Help
Choose how you like GIB to play- note BBO will never be able instill play sense other than follow the simulations into GIB.
Page 1 of 1
GIB Play
#2
- Group: Advanced Members
- Posts: 4,942
- Joined: 2005-January-28
- Gender:Male
- Location:Cologne, Germany
- Interests:Sleeping, Eating
Posted 2012-January-24, 12:39
Silly to have a poll about this, rather it's the sort of thing which needs testing to find the best method.
"Highest percentage line" certainly suffers from GIB making wrong assumptions for its sims.
"Highest percentage line" certainly suffers from GIB making wrong assumptions for its sims.
"One of the painful things about our time is that those who feel certainty are stupid, and those with any imagination and understanding are filled with doubt and indecision"
-- Bertrand Russell
-- Bertrand Russell
#3
- Group: Admin
- Posts: 22,136
- Joined: 2004-August-21
- Gender:Male
Posted 2012-January-24, 16:16
We've actually been thinking about this since that 6♦ hand that it went down in by taking a double finesse to try to make 7.
The problem is that in matchpoint play, this is not winning strategy. This is why some players are good at IMPs, but not MP. Humans are able to tell whether they're in normal or unusual contracts, and then decide whether to play safe or go for overtricks. But we don't know how to program that decision into GIB.
GIB does know the difference between IMP and MP, and plays safer in IMP. Not always the safest line, but it doesn't take as big risks.
The problem is that in matchpoint play, this is not winning strategy. This is why some players are good at IMPs, but not MP. Humans are able to tell whether they're in normal or unusual contracts, and then decide whether to play safe or go for overtricks. But we don't know how to program that decision into GIB.
GIB does know the difference between IMP and MP, and plays safer in IMP. Not always the safest line, but it doesn't take as big risks.
#4
- Group: Advanced Members
- Posts: 4,942
- Joined: 2005-January-28
- Gender:Male
- Location:Cologne, Germany
- Interests:Sleeping, Eating
Posted 2012-January-24, 16:58
barmar, on 2012-January-24, 16:16, said:
The problem is that in matchpoint play, this is not winning strategy. This is why some players are good at IMPs, but not MP. Humans are able to tell whether they're in normal or unusual contracts, and then decide whether to play safe or go for overtricks. But we don't know how to program that decision into GIB.
Well at least for uncontested auctions you could have GIB simulate what would have happened if it had been 2 GIBs bidding. Probably not very useful though.
For NT contracts, you could compare the combined HCP to the level, e.g. if you have 25-31 HCP and are in 3NT that's "normal". Less useful but in the same vein compare combined TP with the level of a suit contract. Throw in some heuristics like play all out when we only found our second-best fit or something...
But it's not easy, right. After all, if GIB had an idea of what the "normal" bidding would be, then it could just bid like that in the first place...
"One of the painful things about our time is that those who feel certainty are stupid, and those with any imagination and understanding are filled with doubt and indecision"
-- Bertrand Russell
-- Bertrand Russell
#5
- Group: Advanced Members
- Posts: 1,529
- Joined: 2008-December-02
Posted 2012-January-24, 19:32
mgoetze, on 2012-January-24, 12:39, said:
Silly to have a poll about this, rather it's the sort of thing which needs testing to find the best method.
"Highest percentage line" certainly suffers from GIB making wrong assumptions for its sims.
"Highest percentage line" certainly suffers from GIB making wrong assumptions for its sims.
Far too complicated ( at least millions of hands and million of possibilities depending on the limited sample of simulations that GIB does) to do testing to work this out- its personal decision of what players want- sure but secure or "optimum" but risky.
Most likely sure and secure of bidding and play would probably get near tops anyway.
Page 1 of 1