Wednesday, October 18, 2017

As the vulture flies

Thanks to Ophelia we have summer-like temperatures lately. Which makes it attractive to work in the garden in the weekends. Despite the global warming, the evenings have become darker every day though, while the winter solstice nears. It is not doable to work in the garden in the dark, so more and more times becomes available for other matters. For instance for thinking about chess.

Robert mentioned a point which has to be clarified, some day. Maybe it is good to give this some attention before I dive head first in the matters which we have talked about lately again.

That point is about to investigate what happens when we stay circling like a vulture while looking at a position, in stead of diving into the first tunnel that smells to road kill time and again. It will require quite some mental discipline to withstand the old habits, of course, so it is actually some kind of meditation. I will give it a try and see what happens.


17 comments:

  1. I watched the video on Bobby Fischer that was up on The Chess Improver a few days ago. It had some things I did not know about Fischer. One of the intriguing things was his rapid rise in playing skill (and subsequent rating increase) - in one year:

    May, 1956: USCF rating = 1756
    May, 1957: USCF rating = 2231

    Bill Wall gives a record of Bobby Fischer's meteoric rise in rating (and a whole bunch of information on his tournament/match results).

    Link: Bobby Fischer's Tournaments and Matches

    It would appear that it IS "possible" to gain over 400 rating points in one year!

    All it takes is:

    (1) genius IQ (180)
    (2) skilled mentors/teachers/opponents
    (3) access to unlimited chess resources (books)
    (4) unlimited study time (unencumbered by school or work)
    (5) total dedication to chess to the exclusion of all else

    Not necessarily in that order, but all of those factors seem to be necessary.

    Little wonder that we adult chess "wannabee" improvers cannot duplicate (or even come close to) that feat!

    (I dropped out of the race as "uncompetitive" on item (1).)

    ReplyDelete
    Replies
    1. https://www.quora.com/What-was-Bobby-Fischers-IQ
      1956 Bobby Fischer was 1956 still at school, he left school with 16
      In January 1956 Fischers performance was already 2157
      Because of the factor K in the formula of elo, a change in performance is averaged with the former ratings: The K of adults is different to the K of kids, a jump in rating is so up to 4 times more difficult for adults as for kids

      Delete
    2. @ Aox:

      In 1956, Fischer experienced a "meteoric rise" in his playing strength.[84] On the tenth national rating list of the United States Chess Federation (USCF), published on May 20, 1956, Fischer's rating was 1726,[85] more than 900 points below top-rated Samuel Reshevsky (2663).[86]

      Source: Bobby Fischer

      In January 1956, Bobby, age 12, won the class B prize of the first Greater New York City Open (January 20-26, 1956). It was held at the Churchill Chess and Bridge Club in Manhattan. Entry fee was $5. The event was won by Bill Lombardy, 6-1, on median tie-breaks over Dr. Ariel Mengarini. They split $50. 3rd-4th places went to Arthur Feuerstein and Edgar McCormick. Fischer won 5 games and lost 2 games (5-2). There were 52 players in this event. Fischer tied for 5th-7th (shared with Anthony Saidy and E.S. Jackson). In the final round, Fischer was playing Rhys Hays (2059). In a particularly difficult position, Bobby thought for a long time, and then decided on a move. Bobby moved a piece then punched the clock on the next table! His USCF rating for the event was 2157. (Chess Review, Feb 1956, p. 36 and Chess Life, Feb 5, 1956, p. 1-2)

      In April 1956, Fischer won the class A championship at the Manhattan Chess Club. He won 7, drew 1, and lost 2. Max Pavey won the Manhattan CC championship.

      Fischer also was the top scorer in the 1956 New York Metropolitan League A team with 4 wins and 1 draw. An award was to be given to him at Highland Park in Brooklyn for his efforts, but he never showed up. Carmine Nigro accepted the award for him. Nigro told Brady that Fischer was master strength.

      At the end of April of 1956, Fischer's USCF rating was 2168. However, his published rating in the 10th national chess rating list (for tournaments ending before Dec 31, 1955) in the May 20, 1956 issue of Chess Life was 1726. One year later, it would be 2231.


      Source: Bobby Fischer's Tournaments and Matches

      Unfortunately, I could not find Fischer using the current USCF ratings records online; they don't go back that far. I realize that his published rating in May 1956 lagged FAR behind his actual rating/playing strength, as evidenced by the performance ratings given by Wall. The "Game of the Century" was played on October 17, 1956, during that meteoric rise in rating. Regardless, it is still remarkable to have such a large rating change in one year, even if it was delayed!

