Very frustrating to see field results where Robot acted differently under non-varying circumstances.
Case in point: Opponents PASS throughout => Auction is 3♠ => 4♠ as most tables... but 3♠ => 6♠ at a few tables. I can accept and understand some "fuzzy" logic built into Robot "thinking / decision making" but I think above example is too fuzzy.
Along same lines: I see Robot opponents when they are Defenders taking EXTREMELY varied lines of defense with minor variations is the auction. This (IMHO) doesn't help the Robot/Human partnership nor does it encourage me to opt to Robot Rental as a form of play.
Page 1 of 1
Robots varying actions under non-varying circumstances.
#2
Posted 2011-July-01, 16:04
Each of the bots simulate hands separately. Some came to the conclusion that 4S was best, others came to the conclusion that 6S is best. As for the defence, I find that the gibs are extremely deep thinkers so can make some pretty bizarre plays that often work out.
Wayne Somerville
#3
Posted 2011-July-01, 16:23
Andy_L, on 2011-July-01, 09:55, said:
Very frustrating to see field results where Robot acted differently under non-varying circumstances.
Case in point: Opponents PASS throughout => Auction is 3♠ => 4♠ as most tables... but 3♠ => 6♠ at a few tables. I can accept and understand some "fuzzy" logic built into Robot "thinking / decision making" but I think above example is too fuzzy.
Case in point: Opponents PASS throughout => Auction is 3♠ => 4♠ as most tables... but 3♠ => 6♠ at a few tables. I can accept and understand some "fuzzy" logic built into Robot "thinking / decision making" but I think above example is too fuzzy.
Welcome to the forum, Andy.
Is this the hand you are referring to?
According to the traveller, only one pair bid 3♠-6♠, and the human player was sitting North at that table; all robots bid 4♠ over 3♠. http://online.bridge...username=Andy_L
#4
Posted 2011-July-02, 03:25
I think there could be a case for giving the robots the same seed at all tables in robot duplicate. So if the conditions are exactly the same at two tables then the robot would do the same.
But when the conditions are only "almost" the same then you should not expect the robots to do the same.
But when the conditions are only "almost" the same then you should not expect the robots to do the same.
The world would be such a happy place, if only everyone played Acol :) --- TramTicket
#5
Posted 2011-July-03, 11:12
I guess the idea is to be more like real duplicate. Often you'll have an identical auction at multiple tables, but different results because the opponents make different guesses or mistakes. Although duplicate bridge is intended to mitigate the extreme randomness of rubber bridge, it doesn't eliminate it totally. Although it's frustrating when an opponent gets a good score because they happened to play a particular hand against the worst pair in the club, while you did nothing wrong, that's a part of the game.
Page 1 of 1

Help
