Oh my god! thanks to this bizarre post i´m finding out natalia pogonina is a chess.com member. I´m commenting the same post she commented... Guys: I think i´m in love 
Can Houdini 3 be beaten?
Sort:
manfredmann wrote:
sloughterchess wrote:
Irontiger wrote:
We are all so fortunate that a 1400-ish player gave us the universal recipe to beat computers. Shame Kasparov didn't know back in 1997.
GM Roman Dzindzichashvili knows I am not a 1700 player, gives my games serious analysis, agrees that my Houdini 3 is a strong program, and accepts the fact that I can outplay Houdini 3.
You are 1728. You've been rated as high as 1799 in the past 11 years. Roman is a pro - of course he takes a student seriously. But I'm pretty sure that he does NOT believe that you can outplay Houdini since no GM seems to be able to do so. You are rated 1000+ points below all these guys who can't beat it. You cannot beat it, unless you are dumbibg it down in some way.
You are using Houdini 3. What platform? x32, x64, how many CPUs/cores, clock rates? how much memory? I'm playing over your Two Knights game with a recent, commercially available Houdini, which gives ...Bb7 as its seventh (!) option at d20. At greater search deepth ...Bb7 looks even worse.
When my game is "off" I am terrible player; when my game is "on" I have already drawn GM Lev Alburt in a Blumenfeld Counter Gambit where he is a leading expert in the only game we ever played so my record against Grandmasters at blitz is 1/2-1/2. As I have indicated many times I have beaten a stronger computer in a game called Magic where I gave the computer 9 moves in a row at the start of the game (see below) and an extra hour on the clock for each of its moves. I took only a few seconds/move relying entirely on intuition. If you think it is easy to do this try it. I have also had one of my games against a Postal Expert published in its entirety in Inside Chess a magazine devoted to chess professionals. Here is the starting position of Magic:
I am using Houdini 3 Pro running on an i7 CPU socket 1155, dual core, I believe, 4 GM Memory 500 GB hard drive. GM Dzindzichashvili says this is a "good" setup.
For some reason the computer is programmed to play inferior openings from the factory setting. I am trying to get this fixed and have contacted the manufacturer through Diego, service representative of Chess King. While 8...Bb
7 is not the best option, it is a rock solid choice that has been analyzed by the opening specialists at Chess King (his wife is the Woman's World Champion). Black plays for an immediate c5/Nc6 immediately bringing the Knight back into play threatening Nd4; I have to play c3 which weakens the d3 square.
This game was played at FIDE level i.e. 2 hours for the first 40 moves; since the first 11 are book, it is actually 2 hours for 30 moves. When we reach the time control the computer has lost on time twice (software glitch), but in the final position when I had the Pro play out both sides of the position at blitz speed, White wins.
manfredmann wrote:
sloughterchess wrote:
Irontiger wrote:
We are all so fortunate that a 1400-ish player gave us the universal recipe to beat computers. Shame Kasparov didn't know back in 1997.
GM Roman Dzindzichashvili knows I am not a 1700 player, gives my games serious analysis, agrees that my Houdini 3 is a strong program, and accepts the fact that I can outplay Houdini 3.
You are USCF 1728. You've been rated as high as 1799 in the past 11 years. Roman is a pro - of course he takes a student seriously. But I'm pretty sure that he does NOT believe that you can outplay Houdini since no GM seems to be able to do so. You are rated 1000+ points below all these guys who can't beat it. You cannot beat it, unless you are dumbibg it down in some way.
You are using Houdini 3. What platform? x32, x64, how many CPUs/cores, clock rates? how much memory? I'm playing over your Two Knights game with a recent, commercially available Houdini, which gives ...Bb7 as its seventh (!) option at d20. At greater search deepth ...Bb7 looks even worse, giving White a substantial advantage from the start. Why did you pick this lousy move for Houdini to make? You should play all of Houdini's first choices - let's call them H1's (Fritz's first choices would be F1's, Stockfish second choice S2, etc). Don't muck around with Houdini's choices! Houdini's book (or what it calculates - either way) would be 7...Be7 H1, 7...Rb8 H2 or 7...h6, a human recommendation from GM Nigel Davies' book 'Play 1.e4 e5!' Of course, these would be at a reasonable depth of search. I chose d20 because I could keep up with it at that rate documenting the choices made in your game.
