Any humans better than the current bots? - Backgammon Forum - Discuss Backgammon Strategy, Clubs and Books

Two Plus Two Forums Sports and Games Chess and Other Board Games Backgammon Forum hosted by Bill Robertie.

Any humans better than the current bots?

Post Reply Subscribe

...

Page 1 of 2

1 2

Page 1 of 2

1 2

10-26-2009 , 04:35 PM

insidebackgammon

centurion

Join Date: Jul 2009 Posts: 104

In money play would you back yourself or any other human against any of the current top bots?
For the bots that would mean playing them at double lookahead so Snowie, BGBlitz and XG at 3-ply gnubg at 2-ply?
I've heard it's possible to beat gnubg by forcing it in to play against your multi checkers back backgame after which it's evaluation goes haywire and so does it's cube handling. Plus it simply doesn't understand how to evaluate an outside prime.
If this is true, what about against the other 3 bots?

Quote

10-26-2009 , 05:32 PM

Robertie

old hand

Join Date: May 2004 Posts: 1,954

Quote:

Originally Posted by insidebackgammon

I wouldn't back any human against Snowie 3-ply. The 'massive backgame' approach was able to beat Jellyfish in the mid-1990s, but it wouldn't work against Snowie with its vastly improved backgame technique. I can't speak for the other bots because I don't use them, but I'd be surprised if any of them were a dog against any human.

Quote

10-27-2009 , 07:16 AM

insidebackgammon

centurion

Join Date: Jul 2009 Posts: 104

Quote:

Originally Posted by Robertie

I can't speak for the other bots because I don't use them, but I'd be surprised if any of them were a dog against any human.

I'm curious Bill, being a world class player, why would you not use the other bots?
There may not be much difference in strength between say Snowie, Gnubg and XG but there are many instances where they disagree on the best play.

Quote

10-28-2009 , 06:14 AM

atrifix

grinder

Join Date: Aug 2004 Posts: 500

Quote:

Originally Posted by Robertie

The 'massive backgame' approach was able to beat Jellyfish in the mid-1990s, but it wouldn't work against Snowie with its vastly improved backgame technique.

I beg to differ.

If you would like to back Snowie 3 or 4 against me with no human intervention feel free.

Quote

10-28-2009 , 06:55 PM

insidebackgammon

centurion

Join Date: Jul 2009 Posts: 104

Quote:

Originally Posted by atrifix

I beg to differ.

If you would like to back Snowie 3 or 4 against me with no human intervention feel free.

Be careful what you wish for lest it come true.

Against S4 for how much and over how many games are we talking?

Quote

10-28-2009 , 11:04 PM

atrifix

grinder

Join Date: Aug 2004 Posts: 500

Quote:

Originally Posted by insidebackgammon

Be careful what you wish for lest it come true.

Against S4 for how much and over how many games are we talking?

How much would you like? The only issue is that I have to ensure that I will get paid if/when I crush it though.

Money game sessions as long as you want.

Quote

10-30-2009 , 08:06 AM

Noir_Desir

veteran

Join Date: Feb 2007 Posts: 2,333

This reminds me of some guys who beat fairly strong chess engines like Fritz 5 back then with crude h-file mating attacks by luring them to take pieces on g5 with their h-pawn.

Quote

10-30-2009 , 01:59 PM

pineapple888

Carpal \'Tunnel

Join Date: Jun 2005 Posts: 13,111

Well the angle shoot is always a possibility. If atrifix has found a specific, replicable, exploitable weakness, the bot can't adjust and he will have the edge. So it's more a bet on whether Snowie is bulletproof than whether he can "outplay" it in a broader sense.

Quote

10-31-2009 , 09:33 AM

atrifix

grinder

Join Date: Aug 2004 Posts: 500

All I'm saying is you would be better off playing me HU yourself or just taking your cash and burning it.

But if you still want to bet feel free. Just don't say I didn't warn you.

Quote

10-31-2009 , 04:24 PM

#10

TheWorstPlayer

HoldEmKillah

Join Date: Dec 2004 Posts: 21,663

Quote:

Originally Posted by atrifix

All I'm saying is you would be better off playing me HU yourself or just taking your cash and burning it.

But if you still want to bet feel free. Just don't say I didn't warn you.

I'll bet you $100, this sounds entertaining!

Quote

10-31-2009 , 09:54 PM

#11

gammoner

adept

Join Date: Mar 2008 Posts: 857

Quote:

Originally Posted by pineapple888

While it's likely true that Snowie has some weaknesses in some spots, I'd imagine the dice really have to take you there. I think you would have to give up quite a bit of equity to explicitly get yourself to a position where Snowie is so bad that you get it back. Maybe I'm wrong, I don't know...I do know that atrifix is solid though. (Good luck to you, crush that bot.)

Quote

10-31-2009 , 11:25 PM

#12

pineapple888

Carpal \'Tunnel

Join Date: Jun 2005 Posts: 13,111

Quote:

Originally Posted by gammoner

It wouldn't be exploiting subtle weaknesses in all likelihood, it would be some exotic position where Snowie thinks it's well ahead when it's actually significantly behind. Then the cube gets to 128 or whatever the limit is, and you wouldn't have to get to that spot very often to make up for the equity you are giving up by trying to get there.

