TrueChess Computer Analysis: Fischer is strongest ever. Botvinnik > Kasparov

[KyleJRM82]

The whole thing just seems to fail to me because it's taking a computer evaluation that is meant only to help computers win chess games, and using it for something it wasn't intended to do: judge the pure accuracy of moves.

The only evaluation that matters is 1-0, 1/2-1/2 or 0-1, and players make their moves with those in mind.

[DaMaGor]

Originally Posted by tchaz

itt we learn that Botvinnik in the 1, 2 and 3 year periods up to 1932(!!!) was a stronger player than any 1, 2 or 3 year period of Kasparov's career

Spoiler:

butnah

Also, that Smyslov was at his best as a 55-year-old and Kramnik was at his best as a 17-year-old, and that each were better than Capablanca, Kasparov, or Karpov ever were.

That's enough to show that this study is not worth paying attention to, and even if I can't pick out the flaws, that it is hilariously flawed.

By the way, the short story I alluded to was "The Jersey Rifle", by Jack McDevitt.

[Do it Right]

Originally Posted by EGarrett

Another relevant comment he made...

"The goal was to be as correct as possible, not to adhere slavishly to the computer's evaulation. For example, I overrode the computers' decision and classified Fischer's 29...Bxh2 in the 1st game of the 1972 Fischer-Spassky match as a blunder -- true, the resulting position is not completely lost, but 29...Bxh2 turned an easy draw into a fight for survival (a fight that Fischer lost)."

BAHAHAHAHAHA. I love it. Ok, so the author creates the format for his study. When his format leads to results even he can clearly see are idiotic he simply "overrides" them and put whats he wants? OOOOKKKKKKKKKKK.

In before another rambling wall of text trying to explain why manually changing data that obviously further invalidates your experiment's methods is actually a great thing to do. Not to mention the person overriding the computer's evaluation and grading world champion caliber games is a 1700 (or unrated) player. But I'm sure there will be more rambling to also support that as well because omg this study is so amazing!

[wlrs]

In more scientific terms, both the reliability and the validity of the study seem weak since a) some of the choices made seem pretty arbitrary or even cherry-picked b) the author has messed with the data after analysis c) it is questionable whether the method would be capable of answering the goal of the research even if was conducted more rigorously. Points for enthusiasm but the results should be taken with a large grain of salt, as already argumented by several others above.

[EGarrett]

Originally Posted by Do it Right

BAHAHAHAHAHA. I love it. Ok, so the author creates the format for his study. When his format leads to results even he can clearly see are idiotic he simply "overrides" them and put whats he wants? OOOOKKKKKKKKKKK.

In before another rambling wall of text trying to explain why manually changing data that obviously further invalidates your experiment's methods is actually a great thing to do. Not to mention the person overriding the computer's evaluation and grading world champion caliber games is a 1700 (or unrated) player. But I'm sure there will be more rambling to also support that as well because omg this study is so amazing!

The sum total of your arguments in this thread appear to be self-contradictions, accusations of bias stupidity or ignorance that were demonstrably incorrect (the only thing of interest on your behalf is something that I purposefully found for you after your own arguments held no water, and have yet to examine further), arguments based on self-confusion of phrases like "what he wanted to show", yelling in all capital letters, ignoring of your own points once they are shown to be flawed, as though you'd never said them, and unwarranted and unjustified attempts to personally attack people who disagreed with you, even though it was simply shown, with no malice, how flawed each of your beliefs are, and seeming complaints about "walls of text," which really means that you don't like that other people formed arguments and expressed themselves more thoroughly, more compellingly and more easily than you could.

That appears to represent the whole of what I've seen from you here. Which is disappointing.

Please remember that this board is entirely voluntary. If something causes you to become upset and/or display this type of behavior, you're probably better off clicking elsewhere unless you can respond in good faith.

Maybe you should take a breather and we can discuss other topics in other threads in the future. I don't think you're able to give a positive contribution to this one.

[EGarrett]

Originally Posted by DaMaGor

Also, that Smyslov was at his best as a 55-year-old and Kramnik was at his best as a 17-year-old, and that each were better than Capablanca, Kasparov, or Karpov ever were.

That's enough to show that this study is not worth paying attention to, and even if I can't pick out the flaws, that it is hilariously flawed.