      Regarding that lag between official rating and actual playing strength: "The job is not finished until the paperwork is done." Or so the statisticians would have us believe!

      Delete
    3. We can't do anything about (1); it's genetic. Also, various studies have concluded that Masters are NOT genetically endowed with genius AS A GROUP (with Fischer and others being notable exceptions to the group norm). With the advent of the Internet and skilled computers, and the plethora (I've always wanted to use that word!) of books and DVDs available, (2) and (3) are adequately taken care of. Those of us with lives outside of chess are constrained by (4) and (5). So, your conclusion seems valid that (4) may be the most significant limiting factor constraining our chess skill improvements. For most of us, that's not really under our control either.

      Life must go on, especially if one has family and work to attend to. I've noted (but am not surprised) that those solitary figures who pursue the ultimate skills and top ranking in any field often undergo a "slide" downward in world ranking after acquiring a family, especially kids. This applies across all fields of endeavor, not just chess or sports. Alas! There are only so many hours in a day allotted to each of us, regardless of our "genius" or our desire to excel in a chosen field.

      Delete
    4. The lag between official rating and actual playing strength is not only caused by slow paperwork, its a "problem" of the formula how new ratings are calculated. The factor how much the rating changes if the performance change is called K. This K is different for kids and adults. So it is possible for a Bobby Fischer to gain 400 points in one year.. but after the age of 18 thats not possibe anymore.

      Delete
  2. (1) genius IQ (180) - You don't have to be a genius, just better than everybody else.
    (2) skilled mentors/teachers/opponents - You need opponents/people who will stretch you, particularly in long, slow games, but this is a given.
    (3) access to unlimited chess resources (books) - I have less than a hundred chess books and think I have too much, too many books, and they are highest quality, I scrubbed my collection thoroughly.
    (4) unlimited study time (unencumbered by school or work) - You could throw this in the "given" pile, since once you get past this obstacle, you'll realize it doesn't say much about what happens next.
    (5) total dedication to chess to the exclusion of all else - Other things can inspire you, so you really need _quality_ training, and the right types of training for where you are at in your development.

    Bobby, like Magnus, like Kasparov, had their own "secret sauce", but they all studied with coaches/others, so they got that out of the way - Fischer w/Lombardy, Collins, Browne, etc., and books. The talent part, 500 point jump in one year part, has to do w/influence of others but primarily has to do with something they are doing different than others, or they just rolled out of bed one morning and their brain was then "wired" correctly all of a sudden for pre-eminence.

    Looking at Magnus' games, he was always a strong defender, except sometimes when he attacked weakly. looking at his games at chessgames, he sure played a heck of a lot of high level tournaments in Norway before branching out from there, and the first games/opponents were not so high-quality.

    One thing I definitely notice in quick-minted Masters is that they always have access to the very strongest local tournaments and players. They will be there bright and early to every weekender, EF paid, possibly hotel room paid (by parents of course). It's this access that I have empirically seen as the sole determiner, if there were such a thing, of who will be what. Who can get to all these tournaments each time, even out of state tournaments. If you look at top players you will see access to top tournaments, and tons of them, prodigious amounts of playing (no pun intended, or maybe intended yes). Underneath all of this buried something special, but that special seed down below is layered with a ton of fertilizer above it. This is why I am more blown away by how they made such progress in the 1800's, but even then it's the same formula - play the top guys in Austrian coffeehouses, then move to Paris, London, etc. These guys didn't step out of "backyard" professorville and then just start taking everyone down in major metropolitan areas from there.

    Strong opponents will stretch you, and force you to work past your limitations. It's not that we need strong opponents so much as that we need someone to point out our flaws, which cracks the whip so to speak, of improvement.

    ReplyDelete
  3. I saw a YouTube video by IM John Bartholomew regarding the Elometer, which estimates your playing strength, so I thought I'd give it a try. There are 76 problems and a bunch of questions regarding past experience and current training.

    Link: Elometer

    Link: Estimate Your Chess Rating! Elometer.net Test

    Well, here's the "shocking" result:

    Based on your move choices, our estimate of your Elo rating is 2238, with a 95% confidence interval of [2102...2374]

    You have to pick the first move only, not work out all the variations. I didn't guess at the moves, but I'm pretty confident (100% level) that I am nowhere near an Elo 2238 rating. Heck, I doubt very seriously I am anywhere near the lower bound of 2102.

    Has anyone else tried that test?

