|
< Earlier Kibitzing · PAGE 26 OF 92 ·
Later Kibitzing> |
Jun-15-06
 | | Gypsy: <keypusher> You are welcome. |
|
| Jun-15-06 | | RookFile: It was an interesting read. |
|
| Jun-15-06 | | whatthefat: <RookFile: what I'm saying is, the Sonas system arbitrarily rewards players for playing more games> And what I'm saying is that the effect is negligible once a reasonable number of games are played. That's patently clear from the mathematics. <but due to the Sonas approach of 'more games is better', Fischer's results get undervalued.> As now pointed out twice, this is false. Fischer played over 100 games in the 4 year weighting period, so the padding effect is negligible. Whether you play 100 games, 1000 games, or more, is mostly irrelevant. It's only when the number of games becomes very low (like after 1972 for Fischer...) that the effect is at all significant. It's a way of naturally excluding inactive players; not separating the active players. Until you understand this, you're just treading water. |
|
| Jun-16-06 | | RookFile: I guess you're arguing with Sonas himself, and am not sure if that is a good idea. Sonas himself says that his system rewards players for playing more games. You can throw in all the 'until you understand' comments you want, complete with condescending tone, but I'm not sure it makes your point. |
|
| Jun-16-06 | | acirce: <Sonas himself says that his system rewards players for playing more games.> What's wrong with you? <whatthefat> doesn't deny this, he says that the effect is negligible when you play as many games as Fischer did. I'd like to see you argue against this. I'm not sure it's possible, but at least it would be better than to keep repeating what everybody already knows. |
|
| Jun-16-06 | | RookFile: I guess what's wrong with me is, I take the time to read what Sonas wrote. You guys don't.
"Anyway, I wanted to reward the players who played more games, and I also wanted to reward the players who faced tougher opposition. I took the same approach again, and experimented with various modifiers to the basic performance rating formula, to see what worked best at predicting future results. I came up with something that not only works well, but I think it also makes a lot of sense too." "Before I go any further I just want to emphasize a very important point here. This system rewards players who play a lot of games. That's because there are always 7 "fake" games that provide the padding. If you play a lot of games, the 7 games don't have much effect on your rating. If you only play a few games, they can play a dominant role in your rating, as they should! Now think about that four-year time window. Your games are weighted linearly, so maybe a particular game gets a 36% weighting one month, but the next month it is slightly farther back in the past, so it only gets a 34% weighting. " What he basically does is "pad" in some non-existant draws against the same level of opposition. But, the results are also weighted - meaning, the fact that Bobby Fischer played 936 games is NOT important - when looking at results in 1972, victories by Fischer prior to 1968 are NOT helpful to him. Even the fact that Fischer is winning everything that matters from 1969 to 1972 is downgraded, because he wasn't like Karpov and travelled around (on the USSR's money, by the way), winning a lot of professional, although not world championship related, tournaments. |
|
| Jun-16-06 | | Dick Brain: I like the idea of dropping the estimated rating over periods of inactivity although I would like it more if it were dropped realistically - i.e. it should fall at the average rate for the decline in skills of an inactive player. It seems to fall by much more than that. |
|
| Jun-16-06 | | RookFile: Take Botvinnik for example. The guy was basically retired after the Petrosian match, but when they needed somebody in 1970 for the USSR vs. the world match, he was strong enough to come out of retirement and defeat Matulovic 2.5 to 1.5. |
|
| Jun-16-06 | | DUS: <All of these are considered GM today. But they did not pass the Ostende GM GM qualifications back then.> It seems you are talking about simply "GM" title and also "GM GM" title? If not, why they have to be considered GM today (or yesterday) if they didn't pass Ostende qualifications. <All of these are considered GM today.> -- By whom? By the chess federation of your country? It is interesting that now you are saying "considered GM" and not "were GM". |
|
Jun-16-06
 | | keypusher: <DUS> Judging from your prior post re Nicholas II, you had a lot to learn (as I did) about the use of the term "grandmaster" in the days before FIDE ruled the scene. Gypsy provided a lot of information. Would it be too hard to express a little gratitude, instead of trying to pick nits? |
