|
< Earlier Kibitzing · PAGE 551 OF 963 ·
Later Kibitzing> |
Jan-09-10
 | | jessicafischerqueen: <Travis> no- your detailed comments on the images I use, how you feel about them, and your analysis of the themes is crucial feedback that helps me understand what I have done- And it also helps inspire me to try to do more of the same things and also try different things. Somebody who has as fine an ear for good music as you should never, ever second guess your value as a critic or appraiser of art in general. What's made me say that about the other videos is a very particular flaw in them that is purely technical- I didn't know how to use the software controls properly, so the image transitions were all jarred and shimmied- and now when I watch them, all I can notice are these "jumping" and "jarring" and "shaking" image transitions- so I think Doh! maybe I should have learned how to use the bloody software before making a dozen videos.... |
|
Jan-09-10
 | | Domdaniel: <Jess> They're still good, for the reasons that impressed Travis and me. Even if your technique -- hmm, that word yet again, whatever it means -- has improved, you don't want to delete your baby steps. Would you destroy all evidence of chess games played before you *got good*? You *would*?!?
Then it's true ... you have Bobby Fischer's genes in your genome ... maybe *all* of them, alongside big chunks of Peter Gabriel. Anyone named after a saint and a boss angel can't be all bad. As Pynchon wrote someplace, a younger self is a fool. You wouldn't loan them money, but neither would you turn them away if they arrived on your doorstep with a saw, a pregnant girlfriend, and a weird story about being followed everywhere by kings, stars, and wise men from the east. You would gently explain that perfessers in Harvard, McGill and MIT don't stalk people, then you'd park the girlfriend in your stable, if you had a stable. And you'd try to wean Younger Self off whatever chemical caused the paranoia. Just as a SWAT team from Cambridge University breaks down your garage door and sacrifices your pet ox to Baal. Margaret Thatcher studied chemistry in Cambridge, so the pieces all fit. Older selves can be fools too. |
|
| Jan-09-10 | | nimh: <I'd guess it had an effective ply depth of about 12-ply, reachable in about five minutes. (In blitz, it fell into cheap traps and was easily mated).> Your program was relatively weaker as compared to humans at shorter time controls? It's also interesting that some players had tried to play tactically against a program. Wasn't it widely known at that time that computers strength reveals more in calculation? |
|
Jan-09-10
 | | jessicafischerqueen: well maybe it's natural to like best the last one of anything you did- At any rate <Dom>, did you see the last one I linked there? Like you, I am also a fan of monochrome film and I enjoyed <Jim Jarmusch's> early work for that reason. And <Paul Coates>, my favorite film academic, wrote in his fine "The Gorgon's Gaze" that "monochrome film stock peels off reality in a way that color film cannot." Also, I recalled "someone is killing all the black and white animals" and decided to fight back, acting most proximally on your suggestion to experiment with the black and white images. So the last video I linked here, "Irish rain", is constructed entirely from black and white photos and I'm very keen to hear what you might think of the finished product. I thought the subject matter of the song (starving grieving farmers) also called for a stark visual treatment. So <Dom> if/when you get time/inclination, please give me your impressions-critique on this video? No rush of course- you are not getting paid for this. Well you are in the real world, but not by me I mean. Although I suppose I could barter something- possibly some background research on the next nubile starlet you interview, something like that. Here is the video link again-
"Irish Rain" (Andrew wrote this while living in Newfoundland) http://www.youtube.com/watch?v=3Zak... |
|
| Jan-09-10 | | crawfb5: <Who breaks a butterfly on a wheel? If it ain't broke, use a bigger wheel.>
I am bound upon a wheel of fire, that mine own tears do scald me like molten lead. |
|
Jan-09-10
