BBO Discussion Forums: bot gives away contract - BBO Discussion Forums

Jump to content

Page 1 of 1
  • You cannot start a new topic
  • You cannot reply to this topic

bot gives away contract

#1 User is offline   thomas c 

  • PipPipPipPip
  • Group: Full Members
  • Posts: 105
  • Joined: 2012-June-27
  • Gender:Male
  • Location:mississippi
  • Interests:golf, bridge, opera

Posted 2020-June-26, 16:00

https://tinyurl.com/y8qnju5h
0

#2 User is offline   smerriman 

  • PipPipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 4,023
  • Joined: 2014-March-15
  • Gender:Male

Posted 2020-June-26, 16:24

As you know, the bot only works double-dummy.. double dummy, GIBs play at trick 3 was irrelevant; your play at trick 6 was not.
A better handviewer
Reverse engineering GIB, part 1: A treatise on the insanity of bidding simulations
0

#3 User is offline   mythdoc 

  • PipPipPipPip
  • Group: Full Members
  • Posts: 114
  • Joined: 2020-January-12
  • Gender:Not Telling
  • Location:Tennessee USA

Posted 2020-June-26, 18:56

Yes, but how much better to ruff on trick 3 then lead out the low spade. Might go down two. South has my sympathies.
0

#4 User is offline   nige1 

  • 5-level belongs to me
  • PipPipPipPipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 9,128
  • Joined: 2004-August-30
  • Gender:Male
  • Location:Glasgow Scotland
  • Interests:Poems Computers

Posted 2020-June-26, 20:30


thomas c
+++++++++++++++

Frustrating :(
GIB doesn't understand McKenny/Lavinthal signals, so you might as well lead ♦2 (although GIB might still fail to ruff).

Guthrie.tech
0

#5 User is offline   smerriman 

  • PipPipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 4,023
  • Joined: 2014-March-15
  • Gender:Male

Posted 2020-June-26, 21:44

I ran the following simulation (extra diamonds with South under the assumption that East wouldn't throw the queen, ruling out North hoping the trick holds):

north SQ75, HT84, DJ, CQT9832
west SJ4, H7652, D954, CAKJ4
south DAK8632
east DQ7

south 14-18 total points
east >=5 spades, >=4 hearts, >=11 HCP, 12-15 total points

Result of 500 deals:

- trump and club equal on 197 deals
- trump better on 139 deals
- club better on 164 deals

Obviously assumes double dummy play from both N/S, but that's a given..
A better handviewer
Reverse engineering GIB, part 1: A treatise on the insanity of bidding simulations
0

#6 User is offline   johnu 

  • PipPipPipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 5,030
  • Joined: 2008-September-10
  • Gender:Male

Posted 2020-June-27, 00:29

View Postsmerriman, on 2020-June-26, 16:24, said:

As you know, the bot only works double-dummy.. double dummy, GIBs play at trick 3 was irrelevant; your play at trick 6 was not.

As a reminder, it is double dummy on a set of hands that meet the specs of the bidding and play so far. It isn't double dummy on the exact hand that is being played.

On this hand, GIBs play didn't cost a trick, but there are plenty where trick 3 would have cost a trick.

e.g. South has ♥QJ. Then North can ruff, cross to ♠A, and the ruff with ♦10 to promote a trump trick for South.
0

#7 User is offline   smerriman 

  • PipPipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 4,023
  • Joined: 2014-March-15
  • Gender:Male

Posted 2020-June-27, 01:47

View Postjohnu, on 2020-June-27, 00:29, said:

As a reminder, it is double dummy on a set of hands that meet the specs of the bidding and play so far. It isn't double dummy on the exact hand that is being played.

On this hand, GIBs play didn't cost a trick, but there are plenty where trick 3 would have cost a trick.