|
Jun-16-06
| | Gypsy: I decided to find out for myself where or, better, how Burn got the points for his high ratings. His games from the Burn variation of French are quite illustrative. You probably will not like some of them, unless you you also find pleasure in games of Michael Adams; or, say, in games of Karpov or of Petrosian himself. Burn was a great 'spider'. Relative the standards of his era, Burn was a superb defender and a superb endgame player. Like Petrosian, he was also a good tactician, and he did attack when the position warranted it. But most of Burn tactics served subtle defensive and endgame objectives. He was a fine master of combining the defense with a counterattack. And I found him surprisingly Karpovian in the way he used dominance in the ways Karpov defines and uses the concept. In a typical Burn game, he survives opening with a worse but playable position. Then he defends throughout the middlegame and strikes back with one or two sorties. His pieces return just in time to cover the danger brewing around his king. This brings on a tour-de-force transition into an endgame -- and the transition itself is the finest aspect of Burn's play. Burn's play in the endgame is of a very high standard, but so often there is simply nothing much left to do, but give oponent enough time to examine the position and realise that it can be safely resigned. And if there is something left to do, Burn is usually up to the task. No need to play this endgame:
Schlechter vs Burn, 1897
A nice queen ending:
Pillsbury vs Burn, 1898
A counterattack in the more classical style:
G Marco vs Burn, 1895 |
|
| Jun-16-06 | | DUS: <Gypsy> - Thank you very much for detailed and interesting chess history. Sorry if I misunderstood something. I guess one can also ask questions. Let me to ask one more question. <That was only a silly pomp-and-circumstance event> Why to use such words for an event with participants like Lasker, Capablanca, Alekhine, Tarrasch and Marshall? Because no names of Burn, Blackburne, Yeats, or Penrose? Or maybe you didn't like that "the tournament was won by Lasker ahead of Capablanca"? I guess that time the great Lasker and the great Capablanca were definitely better than the great Alekhine, the great Tarrasch, and the great Marshall. <keypusher: I guess FIDE only conferred the title on masters who were still alive in 1950. Burn and Blackburne were not.> Of course, FIDE couldn't award them GM title in 1950. And there is no reason to call them GM in our usual terminology. As there is no reason to call a GM the great Paul Morphy even he was one of the best, or perhaps the best, players of all times. |
|
Jun-16-06
| | keypusher: According to <percyblakeney> here is Burn's own favorite of his own games. Burn vs Steinitz, 1898
Since Percy owns a 1,000-page(!) book on Burn, we must respect his authority. Thanks for digging up those other games, <gypsy>. The Schlechter and Pillsbury games were new to me. I was familiar with Marco-Burn because I have the Hastings tournament book. It always reminded me of this game: J Peters vs Seirawan, 1984
|
|
| Jun-16-06 | | DUS: The languages (German, English , or Russian) were rich enough and had the words "grand" and "master" back then and clearly some chess masters and tournaments have been called with such words. But it doesn't mean that today one can use this language "trick" to "award" the title "GM" to even good players who did not have GM title at its time. Thanks for reading my opinion. |
|
Jun-17-06
| | Gypsy: <DUS: ... <That was only a silly pomp-and-circumstance event> Why to use such words for an event with participants ... > Oh the tournament was fine, one of the finest ever. It is all these 'title awarding ceremonies' (pomp and circumstance) that I think of as being kind of silly and amusing. We, humans, like such ceremonies. Heck, I personally like such ceremonies. But when one thinks about it, they are rather silly. My point is this: The standard used to be that the title of a grandmaster was earned, over the board, by winning big, competitive tournaments of masters. And I think it is a generally good idea to respect such conventions and to respect such grand achievements. I think we should not redefine these standarts post mortem; others may differ. Say Rubinstein: he won about 8 big tournaments between 1905 and 1914 (some other tournaments may also count). That safely makes him into a grandmaster in my book, even though he choked and finnished 6th at Petersburg in 1914. Burn won a fair number of big tournaments towards the end of 1800s and that too makes him into a GM in my book. <Or maybe you didn't like that "the tournament was won by Lasker ahead of Capablanca"? > Contrary, I liked that quite a bit. I usually root for the older guy. While I belive that Capablanca would have already prevailed in a match, I quite savor the cunning with which Lasker found a clever winning angle for their key game. Till the end of their days, Lasker was capable of defeating Capablanca in individual encounters; in a match, I would bet on Capablanca, but Lasker was always able to figure something clever for individual games. |