"A one-year period of games represents too small a sample upon which to make a judgment -- interesting, but not very meaningful."

[Noir_Desir]

his analysis is not so bad in itself, although there are certainly weaknesses like his complexity factor against which all results are weighed which seems a little contrived. Furthermore i'm still not convinced that you can simply leave out draws (which make up the majority of today's games) without skewing the rsults somehow.

But it's quite a huge leap from finding which player "has the lowest deviation from a mediocre 2007 engine evaluation over move numbers between 16 and 40" to "who is the stronger player".

[Do it Right]

And in after wall of text defending manipulating your experiment results to fit your desired biased goal.

[Do it Right]

Originally Posted by Noir_Desir

his analysis is not so bad in itself, although there are certainly weaknesses like his complexity factor against which all results are weighed which seems a little contrived. Furthermore i'm still not convinced that you can simply leave out draws (which make up the majority of today's games) without skewing the rsults somehow.

But it's quite a huge leap from finding which player "has the lowest deviation from a mediocre 2007 engine evaluation over move numbers between 16 and 40" to "who is the stronger player".

It's not simply "which player has the lowest deviation from a mediocre 2007 engine evaluation over move numbers between 16 and 40" but also his hand picked manipulations of results he didn't like, decisions to present results arbitrarily without draws and yeah his complexity factor which is inherently silly as it's an antiquated engine's results before and after an arbitrary handicap. Even well done it still results in enormous bias as it arbitrarily provides a huge weighting to positions computers struggle with yet humans find easily. And who knows how many other biases, manipulations and more generally '**** ups' he has self inflicted on the process. Actually even his results presentation seems completely arbitrary. He randomly decided to set 1 blunder equal to 3.5 'better than average grandmaster' notches. Of course 'better than average grandmaster' is a complete misnomer since it has nothing to do with GMs at all but a player's performance against crafty using more a scale with more magic numbers arbitrarily hand picked by the study's author.

It's just a ridiculous study on so many levels. Taking it seriously should be an insult to any intelligent individual's sensibility and mind. It's like listening to somebody try to prove the earth is flat after they start ranting on and stating "well first we need to set pi to 2.7231 and assume gravity is actually 2.4m/s VELOCITY" blah blah. It's absurd. When you use enough magic numbers and arbitrary decisions you can 'prove' anything to somebody naive enough. Citing this as a meaningful source resounds much as somebody citing The Onion as a literal source when writing a research paper.

[DaMaGor]

Originally Posted by EGarrett

"A one-year period of games represents too small a sample upon which to make a judgment -- interesting, but not very meaningful."

It should by no means be close enough for that to matter between Kasparov (or Capablanca) at his best and 20-year-old Botvinnik or 17-year-old Kramnik, who wasn't even close to being the best of his own era yet. (I see from ChessGames.com that Kramnik romped in several open tournaments that year. That may explain it.)

Besides, on the two and three-year analyses it still finds that teenage Botvinnik was better than (or as good as) Kasparov and Capablanca ever were, and that 1968-70 Fischer (who played a couple weak tournaments in 1968, one Manhattan league game against Anthony Saidy in 1969, and legitimately started his rise to the world championship in 1970) was superior to 1970-72 Fischer. No chance.

[EGarrett]

Originally Posted by Noir_Desir

his analysis is not so bad in itself, although there are certainly weaknesses like his complexity factor against which all results are weighed which seems a little contrived. Furthermore i'm still not convinced that you can simply leave out draws (which make up the majority of today's games) without skewing the rsults somehow.

You're welcome to compare the results with draws included to the results with them left out. He provided both. There aren't major differences but there are differences. Do you feel the draw-included table is more accurate?

But it's quite a huge leap from finding which player "has the lowest deviation from a mediocre 2007 engine evaluation over move numbers between 16 and 40" to "who is the stronger player".

I'd like to see the study done again, or something similar conducted using Houdini today. Although if they do, I prefer that they continue to leave out the opening moves, since opening theory seems to dilute the talent factor so much. (though of course some players prepared much better for the openings, but I'm not sure much can be done to take that into account fairly)

[EGarrett]

Originally Posted by DaMaGor

It should by no means be close enough for that to matter between Kasparov (or Capablanca) at his best and 20-year-old Botvinnik or 17-year-old Kramnik, who wasn't even close to being the best of his own era yet. (I see from ChessGames.com that Kramnik romped in several open tournaments that year. That may explain it.)

