x
Chess - Play & Learn

Chess.com

FREE - In Google Play

FREE - in Win Phone Store

VIEW

Stockfish is blind!

  • #241

    Correction, maybe 14...e4 15.fxe4 Ng4 and follow by Bd4+

  • #242

    Agreed with yureesystem

  • #243
    Lyudmil_Tsvetkov wrote:
    bvila19 wrote:
    Lyudmil_Tsvetkov wrote:
    chesster3145 wrote:
    Lyudmil_Tsvetkov wrote:

    Well, you are laughing at my evaluation, but it is precisely that allowing me to win games against SF.

    The twice defended storming pawn on f5 above, worth some 50cps or so, coupled with the d3-e4-f5 pointed chain are factors that easily allow me to quickly assess the position as won for white long before SF will see it.

    That is what refined evaluation means.

    You can see this without convoluted terms: You will win a pawn and Black’s King is short of defenders. A couple of moves earlier, you were the only one that could attack on the Kingside. Also, why 1-minute? Engines are known to be idiots in bullet.

    Because I don't have time.

    Then why SF does not see it?

     

    Please stop talking until you play a game of chess against a human opponent.

     

    I have done so and I have won a range of GMs over the board.

    Really? Any proof? I can't find your name in the fide database? Which GMs did you beat?

     
  • #244

    Stop lying to everyone. You are fake.

     

    Fabricted engine games leading to fabricated books.

    No games played on chess.com.

    No name in the FIDE database.

    "The French is lost for lost for black." Please, I'd bet $1,000 that you would lose to any titled player playing the black side of the French.

    You support engines, and then you don't. Make a decision, please.

    You say you're so mighty and powerful, but you have absolutely nothing to back it up.

     

    All you have is your chessprogamming profile, which is nothing impressive either.

     

    Please, do mine and everybody else's brain cells a favor and stop putting out this garbage,

     
  • #245
    greekgift_221b wrote:

    Engine chess is the strongest, but claiming that engines score badly against other engines in French with black pieces, just kills the beauty of chess IMO.

    On the contrary, the more you know, the more interesting it becomes.

    The deeper the level of chess, the more beautiful it is. That is why, I guess I am fully right in declaring: immediately after my games with the top engines, the Alpha-SF match has produced the most interesting and beautiful chess. happy.png

  • #246

    Well ok then

  • #247
    MARattigan wrote:
    Lyudmil_Tsvetkov wrote:
     

    MARattigan wrote:
    sisu wrote:

    ...

    The example I like to give to my students (whenever they think the engine is god) is Gusev-Averbakh, USSR Championship 1951. Stockfish does not see the Queen sacrifice in this game. And even when you show it, it still thinks the Queen sacrifice is wrong. Only 10 moves or so later does it have a "heart attack" jejejeje

    ... so it is nice to see this new AlphaZero winning the match with Stockfish.

    ...

     

    Very impressive finish to that game, but you don't have to look at anything too esoteric to find positions Stockfish doesn't understand. If it hasn't got an EGTB to consult it doesn't play the White pieces accurately in the majority of KBNK positions.

     

    But there isn't a single Stockfish. The Stockfish I recently downloaded with Tarrasch plays this endgame more accurately than the lichess level 8 version (though still not accurately). I assume this is because the latter's access to computer resources is more limited.

     

    This also makes a nonsense of the much hyped recent match between Stockfish and AlphaZero. My downloaded version would probably score on average two moves better per KBNK position than the lichess level 8 version. AlphaZero apparently had access to a whole supercomputer while Stockfish was running on a PC with limited resources.

    My impression is SF delivers mate with knight and bishop very easily.

    It has problems in bullet TC with this one, though:

     

    How many people on this forum are able to win this with white?

    My point was that Stockfish doesn't understand some relatively simple positions such as KBNK (simpler I would say than KQKBB). While either of the versions of Stockfish I mention can deliver mate, neither plays accurate moves. I tried only one such position against lichess level 8, both as White and Black and mated 6 moves faster than I was mated. I have played only a few such positions against my downloaded Stockfish but these would suggest I should expect at most a two move advantage per pair of games. The fact that Stockfish plays the White pieces inaccurately suggests that it doesn't understand such positions. (Accurate play by Black is not the best defence in this endgame.) 

     

    My other point is that the difference between the Stockfishes is presumably entirely down to availability of computer resources. The difference would suggest that my downloaded version would trounce lichess level 8 if it were to play a game against it. But the difference in allocated computer resources between AlphaZero and the Stockfish that took part in the match between them appears to have been many times greater so cannot be taken as an indicator that AlphaZero is stronger than Stockfish in general. 

    SF level 8 on Lichess uses maybe 50ms on move, so very weak.

    You can not compare it with SF running on general purpose hardware having at least 1 second per move.

    Alpha-SF was not a fair match, of course. SF is at least 400 elos stronger on equal hardware.