|
Jun-17-06
| | Gypsy: <keypusher> Thx, that is a one fine game ... if one has aquired a taste for off-collor bishops, long king walks, bizare endgame traps. The Seiravan game is also a gem! It took me a while to find Burn vs O Chajes, 1911. If I remember correctly, it was a game of the last round of Carlsbad 1911. Burn and Chajes got a special disposition to play the game a day early, during the day day for finnishing adjournments, so that they could depart. Then they played the game for two days straight, interupting only for food and/or sleep. It was the last game finished and they both received a standing ovation from other players and spectators. The tourney organizers voted to create a special prize and award it to both for their fighting spirit. |
|
| Jun-17-06 | | whatthefat: <RookFile>
Perhaps you have read what Sonas has to say. But in that case, your skills in mathematics must be rudimentary at best. For the absolute last time: FISCHER PLAYED OVER 100 RATED GAMES IN THE PERIOD 1968-1972. PADDING EFFECTS ARE THUS NEGLIGIBLE HERE. And I'm not going to continue repeating myself. You've evidently come to this discussion from a biased perspective, but unfortunately for you, that can't be imprinted on mathematics. I'd advise you to hold your tongue when discussing matters where you demonstrate no expertise. Even if you happen to hear that magic word: "Fischer". |
|
| Jun-17-06 | | whatthefat: While I'm at it, here's the average number of rated games during the players' respective 3 year active peaks: Lasker: (1894-1896): 28
Capablanca (1919-1921): 2
Alekhine (1930-1932): 25
Botvinnik (1945-1947): 16
Tal (1959-1961): 67
Petrosian (1961-1963): 52
Fischer (1970-1972): 38
Karpov (1988-1990): 77
Kasparov (1989-1991): 35
Of the WCs, Karpov stands out as the most active, but amusingly, it's Tal, Petrosian and Fischer next! Sorry <RookFile>, but there's no chessmetrics conspiracy against Fischer, and there most certainly isn't one against Petrosian! If anything, it seems pro-Karpov. Yet a moment ago you were saying it was pro-Kasparov... oh dear... |
|
Jun-17-06
| | keypusher: <gypsy> Amazing! I copied your note to the Burn-Chajes game page. Burn was 62 or 63 when it was played... |
|
| Jun-17-06 | | RookFile: I guess it's been conceeded that the fact that Fischer played 936 games isn't going to help him, as I was being raked over the coals not so long ago. Now we need to look at further analysis. I'll do this later, after returning from a wedding. The first error that immediately leaps out at you, is Fischer's 3 year peak is given by Sonas as 1971 to Dec 1973, not 1970 to 1972 as incorrectly given by whatthefat. We'll give this a comprehensive perusal, and help bring the truth to light. |
|
| Jun-17-06 | | RookFile: So, continuing to look at this rationally, Fischer's 3 year rating peak is given as 1971 to 1973 by Sonas, not 1970 to 1972 as incorrectly given by whatthefat. I don't know what Fischer was doing in 1973, other than appearing on tv shows, playing tennis, etc. It does seems bizarre that Sonas chose to leave out 1970. Well, maybe not so bizarre, you have some inconvenient results like Fischer smacking Petrosian 3 to 1, for example. Remember also what Sonas said: the games are 'weighted': "Now think about that four-year time window. Your games are weighted linearly, so maybe a particular game gets a 36% weighting one month, but the next month it is slightly farther back in the past, so it only gets a 34% weighting. " This simply means that the games in 1972 factor a lot more into calculation of the 3 year peak than those a few years past. So - Sonas wants us to look at 1971 to 1973. 1973 is easy, Fischer didn't play any chess during this time. 1971 and 1972 are not hard either, in terms of rated games (not including the forfeit loss to Spassky), here's what you've got as the rated games that by definition will get the vast majority of the 'weight' in calculation of the 3 year peak range: Taimanov 6 - 0
Larsen 6 - 0
Petrosian 6.5 - 2.5
Spassky 12.5 - 7.5
Total: 31 - 10
So, Fischer put 31/41 up on the board.