Quote

11-01-2009 , 06:32 AM

#13

insidebackgammon

centurion

Join Date: Jul 2009 Posts: 104

Quote:

Originally Posted by atrifix

All I'm saying is you would be better off playing me HU yourself or just taking your cash and burning it.

But if you still want to bet feel free. Just don't say I didn't warn you.

If the bet were changed to a series of matches, what's the shortest match length you'd be willing to take on the bet?

Quote

11-01-2009 , 12:28 PM

#14

atrifix

grinder

Join Date: Aug 2004 Posts: 500

I'm really not sure there to be honest.

Quote

11-23-2009 , 03:49 PM

#15

Al Richard

stranger

Join Date: Nov 2009 Posts: 12

I've tried GNU Bckgammon and Snowie. Both have serious weaknesses in middle game.

I would be happy to meet someone betting on one of these programs. Bring a lot of cash

Quote

11-23-2009 , 11:53 PM

#16

sickdice

newbie

Join Date: Mar 2009 Posts: 21

Nack Ballard beat snowie. There have been documented matches between snowie and top players. i dont remember the results but Bill R. will know if you ask him. not too many people can beat snowie though in a large sample regardless.

Quote

11-24-2009 , 07:45 AM

#17

atrifix

grinder

Join Date: Aug 2004 Posts: 500

Quote:

Originally Posted by sickdice

not too many people can beat snowie though in a large sample regardless.

I beg to differ again.

Quote

11-24-2009 , 09:03 AM

#18

franzinator

journeyman

Join Date: Sep 2006 Posts: 271

Quote:

Originally Posted by gammoner

this.
I think that the situations where S4 is weak are very complicated ( hard backgames, outside prime ... ) ant tough to reach sometimes; in most cases you have to give some equity to reach that point.
Plus S4 is not subject to stress, being tired, concentration leaks etc etc, so my bet is on S4.
If atrifix agree to unveil his name i'm ok to a bet till 500$ ( we can fix time, lenght , and give money to Robertie for example for being sure we get paid ) .

Quote

11-24-2009 , 06:55 PM

#19

atrifix

grinder

Join Date: Aug 2004 Posts: 500

Quote:

Originally Posted by franzinator

If atrifix agree to unveil his name i'm ok to a bet till 500$ ( we can fix time, lenght , and give money to Robertie for example for being sure we get paid ) .

My name is Matt Cohn-Geier, no big secret. I post on bgonline.org.

I'd always prefer money games but matches are fine as long as it's long enough. My preference would be a match to 65 points (Snowie's longest setting).

Quote

12-02-2009 , 12:40 AM

#20

AlanDyer

Pooh-Bah

Join Date: Jun 2006 Posts: 4,682

one day people will ask this question about poker

Quote

12-02-2009 , 07:41 AM

#21

franzinator

journeyman

Join Date: Sep 2006 Posts: 271

Quote:

Originally Posted by AlanDyer

one day people will ask this question about poker

maybe... or maybe not...
poker, NL especially is so much based on psychology, gameflow and much different aspects except math.
Bg is basically a math game .

Quote

12-02-2009 , 09:57 AM

#22

Robertie

old hand

Join Date: May 2004 Posts: 1,954

Quote:

Originally Posted by franzinator

maybe... or maybe not...
poker, NL especially is so much based on psychology, gameflow and much different aspects except math.
Bg is basically a math game .

I would dispute this.

Backgammon is basically a pattern recognition game, much like chess. Math plays a role but only a small one. There are plenty of very good players who can't do anything more involved than count pips.

Quote

12-02-2009 , 01:42 PM

#23

franzinator

journeyman

Join Date: Sep 2006 Posts: 271

Bill, i'm sorry but what do you mean by "pattern recognition" ?

Quote

12-02-2009 , 02:23 PM

#24

Robertie

old hand

Join Date: May 2004 Posts: 1,954

Quote:

Originally Posted by franzinator

Bill, i'm sorry but what do you mean by "pattern recognition" ?

When you look at a position on the board, you recognize what type of position it is and relate it to what you know about other positions of that type.

When you're playing a back game, for example, you know that you can slot points in your home board pretty freely. You don't need to do calculations about how often you might get hit, because those calculations won't tell you anything useful.

Quote

12-03-2009 , 08:59 AM

#25

franzinator

journeyman

Join Date: Sep 2006 Posts: 271

Quote:

Originally Posted by Robertie

ok, but i think a position in a bg game is somewhat related to a statistichal concept.
And, for example , cube action, races, and certain posistions are strictly math problems i guess. thats why , imo, an advanced software like snowie is powerfull in such a game, and could'nt be so strong in a non-statistichal game as NLHE ...

Quote

Page 1 of 2

First

1 2

Last

Post Reply Subscribe

...

Page 1 of 2

First

1 2

Last