  • #248
    yureesystem wrote:
    Lyudmil_Tsvetkov wrote:
    greekgift_221b wrote:

     How does White win?

    Easily. Only player that is allowed to play the French as black is Botvinnik.

    It really makes sense the French is a bad opening. Why should black push a central pawn only to the 3rd rank instead of the 4th?

    Are not we taught by chess masters space is very important in chess? Why cede that valuable central space and then believe we are ok?

     

     

     

     

    Ask any uscf expert if the French is losing and they will laugh at you. I know a very strong expert rated 2160 uscf and beat experts and masters with French defense. The problem with players like you is you don't have a strong chess foundation meaning acquire sound chess knowledge and with it players like you will always make dubious statement about chess. If player dedicate to a narrow opening repertoire for black, play the French defense against 1.e4 and Queen Gambit Declined against 1.d4, with this very narrow repertoire a player can become a master. Having a narrow repertoire allow the player to specialize in their opening. The reason I mention this players who specialize in the French were or are {  Tarrasch, Winawer, Lasker, Rubinstein, Capablanca, Reti and others is from older generation } and Uhlmann ( he played only the Frense defense), Botvinnik, Petrosian, Korchnoi { he is testimony why the French is sound and playable, Karpov who is positional genius could not score a win in their match}, IM John Watson, Yusupov, Eingorn, Kamsky, Grischuk, Gurevich and Morozevich and others. Morozevich is perfected the French to make it a deadly weapon and beating many top players; I add Morozevich is incredible creative and he made the Chigorin defense playable ( 1.d4 d5  2.c4 Nc6!) and deadly, almost every GMs was saying that Chigorin defense was not playable and here comes Morozevich and proves them wrong!  

     

     Pawn structures tell a partial story, if we look at King's Indian defense, White has space, active pieces and the center and looks like black will get crush but black has the unbalance nature and dynamic potential, black needs to play active and everything needs to be time accurate and black NEEDS to be flexible ( No black does not attack in their king side always, look at Fischer and Kasparov games, they were flexible); black has very good chances to succeed and especially below IM level. 

     

     This silly age of engines and players believing them to be gospel truth and proclaiming Stockfish said, "The King's Indian is losing and don't play it." Or Stockfish evaluate the French defense dead lost and Super GMs play the French and proving the Stockfish wrong! It comes to this do you have acquire chess knowledge so you can make a sound judgment, otherwise you make ridiculous statement about chess, how this is winning and that is losing.  

     

     GM Peter Svidler mention chess book study is a must and that is soundness way to acquire sound chess knowledge. If you don't know who is Morphy,Lasker Nimzovich, Capablaca, Alekhine,Rubinstein, Tarrasch, Steinitz and Chigorin,and Botvinnik each of these great GMs have contribute a lot to chess and it will give you a strong chess foundation. Every strong player 1900 elo and higher should study Chess Praxis by Nimzovich, without it you have a long ways to go to become a master but this book will shorten your time. That is good chess book do for you accelerate your learning so you don't have to figure it out otb and maybe avoid painful loses.

     

     

     

    Tarrasch, Rubinstein, Uhlmann, Watson: all you are mentioning to me are weak players.

    Most relevant currently is the score of the 100 game match in the French between Alpha and SF: +39 -3 =58 for white. Tremendous! Why do I need to refer to somebody else?

    Acquiring knowledge makes sense, but immensely more important is coming up with ORIGINAL NEW knowledge.

    The first one is just the work of an apprentice, the second the work of a master and an author. There is a difference, is not there?

    All the great books, songs, theorems and paintings have been created taking an original path, a split from the past and the current statusquo, and you want me to be an apprentice...

    I guess Nimzovich will have a lot more to learn from 'The Secret of Chess' than me from 'Chess Praxis'.  happy.png

    Seriously, I have read the most of 'My System' and it is extremely easy, all the games and all the concepts, and frequently the assessments are inaccurate, so what precisely do you want me to learn from there?

    You really believe a book 100 years old can excellently teach still now?

    Have you ever read Asimov's 'Foundation'? I would not like that the Earth becomes Trantor, although it has headed precisely that way...

  • #249

    Rubinstein? Weak? I am done speaking with you.

  • #250
    DMAlphaZero wrote:

    If your good enough to beat Stock fish then play ALPHAZERO.

    You wont have a chance

    Why don't you play a person rated 1600

    Better than you. 

    Alpha is not a very big challenge, even if they double the hardware.

    I am able to score over 60% against SF, so I will score at least 55% against Alpha.

    Still a lot to learn until they reach my level.

  • #251
    Lyudmil_Tsvetkov wrote:
    DMAlphaZero wrote:

    If your good enough to beat Stock fish then play ALPHAZERO.

    You wont have a chance

    Why don't you play a person rated 1600

    Better than you. 

    Alpha is not a very big challenge, even if they double the hardware.

    I am able to score over 60% against SF, so I will score at least 55% against Alpha.