Once you got out of book (I take this to be the point at which no more database games are available, not some arbitrary number like 8 or 10 moves, which would leave both sides continuing to play H1 moves that weren't necessarily their own independent choices - assuming, as we know, that the titled players in the database used H, F, S and other engines in their extensive opening prep)... then you and Houdini both made a long series of H1 moves. Of course, ALL of Houdini's moves should be H1's. On the other hand, as a 17xx player only some fraction of yours should be H1's. Your fraction was quite high, especially after you were out of book. Then towards the end of the game you suddenly starting playing 17xx moves. You voluntarily created weaknesses, withdrew your pieces and played almost a hedgehog, but a very passive one. Houdini maneuvered around and calculated where to stick the knife. No surprise, the 17xx player loses despite making many H1 and not always making H1 for the engine. How does this substantiate your claim that "Roman accepts the FACT (my emphasis) that I can outplay Houdini 3"? You are seriously delusional.
So, I'm not sure what you think this game proves? You played a helluva lot of H1 moves which kept it close, then collapsed. You had a slight advantage at first because of the crummy 7...Bb7? which is not close to an H1, but even then you were never close to winning.
In game 2 you're asking Houdini to defend the Fried Liver Attack, so you forced the choice of 5...Nxd5, not Houdini. That move was not an H1.Then it is a matter of long theory. I'm sure that White played alot of H1s again and I doubt that Houdini played 100% H1s.
You can't beat Houdini. Stop saying you can. If Roman comes out publicly and says a 17xx player can beat Houdini, he'll the chess world will take a very close look at his explanation. Meanwhile, please don't represent what he says. Cheers.
The Houdini 3 Pro routinely chooses the Fried Liver; as I indicated dozens of post ago, I have long since stopped Houdini 3 or the Pro from playing opens I choose; it only plays openings it chooses.
sloughterchess wrote:
(...)While 8...Bb7 is not the best option, it is a rock solid choice that has been analyzed by the opening specialists at Chess King (his wife is the Woman's World Champion). (...)
manfredmann wrote:
I'm capable of answering OP myself, thanks Irontiger. Meanwhile just crawl back into your cage and draw some more cartoons.
uri65 wrote:
EscherehcsE wrote:
sloughterchess wrote:
EscherehcsE wrote:
sloughterchess wrote:
aatkins wrote:
I only had to go through the first 20 moves or so to know you weren't playing a full strength Houdini, or even Houdini at all. A few of the moves played weren't even in Houdini's top 5... for example 10. e4?? is a blunder according to Houdini.
You are making my point! 10...e4 is a concept of World Champion Dr. Hans Berliner who studied the opening for decades. Even an early version of BCO 2 edited by Garry Kasparov devoted a page of analysis to the Berliner Gambit. My point is that the Two Knights' Defense is so poorly analyzed that a complete reworking of the theory is necessary. Meanwhile, I'll just provide wins over Houdini that theory regards as perfectly acceptable for Black!
Yeah, I think we do understand. You're substituting moves for Houdini that Houdini would never play, and then you're claiming that you beat Houdini. Uh, OK...
The point is that several of the generally accepted lines of the Two Knights' Defense are so bad that even Houdini can't salvage them (Just check out TKD in MCO 15 and you will find several). It is dangerous to ignore all theory because then positional "blunders" too subtle to exploit quickly will begin to creep in.
At what point in the first game does your cherry picking of your opening theory end and Houdini is allowed to play for itself? Some of the more blatant poor moves that Houdini would never play are 9, 10, 12, 24, 27, 32, and 39. So your opening theory goes to move 39? Impressive! (I seriously think you should check your Houdini configuration settings to see if it's really playing at full strength.)
Probably this thread should be named "Can sloughterchess-assisted-and-manipulated Houdini 3 be beaten?"
By the way I beat Houdini few times a day while training theoretically won endgames against it
that was tough
RITWIK_11 wrote:
uri65 wrote:
EscherehcsE wrote:
sloughterchess wrote:
EscherehcsE wrote:
sloughterchess wrote:
aatkins wrote:
I only had to go through the first 20 moves or so to know you weren't playing a full strength Houdini, or even Houdini at all. A few of the moves played weren't even in Houdini's top 5... for example 10. e4?? is a blunder according to Houdini.
