Help
I just don't get it. If it's known that I have 4+ spades, then east has 1 more at the most, and west has only 1, meaning it might as well just try to take a trick on spades if it leads a spade since a high spades isn't going to do it any good on any future spade tricks.
Page 1 of 1
Why did it play a low spade on trick 8?
#1
- Group: Full Members
- Posts: 284
- Joined: 2011-December-15
Posted 2012-May-27, 22:58
I just don't get it. If it's known that I have 4+ spades, then east has 1 more at the most, and west has only 1, meaning it might as well just try to take a trick on spades if it leads a spade since a high spades isn't going to do it any good on any future spade tricks.
#2
- Group: Advanced Members
- Posts: 4,155
- Joined: 2003-May-14
Posted 2012-May-28, 02:43
It's going for down 2. If you have SQ and lead a club through it promotes its jack of hearts.
As for why it assumes it's a lock for you to have the SQ, I think that's because of poorly defined 4h bid. I just tested this on my home GIB; it describes the 4h bid as an "aggressive weak jump overcall, 12-hcp". Since East has turned up with HAK DA, it "can't" have the SQ since that give it 13. So my home GIB goes for down 2 also. But if you have East try some other auction, that doesn't deny 12+ hcp, like overcalling 2h then rebidding 4h, GIB cashes out!
So the answer, I guess, is to loosen the restrictions on the 4h call, which is obviously true in real life, game preempts are more wide ranging, and especially at this vul can be on the strong side. Being 4th to bid also widens the range of the action considerably. If GIB is on defense, it should be a bit looser on defining declarer's hand, to cater to unreliable opps, or just opps who have a different idea of what the right bids are in different situations in general. Maybe it needs one database to determine how its side should bid, and a second different database, with looser definitions for E/W, to cater to flaky, unpredictable opponents! Like maybe if East were dealer, instead of 4th hand; if we were GIB's partner perhaps we would not want it to preempt 4h, so we define DB for it not to do it with the 13 hcp. But if an opp chose to open 4h with the hand, it should not dismiss this hand being a possibility!
But it's tricky how much we want to make its assumptions about declarer's hand "fuzzy"; after all both humans and robots should rely heavily on the auction to make assumptions about declarer's assets, in order to defend accurately. But at least on this auction we shouldn't be surprised by an extra Q and should cater to it. And maybe at MP it should give greater weight to contract setting and a little less weight to how much it values the second undertrick.
As for why it assumes it's a lock for you to have the SQ, I think that's because of poorly defined 4h bid. I just tested this on my home GIB; it describes the 4h bid as an "aggressive weak jump overcall, 12-hcp". Since East has turned up with HAK DA, it "can't" have the SQ since that give it 13. So my home GIB goes for down 2 also. But if you have East try some other auction, that doesn't deny 12+ hcp, like overcalling 2h then rebidding 4h, GIB cashes out!
So the answer, I guess, is to loosen the restrictions on the 4h call, which is obviously true in real life, game preempts are more wide ranging, and especially at this vul can be on the strong side. Being 4th to bid also widens the range of the action considerably. If GIB is on defense, it should be a bit looser on defining declarer's hand, to cater to unreliable opps, or just opps who have a different idea of what the right bids are in different situations in general. Maybe it needs one database to determine how its side should bid, and a second different database, with looser definitions for E/W, to cater to flaky, unpredictable opponents! Like maybe if East were dealer, instead of 4th hand; if we were GIB's partner perhaps we would not want it to preempt 4h, so we define DB for it not to do it with the 13 hcp. But if an opp chose to open 4h with the hand, it should not dismiss this hand being a possibility!
But it's tricky how much we want to make its assumptions about declarer's hand "fuzzy"; after all both humans and robots should rely heavily on the auction to make assumptions about declarer's assets, in order to defend accurately. But at least on this auction we shouldn't be surprised by an extra Q and should cater to it. And maybe at MP it should give greater weight to contract setting and a little less weight to how much it values the second undertrick.
#3
- Group: Full Members
- Posts: 284
- Joined: 2011-December-15
Posted 2012-May-28, 03:06
Wow, good answer. I do think that it should give a good amount more value to setting the contract. I got 0% matchpoints for this so it really did hurt.
#4
- Group: Advanced Members
- Posts: 2,577
- Joined: 2009-May-18
- Location:Oz
Posted 2012-May-28, 06:11
Stephen Tu, on 2012-May-28, 02:43, said:
It's going for down 2. If you have SQ and lead a club through it promotes its jack of hearts.
As for why it assumes it's a lock for you to have the SQ ...
As for why it assumes it's a lock for you to have the SQ ...
Because North didn't play the SJ at trick one?
#5
- Group: Advanced Members
- Posts: 4,155
- Joined: 2003-May-14
Posted 2012-May-28, 10:19
sfi, on 2012-May-28, 06:11, said:
Because North didn't play the SJ at trick one?
No, because GIB doesn't understand signalling like J denies Q.
#6
- Group: Advanced Members
- Posts: 2,577
- Joined: 2009-May-18
- Location:Oz
Posted 2012-May-28, 12:55
Stephen Tu, on 2012-May-28, 10:19, said:
No, because GIB doesn't understand signalling like J denies Q.
That would seem to be a major flaw to not pay attention to partner's signals. Here, for example, it's an entirely avoidable two trick difference that rewards declarer for bad play.
#7
- Group: Advanced Members
- Posts: 4,155
- Joined: 2003-May-14
Posted 2012-May-28, 13:13
sfi, on 2012-May-28, 12:55, said:
That would seem to be a major flaw to not pay attention to partner's signals.
Sure, but that's life with GIB. It's pretty complicated to teach a program how to properly interpret signals, and it's not within the skill-set of the people currently working on the program to add this capability.
I actually use it a little to practice defense without signalling, which some authors have suggested as a good exercise. Forces you to not be so reliant, decide what signals are truly necessary, see if you can make plays that force partner to do the right thing instead of needing it to read your signal. Helping a bad partner on defense not make mistakes, which is a valuable skill even with good partner, you save their energy when you make "insult" plays that force them into the right line even when in principle they can work it out themselves.
Page 1 of 1