    Still a lot to learn until they reach my level.

  • #252
    CheesyPuns wrote:
     

    Hey lyudmil,

    i was originally kind of skeptical about your book and ideas, however after reading Mr. Smerdon's review, it seemed alot more reasonable, if you could spare the time, could you walk us throuhg our thought process in the above position? i had it in a game a few months ago and stockfish was indeed blind while analyzing it afterwards...

    Hello, CheesyPuns, thanks for your position.

    Well, this is one position where I am not very strong, as there are not many salient evaluation patterns to recognise, and more will depend on calculation.

    I see the following evaluation factors:

    - the black pawn shelter is almost fully missing, bad

    - the e5 pawn is weak

    - black has very nice f4 advanced pawn, making the g2 pawn semi-backward

    - b2 is backward, but a4 is also weak

    - the particular placement of the pawns in the center, coupled with the big probability black will have to change light-square bishops, could leave white with good vs bad dark-square bishops, especially important in that respect is the control over the e4 central square

    - d5-d6 lever break is an additional opportunity to consider(in 'The Secret of Chess', there is a particular term that would give bonus for unopposed pawn on the 5th rank that is not a passer, just like d5 is, as it can break at any moment with d5-d6, it is called unopposed backward-maker)

    - heavy pieces make the game more drawish

    - pair of bishops each side make the game more drawish

     

    So that, white has clearly more relative advantages, but black has some good terms to rely upon too.

    How would the game end? Difficult to claim with full certainty, this is a position one may analyse for days until a definitive conclusion is reached. This is a borderline position, precisely on the border between win and draw, and such positions are very hard to tell without doing a meticulous analysis.

    I am sorry that I am unable to do full analysis now, and myself am not 100% certain what this is, although I checked it a bit with SF, and it tends to win this at bigger depths in maybe 80% of cases, so I can suppose white wins after all.

    I don't believe there is a single best move for white here, as the position bears a more manoeuvering character, but, as said, this requires further analysis.

    All I can say is this: a borderline position between win an draw.

  • #253
    bvila19 wrote:
    Lyudmil_Tsvetkov wrote:
    bvila19 wrote:
    Lyudmil_Tsvetkov wrote:
    chesster3145 wrote:
    Lyudmil_Tsvetkov wrote:

    Well, you are laughing at my evaluation, but it is precisely that allowing me to win games against SF.

    The twice defended storming pawn on f5 above, worth some 50cps or so, coupled with the d3-e4-f5 pointed chain are factors that easily allow me to quickly assess the position as won for white long before SF will see it.

    That is what refined evaluation means.

    You can see this without convoluted terms: You will win a pawn and Black’s King is short of defenders. A couple of moves earlier, you were the only one that could attack on the Kingside. Also, why 1-minute? Engines are known to be idiots in bullet.

    Because I don't have time.

    Then why SF does not see it?

     

    Please stop talking until you play a game of chess against a human opponent.

     

    I have done so and I have won a range of GMs over the board.

    Really? Any proof? I can't find your name in the fide database? Which GMs did you beat?

     


    The GMs I have beaten was for my Bulgarian rating, which is higher. I have played very few FIDE rated games. I already mentioned some names of GMs I have beaten, don't intend to constantly repeat one and the same info, go find the relevant thread yourself. I myself can't quite find a ranking table from my last tournament for Bulgarian rating, but there I scored 5/6, and won - 2 players over 2300 - 1 player over 2200 - 2 draws with a player over 2400 and a bit weaker one Really, I can't find a table to link.
  • #254
    bvila19 wrote:

    Stop lying to everyone. You are fake.

     

    Fabricted engine games leading to fabricated books.

    No games played on chess.com.

    No name in the FIDE database.

    "The French is lost for lost for black." Please, I'd bet $1,000 that you would lose to any titled player playing the black side of the French.

    You support engines, and then you don't. Make a decision, please.

    You say you're so mighty and powerful, but you have absolutely nothing to back it up.

     

    All you have is your chessprogamming profile, which is nothing impressive either.

     

    Please, do mine and everybody else's brain cells a favor and stop putting out this garbage,

     

    The garbage is all yours.

    I have FIDE rating, and the link was already posted here.

    I have beaten GMs, of course, you can ask one of them.

    Much more info has been posted on me here, it is not my fault you can not read.

    So, basically, it is you who is lying, not me.

    If a GM says the book is worth it, who are you to claim otherwise? : https://www.chess.com/blog/smurfo/the-secret-of-chess

  • #255
     

    yureesystem wrote:

    Correction, maybe 14...e4 15.fxe4 Ng4 and follow by Bd4+

     

     
  • #256

    Please go to Mr. Smerdon's blog to read my last post to know more about the concepts of 'The Secret of Chess'.

    And here the first editorial review for 'Human versus Machine', short one, but to the point: https://www.expert-chess-strategies.com/human-versus-machine.html

Top

Online Now