You are making my point! 10...e4 is a concept of World Champion Dr. Hans Berliner who studied the opening for decades. Even an early version of BCO 2 edited by Garry Kasparov devoted a page of analysis to the Berliner Gambit. My point is that the Two Knights' Defense is so poorly analyzed that a complete reworking of the theory is necessary. Meanwhile, I'll just provide wins over Houdini that theory regards as perfectly acceptable for Black!
Yeah, I think we do understand. You're substituting moves for Houdini that Houdini would never play, and then you're claiming that you beat Houdini. Uh, OK...
The point is that several of the generally accepted lines of the Two Knights' Defense are so bad that even Houdini can't salvage them (Just check out TKD in MCO 15 and you will find several). It is dangerous to ignore all theory because then positional "blunders" too subtle to exploit quickly will begin to creep in.
At what point in the first game does your cherry picking of your opening theory end and Houdini is allowed to play for itself? Some of the more blatant poor moves that Houdini would never play are 9, 10, 12, 24, 27, 32, and 39. So your opening theory goes to move 39? Impressive! (I seriously think you should check your Houdini configuration settings to see if it's really playing at full strength.)
Probably this thread should be named "Can sloughterchess-assisted-and-manipulated Houdini 3 be beaten?"
By the way I beat Houdini few times a day while training theoretically won endgames against it
that was tough
Let me recap: First of all my laptop was running on a celeron processor. This POS was worthless to play chess, hence my preliminary wins were not best play.
Then I got Houdini 3 on my "old" computer; again spurious results.
Then I was told that having the computer play the Black side of known openings e.g. the Berliner Gambit or Wilkes Barre proved nothing because the computer would never choose those openings.
Then I was told that my new Houdini three was giving spurious results so I had a computer expert design a computer with the express purpose of getting the most out of the Houdini program.
Then I bought Houdini 3 Pro and promptly won 3 games in two days because the Pro plays inferior openings. I have a couple of games since then where Houdini played poor, but not necessarily losing openings.
Now I'm just waiting to see what GM Roman Dzindzichashvili and Senior Live Master Russ Potter have to say about my play.
sloughterchess wrote:Then I was told that having the computer play the Black side of known openings e.g. the Berliner Gambit or Wilkes Barre proved nothing because the computer would never choose those openings. Then I was told that my new Houdini three was giving spurious results so I had a computer expert design a computer with the express purpose of getting the most out of the Houdini program.
Then I bought Houdini 3 Pro and promptly won 3 games in two days because the Pro plays inferior openings. I have a couple of games since then where Houdini played poor, but not necessarily losing openings.
Oh boy... your ignorance is really becoming annoying.
Houdini does NOT play any openings by itself. Neither Stockfish does. The only top engine that is bundled with a (very slim) opening book is Critter, but in reality nobody uses this book.
It's the GUI that you use who loads an opening book for engine usage. I hope you will realize this within the next couple of years, instead of making silly claims.
7 months, and more that 360 posts focused on the hole in the doughnut?
Snap Out Of It. PLEASE. 
pfren wrote:
sloughterchess wrote:Then I was told that having the computer play the Black side of known openings e.g. the Berliner Gambit or Wilkes Barre proved nothing because the computer would never choose those openings. Then I was told that my new Houdini three was giving spurious results so I had a computer expert design a computer with the express purpose of getting the most out of the Houdini program.
Then I bought Houdini 3 Pro and promptly won 3 games in two days because the Pro plays inferior openings. I have a couple of games since then where Houdini played poor, but not necessarily losing openings.
Oh boy... your ignorance is really becoming annoying.
Houdini does NOT play any openings by itself. Neither Stockfish does. The only top engine that is bundled with a (very slim) opening book is Critter, but in reality nobody uses this book.
It's the GUI that you use who loads an opening book for engine usage. I hope you will realize this within the next couple of years, instead of making silly claims.
Because the Pro plays inferior openings, I will simply set up known main lines that are regarded as either = or with compensation.
Here in a known variation of the Ruy Lopez, my 9...Re8 appears to be very rare. It is not in ECO Edition 5, MCO 15, or in my Big Data Base 2013. Houdini was able to rely on the Spanish torture to force complications favoring White.