    ReplyDelete
  4. I don't know what happened to the first link (to the Elometer site). I may have entered it wrong or mistyped the HTML. In any event you can find it at "www.elometer.net". Link (another try): Elometer

    I was very surprised to see IM John Bartholomew's rating (at 18:15 in the second link above):

    Based on your move choices, our estimate of your Elo rating is 2240, with a 95% confidence interval of [2051...2430]

    Based on THAT, I conclude conclusively that my "real" rating is NOWHERE near a practicing International Master's rating, much less within 2 points of his rating. Oh well, it was still fun solving the 76 problems.

    I went back to Google and looked for others who have taken the test (Google "Elometer"). The general consensus is that this test rating is "highly over-rated."

    ReplyDelete
    Replies
    1. My estimated rating at Elometer was about 2300 (95% confidence interval). Take notice I am only 1730 ELO rated patzer. This test tests if you basic ideas and simple concepts. I am neither 2300 nor 2100. I would be very proud if I could play at the level of 2000.

      Delete
    2. I just tackled a few of these puzzles at the elometer, these where some famous positions and studies. When you did read a lot, you simply know these positions.

      Delete
    3. I concur with Aox: as I went through the puzzles, I recognized at least 10 of the 76 from previous study. (Hey! Isn't that what we are SUPPOSED to do - grind positions into LTM so that we can recognize the patterns?!?) That is one of the questions that the survey (after the 76 puzzles are completed) asks. If I had known that there were survey questions after the puzzles, I would have counted more accurately the ones that I was familiar with. Given we have little information about the underlying methodology, only speculation is possible about what impact "pattern recognition" plays in the final estimated rating. I suspect that I got a much higher rating than expected simply because I could recognize several of those familiar patterns/positions. [I have a near-photographic memory for things I have studied deeply.] I'm skeptical about any small sample set of problems being sufficient to give a statistically significant "rating," especially if the problem set is primarily composed of widely known positions. But then, I'm neither a statistician nor a psychologist by training.

      I think that the many deep investigations and discussions on this blog have been immensely beneficial by giving me better quality tools with which to analyze positions. Getting a higher than expected "rating" from this experiment is nothing more than encouragement to continue the current study regimen.

      Many thanks to Temposchlucker and the various commenters here!

      Delete
    4. There is a related book: Chess Exam and Training Guide: Rate Yourself and Learn how to Improve! By Igor Khmelnitsky
      He did use 100 positions and a lot of math and statistics to calculate ratings at subskills of a chessplayer. The books was very instructive and entertaining.. but..
      I suspect that he did not have enough pupils and that his pupils had the same american background
      Americans read other chessbooks than germans or dutch. I was very suprised to find a lot of "complicated" puzzles at the tactics trainer very low rated until i recognised that these puzzles where taken from 1001 Brilliant Ways to Checkmate
      and other books of reinfeld, popular in the USA but not in germany
      So a skill like "knowledge of standard positions" is depending of the "standard books" in that country.
      And.. 100 positions .. then you need several hundreds ( better thousands ) of students to make a real statistic, i have doubts that they had so many.

      if you want to know how good you are.. go to chess.com and play some games. Or use an engine , many engine can simulate a given strength. Or use the rating estimate at chesstempo. That is based on more puzzles and more serious statistic.



      Delete
    5. I just made another pass through Elometer, to try to find out exactly how many positions were "familiar" (either through immediate recognition of the specific problem or the general procedure (like setting up a Philidor position in a Rook and Pawn ending.) 45 of the 76 problems were either recognizable or very familiar with the correct approach; 31 were not recognized. That is considerably more than I originally "guess-timated" (at 10). FWIW (not much), my "ELO rating" went up to 2303, with 95% confidence interval of [2159..2448]. I didn't cheat with a computer, but did take my time (since it is no longer timed). My only purpose was to see how many of the problems I actually recognized, not move an already highly inflated rating higher.

      @ Aox:

      I already play Stockfish on "lichess" alternating colors with the time set to 5 minutes (with 5 second interval) and on AI level 5 (which is rated 1700). I win most of my games, if I'm careful. Whenever I get to the point that I don't have to be careful, I'll move up one level, and repeat the process. BTW, this is the major reason why I KNOW the "Elometer" ratings are highly inflated.

      Delete
    6. I just took the endgame test at Elometer. I got 40 out of 48 correct. The ones I missed were positions involving two knights against a single Pawn and similar positions, and rare endings (like two Bishops versus Queen). There were more of the rare endings than I would have liked, and the answers were based on tablebases which extended to 70 moves (in at least one case). Ain't no way I'm EVER going to even try to work out an ending to the 70th move!