I'm not disagreeing with you here. I'm just saying that the minimum sample size for the 1-year table is 10 games. That's extremely small...we can't expect everything to meet expectations.

Besides, on the two and three-year analyses it still finds that teenage Botvinnik was better than (or as good as) Kasparov and Capablanca ever were, and that 1968-70 Fischer (who played a couple weak tournaments in 1968, one Manhattan league game against Anthony Saidy in 1969, and legitimately started his rise to the world championship in 1970) was superior to 1970-72 Fischer. No chance.

Well firstly, let's acknowledge that over those longer samples, the results start to reflect expectations more accurately. The three year table, for example, has Fischer, Kasparov and Capablanca at the top, which makes sense...and Botvinnik right there with them, which is what I found interesting.

It's probably true that Botvinnik was not equal to Kasparov or Capablanca in over quality of play, but the list suggests that, opening theory aside Botvinnik's talent and calculating ability may be underrated. Though his "percent better" score isn't as good as Kasparov, Capablanca or Fischer...he seemed to place highly because his rate of blunders was extremely small.

Interestingly enough, by the way, Capablanca's high placement is the opposite. Higher blunder rate, but much higher accuracy on average than what the average grandmaster was expected to find. I obviously don't know what the various numbers suggest about a player, but I would guess this would indicate that he saw a lot of dazzling combinations and picked up on many things that average players didn't, but in the course of doing so, he also would make sloppy moves on occasion or be overconfident. Perhaps Botvinnik's style was exceptionally sound, to the point of being one of the strongest players ever, without as much brilliancy? It's fun to consider.

Oh, also, Kasparov's opening knowledge was obviously far better than Botvinnik's, so I wouldn't suggest (nor do I think the study would suggest) that young Botvinnik would actually beat Kasparov in a match. Just that his raw talent may be equivalent.

[EGarrett]

Originally Posted by Do it Right

And in after wall of text defending manipulating your experiment results to fit your desired biased goal.

If you're not going to read what people say to you, then you need to exit the thread.

[Do it Right]

Originally Posted by EGarrett

If you're not going to read what people say to you, then you need to exit the thread.

Start writing anything worth reading and I certainly will. And I'd fondly request you stop trying to reference yourself in the third person plural. I'm more than happy to read and logically consider what 'people' say - what YOU say on the other hand is an entirely different matter, in this thread at least.

The reason I berated your wall of texts is because I've generally found on the internet when people find themselves trying to defend an untenable position instead of admitting this they typically choose one of a few options:

1. I'm Einstein - you aren't. Try to obfuscate the subject matter through complication. When the obfuscation attempt fails, claim the detractor simply doesn't understand what he's talking about. Appealing to an outside requisite is a common tactic: "Yeah talk to me once you've gotten a degree in theoretical astrophysics."

2. Argument by attrition - the defense by wall of text. Ignore all arguments and respond in volume. On topic, off topic - it doesn't really matter so long as it's something. When every single comment is not responded to in point, claim the person is cherry picking, ignoring your sound arguments, etc. Your posts in this thread EGarrett are a perfect example of exactly this.

3. Houdini - When you're wrong just disappear. If you never admit it, it never happened!

So no, I'm not going to play your little game. Once it became clear you lost any and all sense of objectivity or rationality I've simply started reading your posts until you state something absurd. It just so happens in an increasing number of your comments this tends to be within the first sentence or two. Eg - you referencing him hand-manipulating his results as somehow being anything but ridiculous. It's absurd, so I stop there - take 20 seconds to respond and move on to something more amusing.

If you don't enjoy reading what I write feel free to click this: http://forumserver.twoplustwo.com/pr...gnore&u=238051 I somehow doubt you'd be the first person to have done so. I, on the other hand, enjoy reading what you're writing. Not in a stimulating fashion but more of in an amusing fashion. Watching you defend this 'study' is honestly like watching somebody desperately try to prove that the earth is flat. It's even better because you're going out of your way to be as eloquent as possible while doing so, which makes it all the more amusing. So no, I'm certainly not going anywhere. But I also have absolutely nothing against you. Regain your senses and I'll happily begin treating you as a sensible individual again. Until then, carry on please.

[YouKnowWho]