 | | Domdaniel: <nimh> -- <Wasn't it widely known at that time that computers strength reveals more in calculation?> Not widely, circa 1989. Most 'chess computers' were actual chessboards + sets aimed at the amateur home player. Also expensive, about £800 for the stronger ones (far beyond my reach at the time). Their 'features' were things like touch-sensitive boards, not analytic ones. Even in the early 1990s, engines as we now understand them -- a software package running on a PC platform -- were relatively rare and little-used by stronger players. I don't think this attitude changed much until the launch of Chessbase and early versions of Fritz. I had *tried* to use the Psion program to construct a small database -- it had a 'save game' feature designed to store an interrupted game for later resumption, or to keep games for playing over. But it was non-PGN, each game was saved as a separate document, with zero search or sort functions. I inputted over 100 games manually, then got bored with the idea. As for the (human) players ... they had reached 2200 or higher without using computers. By then, quite a few people had other experience of non-chess databases etc -- I was programming in Basic in the late 70s, and used serious computers for university work and journalism in the 80s -- yet many people didn't see how computers would revolutionize chess. It took the emergence and convergence of three things -- strong chess engines, searchable databases with large game collections, and the spread of the internet/web beyond colleges and institutions -- to do that. So our hypothetical 2250-rated human in 1989 has perhaps played with a 'home chess computer' and easily beaten it. He knows they're getting stronger but doesn't really believe they can beat him -- maybe the champion programs running on big computers could, but not (he thinks) some commercial chip in a board. So he tries to smash it the way he'd beaten other computers, with gambits and sacs, and gets a nasty shock. What surprised me was the fact that after Mephisto scored four straight wins like this, its 5th opponent took the same aggressive approach, losing again. Only in round 6 did somebody have the wit to play solidly and draw. Human pride was felt to be at stake in those days. I can remember a whole series of 'firsts' -- first comp to score a master norm, first comp to beat a GM in blitz, first to beat a master in a classical match, and so on. Eventually, human pride knew it was beaten. BTW, one anomaly that I've never found a good explanation for: Tigran Petrosian. Even recently, I played through several of his best games with Fritz running in the background. In some cases - enough to be of statistical interest - Petrosian would get a negative eval from the engine almost up to the point where his opponent resigned. The standard explanation is that his 'difficult' positional style, with latent dynamics, would 'mesmerise' his opponents and lure them into a fatal error. At which point Petrosian pounced. This is true in many games. Petrosian was objectively a little worse until the opponent blundered. But in some cases it's hard to find a serious mistake - just gradual drift into a bad position, which even the engine doesn't see until it's too late ... any ideas? |
|
| Jan-09-10 | | Ziggurat: Part of the Petrosian paradox could be this: the heuristics (rules) programmed into the computers are based on "common knowledge" about how chess should be played. I'm talking about the "evaluation function" that tells the program how desirable a position is. The rest is brute-force search and bells and whistles like tablebases and opening books. Now, Petrosian may have been adept at obtaining positions where common sense rules don't apply, or perhaps his evaluations were frequently more accurate than the "common sense" version. Then, Fritz would actually be wrong in the games you mention - Petrosian in fact had the correct evaluation, which was beyond the capabilities of Fritz's efficient but hopelessly dull evaluation function. I don't know if that theory actually explains anything. |
|
Jan-09-10
 | | Domdaniel: <Zig> I'm tempted by that theory too. But, if it were true, then the gap between the comp eval and the <Tigran eval> should shrink as engines get stronger (both in heuristics and brute processing power). I haven't seen evidence of this yet. Admittedly I haven't checked Petrosian on the latest Rybka. I agree that Fritz is probably wrong in the critical cases, but it's beyond my ability to judge. |
|
Jan-09-10
 | | Domdaniel: I've been making ... |
|
Jan-09-10
 | | Domdaniel: ... too many long posts. Also risotto. |
|
Jan-09-10
 | | Domdaniel: So I'll just *twitter* while eating it.
