering algorithms, where obviously "bad" branches are cut off. Computers commonly search the tree-levels iteratively, adding one more ply to its search after it has evaluated (or pruned) the positions at the current ply-level.

When discussing the playing style, strength and weaknesses of computers, what should be noticed is the way computers evaluate a position, and the tree-search. In the evaluation you might put in something like giving a small negative score the further away your knights are from the four central squares (knights on the rim are dim). Similarly you can reward pawns that come closer to the eigth rank, punish pawn islands etc. But the thing is, that most of these evaluations are measures of static aspects of the game, since what could happen two moves further ahead, is resolved when the tree search gets two ply further (if there's enough time to calculate) and not in the evaluation itself. Secondly, you need to have a way of expressing the idea in an efficient manner for the computer to evaluate. Thirdly, there's a limit to how many positional modifiers you can add into the evaluation. Each such modifier adds to the time required to evaluate the position, and therefore means you can evaluate fewer positions in a set amount of time. And finally, it's difficult to implement a way of disregarding these modifiers when they are not relevant, since general rules are just that, general.

Now, the tree search. The obvious problem with it, is that there's a horizon effect. Suppose the evaluation shows there's a rook to be won after 11 ply. Great, unless the opponent takes your queen on ply 12, because the program never got that far before it had to move something because of running short of time. This is partly resolved with things like quiescence search, that makes sure the program searches extra plies as long as the current position has the possibility of captures or checks (non-quiescent). But that still leaves a few long-term problems, like stumbling into positions that are materially good, but positionally bad, because the opponent has either a winning attack coming up in one more move, or the position is completely blocked and drawn.

In addition there's also opening databases, and there's endgame databases that can be applied to play perfectly when there's few pieces on the board (6 or less atm).

Concerning how GMs think, I can only contribute with what I've read, since I'm not a good chess player :-) Top players either a) Spot one of the thousands of "attack patterns" (or something that resembles that "pattern") that they've learned to recognize, and try find a way to set up that attack (or defend against it). The bishop sacrifice and mate on h7 would be a good example for a patzer like me, although probably not for a GM ;-) b) have a plan they wish to achieve, spotting almost immediately 2-6 obvious candidate moves to do that. And from there trying to work out a way to do it, by visualizing and using recognized patterns where possible, calculate where necessary. c) in an opening or endgame position where they can play by memory or a specific strategy.

The difference is that computers simply brute force, working through its search tree, keeping the best current move and substituting it if it finds something better, while GMs usually start with one or more "hooks", something they recognize as good, and try to find a way to achieve that. The weakness of a computer, is that there are things that neither the evaluation nor the tree search will pick up, like the mentioned draw positions, locked down positions without tactics (where its relatively few positional evaluators will let it down compared to a GM) or endgame positions that can't be solved with endgame databases and the program doesn't know the specific strategy for that sort of endgame. The weakness of a GM, is that he (or she) plays a lot better with these patterns, than in a position without them, where pure calculation is necessary. A computer "sees" all short-term tactics, where a GM could miss them if they are not part of a pattern.

To sum up: A computer and a GM don't find their moves in the same way, and they have different weaknesses as well. With that in mind, it might make sense to focus more on the comments from humans when trying to figure out what another human is thinking, even though a computer might be the stronger player. My two cents :-)

set kib 2 = better ICC. set kib 0 = best ICC.

for the record, Jen would tear Dunne a new one.

KB:

Interesting post;

You might be surprised at what a GM would have to say when explaining his thought process. Check the book I mention - "Think Like a Grandmaster" by GM Kotov. He even mentions the "trees" you talk about. I promise you will be surprised, because what you describe as "brute force" in computers is basically what GM Kotov wrote about MANY years ago (way BEFORE personal computers existed) in trying to explain a GM's thought process: select candidate moves, start analyzing the tree of sub-variations ("branching out"), evaluate and eventually play the move that looks more promising.

Isn't it ironic?

Wasn't it GMs Soltis and Nunn who said that what Kotov said is not realistic for a grandmaster in praxis? Nunn wrote his antithesis in "Secrets of Practical Play".

Yes, Nunn is known as having said that the Kotov method had some flaws, specially the fact that Kotov recommended examining every variation only once. Others have criticized the fact that this method is very time-consumming. Something worth pointing out is that in spite of some negative comments (some of them worth paying attention to - like Nunn's), the general opinion of GM's was very favorable.

It would be very interesting for this thread if other master-level players (Fluffy?) describe their thinking process during a game.