1. e4 e5 2. Nf3 Nc6 3. Bb5 a6 4. Ba4 Nf6 5. O-O b5 6. Bb3 Bc5 7. c3 O-O 8. d4 Bb6 9. Re1 Re8 10. Bg5 h6 11. Bh4 d6 12. a4 Bb7 13. axb5 axb5 14. Rxa8 Bxa8 15.d5 Na7 (This is a developing move; Nb8 is an undeveloping move. It is incidental that I am defending b5) 16. Qd2 Qe7 17. Na3 c6 18. h3 Bd8 19. Bxf6! Qxf6 20. Rd1 Bc7 21. Qd3 Rb8 22. dxc6 Bxc6 23. Nc2 Nc8 24. Nb4 Bb7 (The Pro doesn't like this move)
25. Qxb5 Bxe4 26. Qe8+ Kh7 27. Bd5 Rxb4 28. cxb4 Bxd5 29. Rxd5 Ne7 30. Rd1 d5 31. Nh2 Qe6 32. Nf1 d4 33. b5 Bb6 34. Ra1
*
25. Qxb5 Bxe4 26. Qe8+ Kh7 27. Bd5 Rxb4 28. cxb4 Bxd5 29. Rxd5 Ne7 30. Rd1 d5 31. Nh2 Qe6 32. Nf1 d4 33. b5 Bb6 34. Ra1
*
your barely 1400, sloughterchess
Red4444 wrote:
sloughterchess Stupid arrogant crazy psychotic brainless. No one can beat a 3200elo+ engine nowadays and all the chess'world knows it except you, liar mediocre player opening faker. Go, run kid and search someone to play against who actually can choose Houdini's first moves..you'll be dead in less than 20 moves, every game.
It will be an interesting exercise watching you eat your words.
charlesge wrote:
your barely 1400, sloughterchess
It is "you're", not "your" barely 1400, sloughterchess. If you are going to criticize me please you proper grammar.
sloughterchess wrote:
charlesge wrote:
your barely 1400, sloughterchess
It is "you're", not "your" barely 1400, sloughterchess. If you are going to criticize me please you proper grammar.
And here comes the ad grammarem attack.
Was it too hard just to say you don't care about your online rating, that the rating pool is riddled with cheaters, etc. ?
You haven't learnt the basics of trolling. Get back to school, if you ever want to get yours :
Irontiger wrote:
sloughterchess wrote:
charlesge wrote:
your barely 1400, sloughterchess
It is "you're", not "your" barely 1400, sloughterchess. If you are going to criticize me please you proper grammar.
And here comes the ad grammarem attack.
Was it too hard just to say you don't care about your online rating, that the rating pool is riddled with cheaters, etc. ?
You haven't learnt the basics of trolling. Get back to school, if you ever want to get yours :
I have indicated in many posts that I don't care about my on line rating; check multiple posts to see what my playing strength either in correspondence chess against Postal Experts, one Master and other examples of my play when I am playing well.
just show us a screenshot of the evaluation your houdini gave of the moves mentioned by uri65. so moves 9, 10, 12, 24, 27, 32, and 39 of the first game. my houdini 1.5 (free version) evaluates all those moves as bad for black and would play other moves every single time. burden of prove is on you. and thats the only way you can prove that you are playing against (a obvliously faulty) houdini and not just against yourself, making inferior moves.
If you are referring to the first game on this thread, then yes, your Houdini 1.5 is probably right. Forget about the first 200 posts; they were against an inferior version of the software i.e. my computer was not fully operational (as I acknowledged on many occasions).
Once I started playing Houdini 3 Pro rated at 3200, it became clear that something is wrong with the way programmers have allowed the computer to pick openings. Clearly it has been degraded; as I indicated from now on I will only allow Houdini 3 with Black to play openings book regards as = or with compensation. With White I will only allow it play openings that are either = or +/=.
Inspired by your multiply wins against Houdini, I gave it a try and scored a wonderful win against it, I followed your recipe so thank you first and foremost.
GM Viktor Korchnoi validated my result and congratulated me on my beautiful win.
Of course, its opening book might be a bit defective, but who cares. Wonderful win, you just proved that humans can win against computers.