Hiarcs - human play

[herO]

I played many times against Hiarcs and I have to admit that Hiarcs is playing more like a human than a computer on lower ELO settings. I wanted to find out how human-like will Hiarcs play on higher ELO settings. Im only an intermediate player (class B), so I cant win against stronger Hiarcs opponents at all. So playing against stronger Hiarcs opponent will not make any sence.

I made a test. I created two instances of Deep Hiarcs. Deep Hiarcs can use its opening book, no tablebases, pondering was set to off, UCI_LimitStrength was set to ON. One instance had UCI_Elo set to 2640, another one had UCI_ELO set to 2812.

Results: 2812 ELO Hiarcs: 10 wins, 9 draws, 1 loss

I analyze every single game, I can tell that Hiarcs not choose the strongest move after move 20 and more. It plays strong moves with nice iniciative, try to play strategic and positional. If you look how 2700+ ELO players play - their opening preparation is solid, plays "computer enhanced" moves, but after move 20 and more their moves are not the same as the chess engine recommendeds. So Hiarcs try to play like human grandmasters too!

Im very impressed how Hiarcs play even on higher ELO strength. It's really amazing! Im not able to say how precisely Hiarcs play as a human on much higher levels, but I still think its really good.

I post a game, when 2640 ELO Hiarcs won against 2812 ELO Hiarcs.

Code: Select all

[Event "Duelist"]
[Date "2015.02.21"]
[Round "13"]
[White "Deep Hiarcs 2640 ELO"]
[Black "Deep Hiarcs 2812 ELO"]
[Result "1-0"]
[ECO "B40"]
[Opening "Sicilian defence"]

1.e4 c5 2.Nf3 e6 3.b3 b6 4.Bb2 Bb7 5.Nc3 Nc6 6.d4 cxd4 7.Nxd4 Qb8 8.Nxc6 Bxc6 9.Bb5 Bxb5 10.Nxb5 a6 11.Nc3 Nf6 12.Qe2 Bd6 13.O-O-O Bf4+ 14.Kb1 Be5 15.g3 Bc7 16.Rhe1 d6 17.Ba3 O-O 18.e5 dxe5 19.Bxf8 Qxf8 20.Ne4 Nd5 21.c4 Ne7 22.Rd7 Rc8 23.Red1 h6 24.g4 a5 25.g5 Nf5 26.gxh6 Nxh6 27.h4 Bd8 28.Nd6 Rb8 29.h5 Be7 30.Qe4 a4 31.Rd3 axb3 32.axb3 Bf6 33.Rb7 Rd8 34.Rxb6 Be7 35.Qxe5 Nf5 36.c5 Bf6 37.Qe4 Ra8 38.b4 Nxd6 39.Rbxd6 Ra4 40.Qb7 Ra8 41.Kc1 Qe8 42.Kd2 Rb8 43.Qd7 Qf8 44.Qa4 Qe7 45.Ke2 Bh4 46.Rf3 Qg5 47.Qa7 Rf8 48.Rdd3 e5 49.Rde3 Qg4 50.Kf1 Qc4+ 51.Re2 Qxb4 52.Qa3 Qb5 53.h6 Rc8 54.Qa2 Qb7 55.Rh3 Bf6 56.h7+ Kh8 57.Qc4 Rc7 58.Ra2 Bg5 59.Ke2 Qc8 60.Rh5 Rxc5 61.Qxf7 e4 62.Ra1 Rc2+ 63.Kf1 Bh6 64.Qg6 Qc6 65.Qxc6 Rxc6 66.Ra4 Re6 67.Rh4 Kxh7 68.Raxe4 Rxe4 69.Rxe4 Kg6 70.Kg2 Kf5 71.Re7 g6 72.Rf7+ Kg4 73.f3+ Kh5 74.Rh7 g5 75.Kg3 Kg6 76.Rh8 Bg7 77.Rg8 Kf7 78.Rc8 Be5+ 79.Kg4 Bf4 80.Rc6 Ke7 81.Kf5 Kd7 82.Rg6 Bc1 83.Rxg5 Bb2 84.Rg2 Ba1 85.Ra2 Bc3 86.f4 Ke7 87.Kg6 Kd6 88.f5 Bb4 89.f6 Kc7 90.f7 Kc8 91.Ra8+ Kd7 92.f8=Q Be7 93.Ra7+ Kc6 94.Rxe7 Kc5 95.Kf5 Kc4 96.Ke4 Kb3 97.Kd4 Kc2 98.Re2+ Kb3 99.Qb8+ Ka3 1-0

[Fernando]

I played many times against Hiarcs and I have to admit that Hiarcs is playing more like a human than a computer on lower ELO settings. I wanted to find out how human-like will Hiarcs play on higher ELO settings. Im only an intermediate player (class B), so I cant win against stronger Hiarcs opponents at all. So playing against stronger Hiarcs opponent will not make any sence.

I made a test. I created two instances of Deep Hiarcs. Deep Hiarcs can use its opening book, no tablebases, pondering was set to off, UCI_LimitStrength was set to ON. One instance had UCI_Elo set to 2640, another one had UCI_ELO set to 2812.

Results: 2812 ELO Hiarcs: 10 wins, 9 draws, 1 loss

I analyze every single game, I can tell that Hiarcs not choose the strongest move after move 20 and more. It plays strong moves with nice iniciative, try to play strategic and positional. If you look how 2700+ ELO players play - their opening preparation is solid, plays "computer enhanced" moves, but after move 20 and more their moves are not the same as the chess engine recommendeds. So Hiarcs try to play like human grandmasters too!