I don't know, Shahade's problem seems to be that there are class players shouting out computer analysis during top level games, whereas my problem is that I can't filter kibitz by rating so I have to hear them at all!!!! :-) It's quite easy to show how rediculous it is to take a computer's analysis at face value. Perfect example: take any position that isn't a dead draw (or sometimes even one that is) give it to the computer, let it spit out it's stupid little number. The make the move it suggests and repeat. This little stupid number fritz tells you keeps changing... Well, then the original number wasn't an evaluation of the position at all since if you're making best moves the evaluation of the position doesn't change!!!!! I promise you won't see a master write: "this position is +=", then makes a move, "with this move black equalizes". This just shows how worthless computer analysis can be. Another example: take some amazing sacrifice annotated from some Informant and go over it with fritz, it'll think the sacrifice is unsound, and try to "refute" it and lose, even though it's evaluation claims it is better it is actually losing, that's a true role reversal.

My two cents...

I think understanding of the human thought process has improved since Kotov's days. It is true that humans do some level of calculation, tree search, pruning etc - but the level at which they can do this is only a fraction of that which a computer can do. A human analyzes a few dozens of possible positions per move (maybe 1,000 in extreme cases), a computer analyzes millions. The very fact that a handful of humans can actually *beat* a computer shows there is something else going on besides raw calculation.

I don't blame Kotov for not knowing exactly how heis own mind worked. But the fact is, there is a lot more to GM play that tree search. In the very least, human pruning algorithms are much better than computer algorithms, and it pays to get such pruning info through human commentary.

I have read Kotov's book several times and I'm not under the impression that Kotov was trying to explain "how his mind worked". He starts the book by explaining that he was relatively weak in the area of calculcation and developed a thinking & training method to improve. That's where the trees came in.

I used his training suggestion when I was in college -- play through GM game until you reach a complex position, the put down the book and calculate in your head for an hour or 45 minutes. Then write down your tree of variations and compare it against the book analysis. This worked wonders for my ability to visualize and also helped me organize my thoughts and use my time more efficiently. It's a useful technique for those of us who (like Kotov apparently), left to our own devices, don't think very clearly or efficiently.

Fritz is an Oracle, but used correctly, Fritz is also a teacher. Fritz gives a line and you have no idea why those moves were chosen. Then you try the alternatives, see how fritz analyzes and can see in concrete terms why they aren't as good. Usually, it's a matter of there being some problem with the normal looking moves that fritz finds through brute force calculation, and that the wierd looking computer move avoids. Is this helpful to chess? I believe that's how Naka trains isn't it? Seems to work for him.

Peace...

Basically, Fritz as a tool is useful. Fritz as a crutch is not. I have a friend who goes over all of his games EXCLUSIVELY using Fritz. He does not try to identify any new ideas or any such time-wasting measure. Rather, he goes over the game, move by move, and writes down which move Fritz rates highest. If there is a move which Fritz rates .25 better, then he declares the text to be "a mistake" and annotates it with a "?", which is quite ridiculous. When asked why he plays h3 in a certain position, he answers me with "Fritz says it is best." This man was a Class A when I met him, and he has made really NO progress since that point in 2001. Sure, he has gone up, and he has gone down, but he is basically at the same point rating-wise and certainly has not gotten more formidable. No matter how people try to help him, he is stubborn and insists only on going by Fritz evals and maxims such as "I have the two Bishops, which is worth half a pawn!" It is terrible to see people thinking in these mechanical ways, and this is where Fritz poses a potential danger to those who are not schooled on how to use it as just one of several tools.

Hotep,

Maliq

Fritz control? Nobody needs no Fritz control. Jen's hair looks great.

When commenting before the live audience on one of the X3D Fritz vs. Kasparov's games, Yasser Seirawan predicted Kasparov's move g5!, clinching the draw in a rook endgame, and the audience applauded: it was a thrilling human moment. Seirawan just "knew" the move was right, he felt it: maybe it was something that had come up in a game of his vs. Portisch or someone.

Also, when commenting on ChessFM during the same match, at a point where Kasparov had white with a winning position, Fedorowicz said, "It's time to play good chess moves," (meaning it was a matter of technique) and he predicted c3 and Kasparov played it. Another great moment in the annals of chess spectatorship!

It's a chess thang. The programmers responsible for the extraordinary development of chess computers (with the exception of Hans Berliner!), and those who are slavishly dependent on chess computers, wouldn't understand.

"It would be very interesting for this thread if other master-level players (Fluffy?) describe their thinking process during a game."

You're a master, why not give us your thoughts?

:-)

Especially as a now qualifier for the US Championships.

MD wrote:

>>
You're a master, why not give us your thoughts?
>>

I gave you my thoughts already (you might want to read the entire thread - there's a lot of information there, on both sides of the argument), but I'll repeat it, just in case: you look at the position, evaluate, examine attractive-looking moves, branch out - evaluating at every step - and eventually settle for a move, hoping that your analysis is correct. Those who "see" deeper and calculate better are stronger. Kaparov sees far deeper and calculates more accurately than your average GM, so he consistently outplays them, with the occassional - and statistically expected - upset.

