Help
Currently, the human player always has the most points (or tied for most points) at the table. This allows some "exploits" to take advantage of the fact that partner does not have more points than you do.
Instead of the human player always having the most points, I think an easy change to the algorithm would be that either the human or the GIB partner have the most points. If you have less than 10 HCP, then you would know that partner has the most points, but if you have 10 HCP or more, you don't know whether you or GIB partner has the most HCP.
This should mostly eliminate "safe" deviations/psyches like opening 1NT with 13 HCP in 1st or 2nd seat, since partner may have a big hand and either bid a slam or make a slam try. Or you open on a so-so 11 HCP in 1st or 2nd seat and pass responder's bid with a fit because responder has at most 11 HCP and may make a game try if given the chance.
Page 1 of 1
Improvement to best hand format?
#1
- Group: Advanced Members
- Posts: 5,030
- Joined: 2008-September-10
- Gender:Male
Posted 2017-July-09, 16:37
#2
- VIP Member
-
- Group: Advanced Members
- Posts: 6,971
- Joined: 2009-November-14
- Gender:Male
- Location:Houston, TX
Posted 2017-July-09, 17:41
Agreed.
"Genius has its own limitations, however stupidity has no such boundaries!"
"It's only when a mosquito lands on your testicles that you realize there is always a way to solve problems without using violence!"
"Well to be perfectly honest, in my humble opinion, of course without offending anyone who thinks differently from my point of view, but also by looking into this matter in a different perspective and without being condemning of one's view's and by trying to make it objectified, and by considering each and every one's valid opinion, I honestly believe that I completely forgot what I was going to say."
"It's only when a mosquito lands on your testicles that you realize there is always a way to solve problems without using violence!"
"Well to be perfectly honest, in my humble opinion, of course without offending anyone who thinks differently from my point of view, but also by looking into this matter in a different perspective and without being condemning of one's view's and by trying to make it objectified, and by considering each and every one's valid opinion, I honestly believe that I completely forgot what I was going to say."
#3
- Group: Advanced Members
- Posts: 5,030
- Joined: 2008-September-10
- Gender:Male
Posted 2017-July-09, 18:01
Currently, the human player always has the most points (or tied for most points) at the table. This allows some "exploits" to take advantage of the fact that partner does not have more points than you do.
Instead of the human player always having the most points, I think an easy change to the algorithm would be that either the human or the GIB partner have the most points. If you have less than 10 HCP, then you would know that partner has the most points, but if you have 10 HCP or more, you don't know whether you or GIB partner has the most HCP.
This should mostly eliminate "safe" deviations/psyches like opening 1NT with 13 HCP in 1st or 2nd seat, since partner may have a big hand and either bid a slam or make a slam try. Or you open on a so-so 11 HCP in 1st or 2nd seat and pass responder's bid with a fit because responder has at most 11 HCP and may make a game try if given the chance.
Moderator - Feel free to delete duplicate post in GIB Robot forum since this isn't about the GIB bid/play engine.
Instead of the human player always having the most points, I think an easy change to the algorithm would be that either the human or the GIB partner have the most points. If you have less than 10 HCP, then you would know that partner has the most points, but if you have 10 HCP or more, you don't know whether you or GIB partner has the most HCP.
This should mostly eliminate "safe" deviations/psyches like opening 1NT with 13 HCP in 1st or 2nd seat, since partner may have a big hand and either bid a slam or make a slam try. Or you open on a so-so 11 HCP in 1st or 2nd seat and pass responder's bid with a fit because responder has at most 11 HCP and may make a game try if given the chance.
Moderator - Feel free to delete duplicate post in GIB Robot forum since this isn't about the GIB bid/play engine.
#4
- Group: Advanced Members
- Posts: 2,871
- Joined: 2007-April-22
- Location:Santa Clara, CA
Posted 2017-July-10, 00:16
I think best would be full random, but if you aren't going to do full random, a reasonable way to goose the points without restricting the opponents information is that the human player always has 10+ hcp. Then you don't know, the opponents may both have better hands than you, or your partner may have better. You'll still declare and have more of the points since you always have at least the average number of points, but you can't count on anything for the other hands. You know the opponents will never have 31+ hcp, but you know that by looking at your hand, not by applying another rule.
#5
- Group: Advanced Members
- Posts: 3,162
- Joined: 2010-April-22
- Gender:Male
- Location:Hamilton, Canada
- Interests:Bridge duh!
Posted 2017-July-10, 06:18
is a good idea. One of the goals of best hand is for the player to be making most of the key decisions, so this would be lessoned. But would make certain free out of system bids in 1st/2nd risky.
Sarcasm is a state of mind
#6
- Group: Admin
- Posts: 21,589
- Joined: 2004-August-21
- Gender:Male
Posted 2017-July-10, 08:33
johnu, on 2017-July-09, 18:01, said:
Moderator - Feel free to delete duplicate post in GIB Robot forum since this isn't about the GIB bid/play engine.
I merged the two threads, since there were replies in both places.
#7
- Group: Admin
- Posts: 21,589
- Joined: 2004-August-21
- Gender:Male
Posted 2017-July-10, 08:39
I've had a similar idea myself, but I don't think I'd brought it up with management. I think the current design was done before we also had human-declares. Since the goal is to have the human play more than defend, with human-declares you'll get that regardless of whether the human or his robot partner gets the most points.
Page 1 of 1