[pigpenz]

---

in robot tourney auction was the same, cards played the same at tricks one two three
decalrer wins dimaond plays A then Q♥
west winning Q♥ then shifts to spade which east wins.

then the fork in the road happens.

half(7 easts) returned a spade and half(6 easts) returned a diamond.
Now what I dont understand if GIB is running simulations shouldn't they be the same?
same auction
same cards played in same order for first three tricks.
shouldn't the input be the same for the simulations??

So if the bots are on the server there is no settings to predetermine this, so what I am curious
about is there just randomness built in to simulate human stupidity.
cause it sure looks like it!

It interesting cause if you dont go look at the rest of the results on alot of hands you just wont notice this,
maybe I have too much time on my hands.

[pigpenz]

heres another example I saw on results today

---

West starts J♥ some switch some continue

[barmar]

Each bot generates different hands in its simulations. If an auction doesn't provide lots of information about the hands, they may come to different conclusions about the best play. They would probably have to do simulate thousands of hands to minimize this random factor, but that's not possible in the time allowed.

[pigpenz]

Its a bot on a server does alot more calculations than we can inten seconds.
Data in
Same data in
Sometimes garbage out

The calulations should be the same, i didnt realize the bots were under time constraints

So are you saying if we play faster that has an effect on play, if not then I dont understand the time problem you are talking about.

[manudude03]

pigpenz, on 2011-September-05, 08:55, said:

Its a bot on a server does alot more calculations than we can inten seconds.
Data in
Same data in
Sometimes garbage out

The calulations should be the same, i didnt realize the bots were under time constraints

So are you saying if we play faster that has an effect on play, if not then I dont understand the time problem you are talking about.

No, each play is confined to about 5 seconds. It can only analyse so much in that time period. I know it sometimes pauses for longer, but I think that's just server lag.

[pigpenz]

yes cause I have never noticed GIB to be lacking on speed
but what I would question then is this really good for people who are paying to play in robot tourneys?
It would seem to me that it is making the results random, if people play the same the results should be the same,
if a person takes a different line of play then his results should reflect his success or failure, but having
the computer doing it doesnt seem quite right.

so it like the bots have programmed into them the human stupidity factor
so do people who play in these pay robot tournies actually go back and see the results of each hand, I guess not...
this is almost like the random sectioning and flighting that they do or used to do in ACBL BBO games, its the luck of
the draw...

[Bbradley62]

pigpenz, on 2011-September-03, 10:22, said:

in robot tourney auction was the same, cards played the same at tricks one two three
decalrer wins dimaond plays A then Q♥
west winning Q♥ then shifts to spade which east wins.

then the fork in the road happens.

half(7 easts) returned a spade and half(6 easts) returned a diamond.
Now what I dont understand if GIB is running simulations shouldn't they be the same?
same auction
same cards played in same order for first three tricks.
shouldn't the input be the same for the simulations??

So if the bots are on the server there is no settings to predetermine this, so what I am curious
about is there just randomness built in to simulate human stupidity.
cause it sure looks like it!

It interesting cause if you dont go look at the rest of the results on alot of hands you just wont notice this,
maybe I have too much time on my hands.

The various lines of play on this hand were:
5 Souths (chmtc, madrigal68, leftfoot, boldrini, cash2009) won ♦K then played ♥A and ♥Q. West won ♥K and shifted to a spade. In all five of these cases, East returned ♠2.
2 Souths (jdahlsbcgl, carbonik) won ♦A then played ♥A and ♥Q. West won ♥K and shifted to a spade. In both of these cases, East returned ♦7. Things may change based on South's first card.
1 South (maxi108) Played ♦A, ♥A, ♥8, so different things happen.
2 Souths (mikend, ndavies) won ♦A then played ♥Q to West's ♥K. In both of these cases, East/ West then played three rounds of ♠.
3 Souths (jbr_dk, Bastiaang, pigpenz) won ♦K then played ♥Q to West's ♥K. In all three cases, West returned a ♥. Again, the difference of South's first card in an influence.

So, GIB always made the same decision when faced with exactly the same circumstances.

http://online.bridge...sername=pigpenz

[pigpenz]

2 Souths (jdahlsbcgl, carbonik) won ♦A then played ♥A and ♥Q. West won ♥K and shifted to a spade. In both of these cases, East returned ♦7. Things may change based on South's first card.

really the only card that was different was weather south played the small spade or 9 spade, I am specifically talking about the hands just like a described where A♦AQ♥were played in that order
shift to spade only difference could what card south plays on High spade from east now.
The only difference was the defense between 1NT and 2NT.

[Bbradley62]

pigpenz, on 2011-September-05, 15:10, said:

2 Souths (jdahlsbcgl, carbonik) won ♦A then played ♥A and ♥Q. West won ♥K and shifted to a spade. In both of these cases, East returned ♦7. Things may change based on South's first card.

really the only card that was different was weather south played the small spade or 9 spade, I am specifically talking about the hands just like a described where A♦AQ♥were played in that order
shift to spade only difference could what card south plays on High spade from east now.
The only difference was the defense between 1NT and 2NT.

Your original post seemed to say that 13 Souths played ♦A, ♥A♥Q, but only 2 actually did. This is different than the 3 cases where South played ♦K♥A♥Q. There is not one example on this hand where GIBs did different things after their Souths played the same.

[barmar]

We've had this discussion many times before. BBO probably could program the system to use the same random numbers at all tables, but they've chosen not to.

So it's like real world duplicate bridge: even if you bid and play the same, you may get different results because of different opponents.