      The absolutely WORST ending I've ever had to defend was a Rook+Bishop versus Rook ending. My opponent and I were both rated in the low USCF 1800s, and were the two highest rated players. It was an OTB classical time rated game, and lasted almost 6 hours until I finally forced the drew by exchanging off the Rooks. As a result, I dropped out of tournament play and have never played classical time rated tournaments since then (AUG 1975). The only rated games I've played since then are "Quick" games (under 10 minutes for each side).

      In my games against Stockfish, I often find myself grinding out a win (somewhere between 50-75 moves). I guess it's good practice for "technique." I never been very good at "risky" sacrificial attacking chess.

      Delete
  5. I went through the first 14, and since I can see I won't get their solutions at any point I stopped. Also, I realize that they don't have the 30 second timer anymore, no timer. Well, with no timer I'm gonna totally max out this test (recognized quite a few of the tactical problems), and with 30 seconds I'm gonna get like 1600 last place or something. All you need to verify is that a Master is a Master because they can solve hard problems in under 30 seconds as compared to the rest of us. Speed, consistency, instincts are a big part of being a Master. Masters can calculate fast and deep, but they still know what to look for, and have those instincts of what's worth noticing instantly and what's worth never bothering to notice.

    ReplyDelete
  6. I went in for a sleep apnea study last night. After getting hooked up with all the electrodes (looking like a poorly wired Borg from Star Trek), I had a difficult time getting to sleep. So, I decided to work out variations of a relatively simple endgame that (I think) was on the Elometer test. Effectively, I was solving "blindfold" since the lights were off and I had no board with me.

    FEN: 8/p7/8/PP1k4/1K6/8/8/8 w - - 0 1

    The first approach that came to mind was to gain the opposition, and (eventually) force the Black King back until the b-Pawn can be advanced. My first thoughts involved two generalities:

    (1) I did not want to be left with an a-Pawn (Black can easily draw by either occupying the corner or by forcing the White King to occupy the corner without letting him escape.

    (2) The White King MUST get in front of the (remaining) White Pawn (after the inevitable exchange of the Black a-Pawn), which seems impossible to accomplish.

    After experimenting with various White King moves (especially 1. Kb3 to take the direct opposition 1. ... Kc5 2. Ka4 [Black can never cross to the 4th rank because a simple advance will promote one of the White Pawns] 2. ... Kd6 3. Kb4 Kc7 and the Black King can prevent Pawn promotion.), I came to the (correct) conclusion that Black could draw with best play. He simply heads for a8, and White will be unable to dig him out of the corner.

    That leaves White with the two possible Pawn moves: 1. b6 and 1. a6. I first rejected 1. a6 because it seemed to be "backwards" to the "rule" about creating a passed Pawn by advancing the "candidate" (the unopposed Pawn). However, that's easily shown to allow Black to get his King back in front after 1. b6? Kc6 2. bxa7 Kb7 with the Black King stopping the a-Pawn.

    Finally, I set my mind to analyzing 1. a6. After all, it is the "outside" Pawn (even if NOT passed at this point). Here I could have saved myself a lot of calculation by simply relying on the "rule of the square". Black will have to move his King outside of the promotion "square". Surprisingly, this is the only way to win because of a neat tactical consideration at the end! 1. ... Kd6 2. b6 Now Black has a choice of which way to lose.

    2. ... axb6 3. a7 and the Pawn queens.

    2. ... Kc6 and it appears that Black will be able to stop the b-Pawn if it advances: 3. b7 Kc7 and Black can never be dislodged from c7 and b8. BUT: 3. bxa7! and now the Black King cannot get to the promotion square because the a6-Pawn covers the b7 square! 3. ... Kb6 4. a8(Q) and Black can still move to c7 (no stalemate, although it is close).

    What did I learn from this exercise? (1) If you can "see" the right approach, you can save yourself a lot of calculations. (2) When you have the opportunity to promote, ADVANCE THE PAWN. (3) Generalities (like the opposition) often don't get to the heart of a concrete position. You still have to consider the nuances. (4) I CAN "see" the board clearly in my mind and "see" what the ramifications of various moves are, IFF I focus on the position INTENTLY. I suspect (based on previous tactical training exercises) that this is as important as (or more so than) trying futilely to cram 100,000 positions into my LTM. (5) THIS is THE way to "study" the endgame (and tactics and ...), rather than "reading and nodding" superficially based on other peoples work.

    When I got home this morning, I loaded the position into the Nalimov endgame tablebase and saw that I had analyzed correctly. If you want to see the alternatives (and results) that the Nalimov tablebase recommends:

    Link: Nalimov 6-Man Endgame Tablebase

    ReplyDelete