Correct, but there are even more times when trick 3 gains a trick, because GIB cannot rule out South having 7 diamonds. See sim above. Eg:


A better handviewer
Reverse engineering GIB, part 1: A treatise on the insanity of bidding simulations
0

#8 User is offline   johnu 

  • PipPipPipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 5,030
  • Joined: 2008-September-10
  • Gender:Male

Posted 2020-June-27, 03:35

View Postsmerriman, on 2020-June-27, 01:47, said:

Correct, but there are even more times when trick 3 gains a trick, because GIB cannot rule out South having 7 diamonds. See sim above. Eg:



Where is ♦10 on the 3rd round of diamonds??? A decent human player would know 100% that declarer had the 10. If partner had ♦10, they would have led it and North could pitch another spade. If partner had ♥AQ, North would eventually get a spade ruff.

But maybe you are right. Maybe "GIB" can't rule out that South had 7 diamonds even though a human player would know.
0

#9 User is offline   thomas c 

  • PipPipPipPip
  • Group: Full Members
  • Posts: 105
  • Joined: 2012-June-27
  • Gender:Male
  • Location:mississippi
  • Interests:golf, bridge, opera

Posted 2020-June-27, 06:20

View Postsmerriman, on 2020-June-26, 16:24, said:

As you know, the bot only works double-dummy.. double dummy, GIBs play at trick 3 was irrelevant; your play at trick 6 was not.

while i agree the a of spades was not best, if the bot ruffs the third diamond he has to lose 2 more tricks a of spades and k of hs
0

#10 User is offline   thomas c 

  • PipPipPipPip
  • Group: Full Members
  • Posts: 105
  • Joined: 2012-June-27
  • Gender:Male
  • Location:mississippi
  • Interests:golf, bridge, opera

Posted 2020-June-27, 06:22

View Postmythdoc, on 2020-June-26, 18:56, said:

Yes, but how much better to ruff on trick 3 then lead out the low spade. Might go down two. South has my sympathies.

actually leading the 2 of diamonds causes the bot to ruff. i checked others and sure enough he ruffs when the 2 is led. i dont like this but i can live with it
0

#11 User is offline   johnu 

  • PipPipPipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 5,030
  • Joined: 2008-September-10
  • Gender:Male

Posted 2020-June-27, 13:35

View Postthomas c, on 2020-June-27, 06:22, said:

actually leading the 2 of diamonds causes the bot to ruff. i checked others and sure enough he ruffs when the 2 is led. i dont like this but i can live with it

It makes no sense that leading the 2 causes GIB to ruff. GIB does not understand card signals. Some kind of odd glitch in the play engine.
0

#12 User is offline   mythdoc 

  • PipPipPipPip
  • Group: Full Members
  • Posts: 114
  • Joined: 2020-January-12
  • Gender:Not Telling
  • Location:Tennessee USA

Posted 2020-June-27, 15:29

View Postthomas c, on 2020-June-27, 06:22, said:

actually leading the 2 of diamonds causes the bot to ruff. i checked others and sure enough he ruffs when the 2 is led. i dont like this but i can live with it


Somehow I am reassured by this.
0

#13 User is offline   thomas c 

  • PipPipPipPip
  • Group: Full Members
  • Posts: 105
  • Joined: 2012-June-27
  • Gender:Male
  • Location:mississippi
  • Interests:golf, bridge, opera

Posted 2020-June-28, 04:45

View Postsmerriman, on 2020-June-26, 16:24, said:

As you know, the bot only works double-dummy.. double dummy, GIBs play at trick 3 was irrelevant; your play at trick 6 was not.

just curious. what would u have lead at trick 6. it is obvious the bot is void in clubs. a diamond gives a ruff sluff. as it turns out this is best but i dont think its so easy to see.
0

#14 User is offline   smerriman 

  • PipPipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 4,023
  • Joined: 2014-March-15
  • Gender:Male

Posted 2020-July-08, 16:15

View Postthomas c, on 2020-June-28, 04:45, said:

just curious. what would u have lead at trick 6. it is obvious the bot is void in clubs. a diamond gives a ruff sluff. as it turns out this is best but i dont think its so easy to see.

Oh, I'm definitely not saying you *should* have done something differently. Simply that GIB is unable to distinguish lines which require you to guess, and lines which don't.
A better handviewer
Reverse engineering GIB, part 1: A treatise on the insanity of bidding simulations
0
Page 1 of 1
  • You cannot start a new topic
  • You cannot reply to this topic

1 User(s) are reading this topic
0 members, 1 guests, 0 anonymous users