Im very impressed how Hiarcs play even on higher ELO strength. It's really amazing! Im not able to say how precisely Hiarcs play as a human on much higher levels, but I still think its really good.

I post a game, when 2640 ELO Hiarcs won against 2812 ELO Hiarcs.

Code: Select all

[Event "Duelist"]
[Date "2015.02.21"]
[Round "13"]
[White "Deep Hiarcs 2640 ELO"]
[Black "Deep Hiarcs 2812 ELO"]
[Result "1-0"]
[ECO "B40"]
[Opening "Sicilian defence"]

1.e4 c5 2.Nf3 e6 3.b3 b6 4.Bb2 Bb7 5.Nc3 Nc6 6.d4 cxd4 7.Nxd4 Qb8 8.Nxc6 Bxc6 9.Bb5 Bxb5 10.Nxb5 a6 11.Nc3 Nf6 12.Qe2 Bd6 13.O-O-O Bf4+ 14.Kb1 Be5 15.g3 Bc7 16.Rhe1 d6 17.Ba3 O-O 18.e5 dxe5 19.Bxf8 Qxf8 20.Ne4 Nd5 21.c4 Ne7 22.Rd7 Rc8 23.Red1 h6 24.g4 a5 25.g5 Nf5 26.gxh6 Nxh6 27.h4 Bd8 28.Nd6 Rb8 29.h5 Be7 30.Qe4 a4 31.Rd3 axb3 32.axb3 Bf6 33.Rb7 Rd8 34.Rxb6 Be7 35.Qxe5 Nf5 36.c5 Bf6 37.Qe4 Ra8 38.b4 Nxd6 39.Rbxd6 Ra4 40.Qb7 Ra8 41.Kc1 Qe8 42.Kd2 Rb8 43.Qd7 Qf8 44.Qa4 Qe7 45.Ke2 Bh4 46.Rf3 Qg5 47.Qa7 Rf8 48.Rdd3 e5 49.Rde3 Qg4 50.Kf1 Qc4+ 51.Re2 Qxb4 52.Qa3 Qb5 53.h6 Rc8 54.Qa2 Qb7 55.Rh3 Bf6 56.h7+ Kh8 57.Qc4 Rc7 58.Ra2 Bg5 59.Ke2 Qc8 60.Rh5 Rxc5 61.Qxf7 e4 62.Ra1 Rc2+ 63.Kf1 Bh6 64.Qg6 Qc6 65.Qxc6 Rxc6 66.Ra4 Re6 67.Rh4 Kxh7 68.Raxe4 Rxe4 69.Rxe4 Kg6 70.Kg2 Kf5 71.Re7 g6 72.Rf7+ Kg4 73.f3+ Kh5 74.Rh7 g5 75.Kg3 Kg6 76.Rh8 Bg7 77.Rg8 Kf7 78.Rc8 Be5+ 79.Kg4 Bf4 80.Rc6 Ke7 81.Kf5 Kd7 82.Rg6 Bc1 83.Rxg5 Bb2 84.Rg2 Ba1 85.Ra2 Bc3 86.f4 Ke7 87.Kg6 Kd6 88.f5 Bb4 89.f6 Kc7 90.f7 Kc8 91.Ra8+ Kd7 92.f8=Q Be7 93.Ra7+ Kc6 94.Rxe7 Kc5 95.Kf5 Kc4 96.Ke4 Kb3 97.Kd4 Kc2 98.Re2+ Kb3 99.Qb8+ Ka3 1-0

This "human style" thing is more and more dubious to me now. It had sense years ago, but not so much now. The kind of things "human style" engine did was to make use of tricks and positional nuances that today are part of any top engine. There is a movement of coincidence as much every one has learned any good algorithm that has been developed to date.
Now, it is human and nice to accept these legendary features in order to give more pepper to a game, of course. I myself, when playing Hiarcs, prepare to play differently and so i get more kick.....

Fern

[fourthirty]

This "human style" thing is more and more dubious to me now. It had sense years ago, but not so much now.

Good point Fern. Is it the case of an engine that has a higher bias towards positional play versus tactical? (However, just about any chess engine running on modern day hardware can search 20+ ply deep in a matter of minutes - How can this not be considered positional?)

Seems like long gone are the days of Fritz 2 fiercely attacking right out of the gate just to gain material.

Are there any good articles posted that demonstrate positions played by various chess engines to reflect this categorization?

Greg

[Fernando]

Maybe so, but remember this is NOT the place to touch these engine things...!!!!
Blincoe could send his heavily armed Mossad mercenaries to kill you for much less.
Even I am tempted to send my Serbian bodyguards.
Please, make fast amend talking of, say, Champion chess Challenger that went 6 ply in middle game, with lot of time...

Just 5 ply regards
Fern

[fourthirty]

Haha. Forgive me Fern, for I have sinned.

What was my 1-Ply mind thinking? Discussing chess engines on the HIARCS chess engine forum? Blasphemy!

Of course, any dedicated unit, even the minuscule SciSys Mini Chess, could easily annihilate every modern engine running on unholy pc hardware! What he lacks in physical ability he makes up with heart and soul.

No need to unleash the Blincoe Brigade or Fernando Fighters.

[Fernando]

Sorry, but my men are already on the way. Perhaps if you send to me at once a Paypal deposit of say a couple of Kbucks I could stop them before too late.
Do not open the door in the meantime ....

[fourthirty]

As I cannot trust e-commerce, I am delivering your Kbucks via a Wells Fargo wagon. It is guaranteed to arrive in Santiago prior to the next apparition of Halley's Comet.

[Fernando]

Maybe it will never arrive. Sioux are very active these days...

Fern