from the previous topic on GIB and GIB analysis(deep finesse like)
what is GIB missing that doesnt allow it to evaluate the proper play of the hand
since it is a computer? On the actual hand it was played 16 times and 12 of those times
GIB played in ♦'s and not once did it make 11 ♦ tricks.
http://online.bridge...sername=pigpenz
on all the♦ contracts GIB played a high ♠ from dummy dooming the contract (endplay strip)
the actually hand is a pretty instructional hand on proper card play technique
Page 1 of 1
GIB analysis continued what is GIBs declarer play missing?
#2
Posted 2011-August-14, 10:15
What settings was the gib on? The basic (or fast robots on download) do very few simulations, and those simulations probably told it there was no danger of ruffs.
Wayne Somerville
#3
Posted 2011-August-14, 10:38
The GIB analysis is double dummy. That is the robots KNOW where all 52 cards are laying on the hand in question. The play is not double dummy. GIB deals a lot of random hands (the number depends on the GIB setting) and then tries different lines. He chooses the line that works best in the simulations.
To make this hand, first GIB has to win the spade King in hand. Then cash two clubs, before touching trumps. Then when they win the trump ACE and ruff a spade, the GIB that ruffed the spade is out of spades and trumps and is thus enplayed. Rather he returns a heart or a club, they will get no more tricks. On a club return, dummy ruffs -- declarer throws a heart, trumps are pulled and then AQ of spades throw away last two "small" hearts.
On a heart return, one heart loser goes away (win queen), then pull trumps and throw other two away on spades.
I am sure most of us would find win the SPADE and unblock clubs. But GIB doesn't analzye NS hands like that. it deals hands and trys a lot of different lines to figure out which one works most often. In the trials, a 6-1 spade split wouldn't occur that frequently, and even when it did, the other hand is likely to have three trumps so trump exit was possible or the heart king was onside. We can see the dangers and play more carefully, GIB can not.
To make this hand, first GIB has to win the spade King in hand. Then cash two clubs, before touching trumps. Then when they win the trump ACE and ruff a spade, the GIB that ruffed the spade is out of spades and trumps and is thus enplayed. Rather he returns a heart or a club, they will get no more tricks. On a club return, dummy ruffs -- declarer throws a heart, trumps are pulled and then AQ of spades throw away last two "small" hearts.
On a heart return, one heart loser goes away (win queen), then pull trumps and throw other two away on spades.
I am sure most of us would find win the SPADE and unblock clubs. But GIB doesn't analzye NS hands like that. it deals hands and trys a lot of different lines to figure out which one works most often. In the trials, a 6-1 spade split wouldn't occur that frequently, and even when it did, the other hand is likely to have three trumps so trump exit was possible or the heart king was onside. We can see the dangers and play more carefully, GIB can not.
--Ben--
#4
Posted 2011-August-14, 14:23
inquiry, on 2011-August-14, 10:38, said:
The GIB analysis is double dummy. That is the robots KNOW where all 52 cards are laying on the hand in question. The play is not double dummy. GIB deals a lot of random hands (the number depends on the GIB setting) and then tries different lines. He chooses the line that works best in the simulations.
To make this hand, first GIB has to win the spade King in hand. Then cash two clubs, before touching trumps. Then when they win the trump ACE and ruff a spade, the GIB that ruffed the spade is out of spades and trumps and is thus enplayed. Rather he returns a heart or a club, they will get no more tricks. On a club return, dummy ruffs -- declarer throws a heart, trumps are pulled and then AQ of spades throw away last two "small" hearts.
On a heart return, one heart loser goes away (win queen), then pull trumps and throw other two away on spades.
I am sure most of us would find win the SPADE and unblock clubs. But GIB doesn't analzye NS hands like that. it deals hands and trys a lot of different lines to figure out which one works most often. In the trials, a 6-1 spade split wouldn't occur that frequently, and even when it did, the other hand is likely to have three trumps so trump exit was possible or the heart king was onside. We can see the dangers and play more carefully, GIB can not.
To make this hand, first GIB has to win the spade King in hand. Then cash two clubs, before touching trumps. Then when they win the trump ACE and ruff a spade, the GIB that ruffed the spade is out of spades and trumps and is thus enplayed. Rather he returns a heart or a club, they will get no more tricks. On a club return, dummy ruffs -- declarer throws a heart, trumps are pulled and then AQ of spades throw away last two "small" hearts.
On a heart return, one heart loser goes away (win queen), then pull trumps and throw other two away on spades.
I am sure most of us would find win the SPADE and unblock clubs. But GIB doesn't analzye NS hands like that. it deals hands and trys a lot of different lines to figure out which one works most often. In the trials, a 6-1 spade split wouldn't occur that frequently, and even when it did, the other hand is likely to have three trumps so trump exit was possible or the heart king was onside. We can see the dangers and play more carefully, GIB can not.
Yes Ben that is whatmakes the hand pretty neat. If we really stop and think about it the hand is fool proof barring a 8-1 Club break. When i watch a lot of hands gib plays in hands played listi keep wondering why it doesnt seem to count its tricks or so it appears....so maybe its double dummy simulations lead it astray more than it helps.....mzybe i shouldnt call them double dummy simulations.
Page 1 of 1