I don't know what the average rating of these 4 players was, but let's say it was 2662. Now, here comes the problem: Sonas wants to 'reward' players who play 'more' games. He does this by 'padding' the weighting results with 7 non-existant draws against the same caliber opposition. In a sample size that consists predominately of only 48 games, adding in 7 drawn games against 2662 rated opposition means that you are effectively 'diluting' Fischer's achievement by a factor of 17 percent. Fischer played some more games in 1970, of course, winning everything. Palma and beating Petrosian come immediately to mind. (You also have to be careful when scanning chessgames to remember to leave out Fischer's impressive wins at Herceg Novi in 1970 or the Manhattan chess club, as these were blitz games that Fischer won by the ridiculous scores of 19/22 and 20.5/21 against some of the best blitz players around.) However, because of the 'weighting' of result, these result mean less under the Sonas system than Fischer's wins against Spassky. So yes - padding in 7 non existant draws does hurt Fischer's calcuation of his 3 year peak, which was 1971 to 1973, according to Sonas. |
|
Jun-17-06
 | | Benzol: What does the 1992 Match do to Fischer's rating on the Sonas site? |
|
Jun-17-06
| | keypusher: <benzol> His old rating had expired by 1992. His rating for the 1992 match was 26-something (chessmetrics ratings are generally higher than Elo ratings). Here, you can look it up! http://db.chessmetrics.com/CM2/Play... |
|
| Jun-18-06 | | whatthefat: <not 1970 to 1972 as incorrectly given by whatthefat> If you'll read what I said, I said "3 year active peak". The fact that the Sonas system includes 1973 is not remarkable at all. You yourself have been claiming that chessmetrics unfairly punishes inactivity. Supposing inactivity weren't punished at all - as you might prefer - Fischer's rating would remain unchanged from 1972 until 1992. In this case his 3 year peak could be given as 1972-1974, or even 1981-1983 if you like. <In a sample size that consists predominately of only 48 games, adding in 7 drawn games against 2662 rated opposition means that you are effectively 'diluting' Fischer's achievement by a factor of 17 percent.> This is a naive (or just plain incorrect) approximation. A better approximation is to weight the games in each preceding year in the ratio 1:0.75:0.5:0.25, due to the linear weighted averaging. His rating in each year is then determined (along with the 7 game padding) by: 1970 - 86 games
1971 - 81 games
1972 - 73 games
1973 - 44 games
1974 - 16 games
1975 - 5 games
1976-1992 - 0 games
The padding starts to noticeably hurt his rating in 1973, when he becomes <inactive>. It negligibly affects his rating in 1972, 1971, 1970, or before then, when he was <active>. Why is this so difficult for you to understand? The reason Fischer's overall 3 year peak includes an inactive year, is simply because he had no other sustained periods of brilliance in his career, unlike most of the other greats. And that's nobody's fault but Fischer's. <For reference in the last point, Fischer's great run began in 1970, with either Palma (November); or if you want to try and go further back, you could arguably include his 3/4 in the URS-World match (March), and Buenos Aires (July). It finished in July of 1972 with the Spassky match. So it's certainly no more than 27 months.> |
|
| Jun-18-06 | | RookFile: keypusher: I agree that if you're just looking only at the Fischer vs. Spassky games of 1992, a strength of 2650 on average is reasonable for Fischer. This 'on average' takes into account 'good days' when he plays much stronger than that, and 'bad days' when his strength is weaker than that. |
|
 |
|
|
< Earlier Kibitzing · PAGE 26 OF 92 ·
Later Kibitzing> |
|
|
|
Kibitzing