Or try dignified silence. |
|
| Jan-09-10 | | Red October: <This is true in many games. Petrosian was objectively a little worse until the opponent blundered. But in some cases it's hard to find a serious mistake - just gradual drift into a bad position, which even the engine doesn't see until it's too late ... any ideas?> the engines are still a bit off their eval in closed positions.. the tactical ideas too are often too deep for them in such positions, so since the "objective" strengths and weaknesses lead to a different eval.. recently I have played against a guy called <Highendman> who defeated the Rybka forum and he often aims for paradoxical positions that are difficult for the engine to evaluate... White: Highendman
Black: Rybkaforum
1. e4 c5 2. b4 cxb4 3. Nf3 Nf6 4. e5 Nd5 5. a3 e6 6. axb4 Bxb4 7. Na3 Nc6 8. c4 Nde7 9. Nc2 O-O 10. Nxb4 Nxb4 11. Ba3 Nec6 12. Qb1 a5 13. Bd3 f5 14. Be2 d5 15. exd6 Qxd6 16. d4 Bd7 17. Qb2 Be8 18. O-O Bh5 19. Rfd1 Rae8 20. Rd2 b6 21. Rad1 Re7 22. h3 h6 23. Qb3 Rc8 24. Bb2 Nd8 25. g4 fxg4 26. Ne5 Be8 27. d5 exd5 28. Bxg4 Ndc6 29. Bxc8 Nxe5 30. Bxe5 Rxe5 31. cxd5 Bf7 32. Be6 Bxe6 33. dxe6 Qxe6 34. Rd8+ Kh7 35. Qxe6 Rxe6 36. R8d6 Re4 37. Rxb6 a4 38. Ra1 Nd5 39. Rb7 h5 40. Kg2 Rd4 41. Ra7 Nc3 42. Rc1 Nd5 43. Rc6 Nf4+ 44. Kg3 a3 45. Rcc7 Ne6 46. Re7 Rd6 47. Rxa3 Kh6 48. Rea7 Nc5 49. f3 Rg6+ 50. Kf2 Rd6 51. h4 Nd3+ 52. Kg2 Rg6+ 53. Kh2 Nc5 54. Rc3 Ne6 55. Ra5 Nf4 56. Rf5 Ne2 57. Rcc5 Ra6 58. Rxh5+ Kg6 59. Rhg5+ Kh6 60. Rc8 g6 61. Rgc5 Ra4 62. Rc4 Ra3 63. R4c7 g5 64. Rc6+ Kg7 65. R8c7+ Kf8 66. Rh6 Kg8 67. Re6 Kf8 68. Rxe2 g4 69. fxg4 Ra4 70. h5 Ra8 71. h6 Kg8 72. Ree7 Kf8 73. Rg7 Ra2+ 74. Kg3 Ra3+ 75. Kh4 Re3 76. h7 Rh3+ 77. Kxh3 Ke8 78. h8=Q# 1-0 |
|
| Jan-09-10 | | achieve: Allow me a tweet, inspired by this high-end discussion. Maybe it's interesting to draw attention to the choice of <highendman> to play the Wing Gambit here, and the fascinating apparent contra indication to open the game up like this, when facing massive B-Force from Centaur opposition.. However, since HEM was of course *also* allowed to use his maximum HW/SW recources there is this fascinating challenge to establish a lasting positional advantage, mainly due to swift development and (thus) an opportunity for better piece placement- and coordination. If the Gambit is accepted. In a Human vs Engine match with standard TCs, this is of course much more difficult, and a different strategy is likely more fruitful. I chipped in on this discussion because of my CC experience with the Smith-Morra, and a subsequent study I did following a Wing Gambit with <Hitman84>, where I had very strong play as white. Later I played a number of test games with the help of my engine vs The Engine, where it became clear to me that the evaluations were mostly useless, as, - like Deffi said - the long term advantages of the dominating and dynamic position only dawned on the "opposing" engine "when it was too late." How can it be otherwise.
So - very interesting that in Human vs Engine games, there are several approaches that may lead to disaster/success, emanating from <both> ends of the "closed--Open" spectrum, depending on the conditions, amount of (Opening-)preparation, and time. Discussing this and experimenting with it is most illuminating from various angles, actually. Hence here are a few links that may be of interest - in case anyone wants to look at the history of GM-Engine Chess, and eg Bronstein's amazing head-to-"head": Bronstein vs Chessmaster, 1995
<Bronstein vs Computers.>
Compiled by lostemperor
<Creator vs Creation? Creativity vs Calculator? Sorcerer vs Apprentice. The old man, 4 decades after his title match, vs the best machines. 39 games; 36 wins!> Game Collection: Bronstein vs Computers. Game Collection: Computer - GM games 1963-2002 Maybe I wasn't entirely clear in getting my points acroos due to time limitation, but having looked up some material and links I thought: PRESS POST KIBITZ button, et voila. |
|
| Jan-09-10 | | Red October: < Domdaniel: So I'll just *twitter* while eating it.
Or try dignified silence.> only <twinlarks> are allowed to twitter...
moi aussie... |
|
Jan-09-10
 | | Domdaniel: All very intriguing. I should point out that my *dignified silence* is a strategy, not a tactic, and my engine doesn't like it. I'll be giving evals to Risotto next ... |
|
| Jan-10-10 | | achieve: Ahh - apologies... Though I couldn't really foresee an escalation like this-- maybe I should have. |
|
Jan-10-10
 | | Domdaniel: A bit of old traditional music ...