The key (in addition to chess "vision" and tactical skill) to being a very strong player is consistency. Very strong players very seldom make serious mistakes (this accounts for the high rate of draws between world-class players). Computers have reached that point, too, and it's proving very tough for humans to keep pace.

(Note: the above is just a generalization - there are other factors, like chess knowledge, memory, training, etc, but I just wanted to give a succint explanation, assuming - for argument's sake - that all those other factors are the same among different players).

Sorry LBacan, while your argument for deeper calculation proving stronger might work for the difference between 8 and 10 ply, it makes much less of a difference for 14 and 16 ply. Top grandmasters probably calculate fewer total variations than weaker grandmasters because their experience enables them to judge a position much better. In fact a player like Shirov probably can calculate deeper than Kasparov, but Kasparov has always toted his understanding without any real need for deep analysis in the position. I realize your positional understanding is probably nothing spectacular if you fail to understand this, but there are pleanty of situations where all the calculation in the world won't find you the right solution and you have to think more schematically. Certainly with equal understanding a player who can calculate deeper will prevail more often. Plus you failed to mention more practical aspects such as speed of calculation. If you are unnecessarily calculating deeply in a position that doesn't require it, this takes time, eventually you might be put in a position where suddenly your ability to calculate makes no difference as you do not have enough time. I believe the largest difference between a 2400 and a 2600 on average is in fact not their ability to calculate, but their ability to pick which variations to calculate which is a form of pruning that must be based on positional understanding.

Jeguteman wrote:

>>>In fact a player like Shirov probably can calculate deeper than Kasparov...>>>>

"Dear Virginia, we have heard enough."

:-)

If you have ever tried to program computer to play games, you know why they are so stupid :)

With computer chess, the most important strength factor has always been Hardware speed, not knowledge. And hardware speed goes together with pruning speed an so on...

So, what LBacan says, is correct for computers, but does not have anything to do with humans. You just can't make human mind linear!

Actually, it has been observed in AI field,that the difference between good human players and GREAT human players is how good their memory is. Better players were able to remember certain positions much better than lower rated players.

It's not likely that Kasparov's brain is working faster than for example a 1800 ELO player's brain, but he has exceptional memory of certaing positions, certain situations and also the logic's that is included in them. Kasparov will remeber ideas and positions from his previous games and also from other's games and is able to react to them the way he remembers to be correct.

But this is way off topic now. Personally, I do like computer evaluations as a support tool for analysis. I think it's also quite spectacular to see a top human player playing also tactically sound moves! And what is better tool to confirm that than "fritz"? But we need GM commentary too!

I hope you can rest the case :)

This thread is too long for me to read through. I stopped after the first 20 or so posts. i think the first critical line in the original post is:

"Hydra's brilliant combination at the end of the second game was not unveiled gradually in a dramatic way, but instantly cited by computer programs."

I read this and said alright maybe she likes mystery novels where the plot slowly unravels while someone else may not have the time for that. If i am at work and am checking on a game, computer analysis and lines tell me a whole lot in a brief time as opposed to waiting a long time. But nowhere in her post does she say or hint at the things Mig has taken it upon himself to champion while apparently coming to the damsel's rescue. She doesnt need rescuing. She never even accuses those turning on Fritz of being less than stellar. This is LBacan's assumption only given importance because of the hordes of Shahade defenders. All she says is:

"If someone wants to use computers to analyze top level live games, that's fine, but why not be courteous?"

Actually in my view, this part of her post is not even necessary since she can always ignore lines with analysis by censoring and glossing over. She can set kibs to her preference. Finally, the moderators (like zek) themselves do a good job of censoring if someone keeps shouting comp lines.

The second part of her post is a probably more important:

"Much of using Fritz well is knowing when to turn it off. (or ignore it)".

I read this and I thought: Maybe for her, for several others it is turning it on more often and spending time going over their games with Fritz. Many times Fritz throws up new lines that you would never consider and going over them adds to your chess knowledge. However there was no need to attack her for this opinion and far less reason to come to her defence.

Now coming to this "debate" between LBacan and Mig, which has very little to do with the original post:

I totally disagree with Mig (and agree with LBacan's criticism) when he says Fritz can suggest move after move of accurate but completely illogical moves and without any way to explain them other than one being .24 better than the other. Please dont insult an 1800 rated player by saying he or she cannot understand what Fritz is talking about. How did a 1300 + 1600 + computer beat all comers in the recent freestyle chess tournament? In fact Mig, you should know that you are able to converse intelligently about a chess game with someone at least 400 points above you only because of the help of our silicon friend. Sure there may be instances when a particular move is too deep for a computer to understand while it is obvious to a human, but arent you generalizing this too much?