<From the Erse “Ba mhaith liom bheith i Mhá mhá mhá”, anon> I wish I was in Wawawa
On the world wide web with you
Where cypher text meets hypertext
And no-one has a clue.
I’d lay you down on fields of cyan
And hold on to your handle
Like something out of Life of Brian
To which none can hold a candle.
“Aitch tee em ell” we’d cry as one
“Slash slash (the forward kind)”
The URL of Moray and his son
Ne’er saw such data mined.
I wish I were in Wawawa
Subjunctive verbs are tricky.
But there’s a site - it’s quite a sight
Beyond the reach of Wiki.
The dark web falls. Night closes in.
The year’s midnight is passing
Yet pixels whisper in my ear
And evil trolls are massing.
I wish we were in Wawawa
It’s far from Lonely Street
The desk clerks wear chrysanthemums
And wings, on slippered feet … |
|
Jan-10-10
 | | Domdaniel: <Ohio> -- < Very interesting, even if it was about chess.> Thanks. It was a bit *chessy*, wasn't it? I admit I get carried away sometimes: I go so wildly off-topic that I wind up back where I'm supposed to be. Chess?? Whatever next? |
|
Jan-10-10
 | | Domdaniel: <crawfb5> Thou art a soul in bliss, then? Must be nice. |
|
| Jan-10-10 | | nimh: I looked briefly at Petrosian's games I analyzed, but I didn't see any statistical anomalies as to unusually high amount of negative evals. Perhaps we looked at different games. But I think it wouldn't be very surprising because in order to win in chess one must have a significant advantage, a slight one won't do.
Petrosian may have intentionally lured his opponents into positions where defensive play is easier. Petrosian's brick-wall-solid defense with waiting play surely frustrated his opponents - especially with white - and they banged their heads against the Wall out of desperation. |
|
| Jan-10-10 | | nimh: In late eighties comparative computer analyses surely weren't as obvious as today. I wouldn't have been doing what I am now if B&G hadn't published their famous study. Where did you get the idea of analyzing games? How did the others take it? |
|
Jan-10-10
 | | Domdaniel: <nimh> It was just something I did for amusement: out of interest, looking at the performance of the engine as much as the players. I don't recall even showing it to anyone else, though it was something I did sporadically for a few months. As I think I wrote in my profile, I tend to prefer playing *with* chess to playing chess. btw, what do you think of the survey on chess aesthetics in which CG members participated? There's a link to its preliminary results on the homepage. Unless these are *very* preliminary, it looks to me like the most pointless kind of data mining. There are too many hidden variables, such as the difference between real games and compositions -- some players dislike compositions because of their unreality, however 'pretty' the solution. And how can anyone place a numerical value on a standard smothered mate: it's familiar, it can still excite wonder in people seeing it for the first time, but I don't think it can be compared to a 21st century combination. I just don't see how the numbers they extract tell us anything at all. |
|
| Jan-10-10 | | crawfb5: <Dom> You can download Schiller's Caxton classification system in PDF format at: http://www.chess.com/download/view/... and also see:
http://www.ericschiller.com/resourc... |
|
| Jan-10-10 | | achieve: <[...]but I don't think it can be compared to a 21st century combination.> I don't think there exists such a "thing" as a 21st century combination, yet, but I understand your point, though surely in the last 40 years or so there wouldn't have been a significant change, evolution or "shift" re the type of combinations? hmmm... I do agree fully though on the pointlessness, many times, of data mining. Seems I run in to a screwy one more often by the week. |
|
Jan-10-10
 | | Domdaniel: <Niels> Maybe -- and you're infinitely better able to judge this than me -- it's a little bit like music. The music of Bach and Vivaldi is very beautiful, and original for the time it was composed. But we've heard it so often, plus many variations and riffs and even jazz/rock versions, that it's now very familiar. One aspect of the beauty, its novelty to the ear, inevitably weakens with time. And we're possibly more excited by new music: not because it's better, but because we haven't heard anything like it. Doesn't really hold up for chess, does it? Hmmm ... |
|
 |
 |
|
< Earlier Kibitzing · PAGE 551 OF 963 ·
Later Kibitzing> |
|
|
|