Regarding the learning value, maybe Mig is right. For some people, there may be a lot more learning without computer analysis done by the kibitzers. But shdnt we leave it to the individuals who frequent the live broadcasts to decide this rather than adopt a holier than thou stance and preach what is good and what is bad. In fact it can be damaging to the game.

"This practice has always bothered me and my brother [IM Greg Shahade], who went as far as to instate a no-computer analysis rule on ICC coverage of the New York Masters."

Chess needs to attract people to the game. If you impose such a rule, many people will be turned off. I think the people who frequent the boards are better qualified to decide whether to ignore the comp analysis or not rather than one IM taking the decision for everyone. It is not even clear to me that people watch live broadcasts to improve their chess game. I think it is this part of her post that probably riled LBacan and some others.

Personally I find the kibitzers' comments on ICC far better and interesting than any of the commentators on Chess FM (GM or WGM). In fact, I will go so far to say that I have rarely seen an original move suggestion on Chess FM. Usually it comes out a lot earlier through a kib. Even explanations for moves come out a lot sooner on the kib board and are more comprehensive than any of the commentators' comments. There is a simple reason for this: fifty 2000 players can offer far better commentary than any one 2300 or 2400 or even 2600 player. Not to mention that half of them are armed with Fritz.

Despite this, I do find LBacan's alleged "less than stellar" analysis comment totally uncalled for.

Raindeer I believe your point about memory is actually an example where correlation does not imply causation. They did the same experiment with the pieces randomly arranged on the chessboard and stronger players did not perform any better than weaker players so it is fact likely that the stronger understand made it able for stronger players to break down the position more logically and made it easier to remember.

Very good that you brought that up!

I think what is confusing is the concept of "memory", as something which has "performance" and a structure of memory, or collection of memories, which makes it easier to remember things, and thus improves the overall performance.

If better chess players would have performed better also with randomly arranged boards, this would have demostrated that what they had learned has nothing to do with the results, and they have natural ability to break down any chess position into smaller logical set's, which are easier to remember.

In case of humans, memory improves understanding, and you remember best things you have feelings about. So, if you want to become a GM, you have to have very strong feelings about chess :-)

Peace...

For his undergraduate honors research, my brother conducted a study on whether better chessplayers are stronger because they see less. It was an interesting experiment, conducted with absolutely NO chess diagrams or even references. Rather, he tested other areas of pattern recognition and other relevant abilities and then compared how well they performed in these independent tasks with how accomplished they were over the board based on USCF rating. This was so that he could tell whether innate abilities to recognize patterns and differentiate from various conflicting stimuli with regard to which is the more important are responsible for chess prowess. Although he is now at a different institution, he has continued to work on this research with his undergrad advisor with an eye toward publishing his findings. He also noted the studies quoted above in his paper, which is what brought this to mind. Hopefully, he can get it published and then we can all have a look to see what the results bore out.

Hotep,

Maliq

Very interesting research! I can see there are a lot of problems to be solved in normalizing the results, but certainly it would be nice to know what he has concluded.

For those who do not enjoy the Tony Rook Chess.fm shows, I have a simple question, why do you listen?
Reminds me of all the folks who complain about Howard Stern, but still continue to listen. We all know that Tony is not master strength, but I feel he does a supurb job of holding listener interest. Yes he does have catch phrases, but so what? He probably went to radio school, has his license, and could work in a number of fm or am radio stations. But he's at least trying to get chess out to the public. It's easy to sit back and be critical, but if you think it's so easy to go out and do what he does, and do it better, then give it a go. Good luck and keep checkin';-)

I like using my fritz when I am watching games on playchess. Why? Be cause i tend to find that there is a lack of GM commentary on playchess. So use my very own portable GM.

Even if there is commentary, it is usuulay not rapid fire. Amateurs want an explanation for every GM move made...or at least pretty geular comments. What usually happens on playchess is that someone like Sereiwan give his opinion and then we have to wait 2 minutes more for the next comment. Thats boring.

I think a solution would be if 5-6 GM's watching the game got into a heated argum,ent over a plan...we would be getting opinions from them every 20-30 seconds. That would simply kills any thought of anyone giving a computer evaluation.

Another suggestion is, is inaddtion to this heavy GM (or at least IM) analysis, why don't playchess give a permanent computer evaluation on one of the windows? This way no one will bother to fire up their fritz...the eval has already been provided by the server. I saw this being done at the official site of the kramnik-leko match and I just didn't feel the need to turn on fritz.