[jammen]

After the bot false carded me twice in the first four tricks I conceded the hand even though we could still have beaten the contract if North had ruffed the first discard and then switched to a heart. Typically, however, when the bot is partner it waits until the second discard before ruffing in, or saves its high trump til the end.

1) Why did it lead the 4♦ from J4?
2) Why did it false card the J♠? This can only fool partner.

---

[Stephen Tu]

Basic or advanced bot?
(1) Who knows, probably insufficient thinking time/sample size to come up with a better lead? Don't know why it wouldn't lead high if leading a diamond. It's possible to construct hands where low from the Jx is necessary, but probably low probability.

(2) GIB doesn't have any notion of falsecards or signalling or whether it is fooling partner more than declarer. It is simply selecting whatever card it thinks is best on average double dummy, dealing a random bunch of hands consistent with the bidding and previous play. When cards are equivalent, it will usually choose randomly between them. In some cases it is playing high from sequences.

I would never concede based on assumptions of a robot partner's "signalling". This hand it should be able to find the heart switch.

Besides when conceding don't you see all cards?

[johnu]

jammen, on 2018-April-01, 11:21, said:

After the bot false carded me twice in the first four tricks I conceded the hand even though we could still have beaten the contract if North had ruffed the first discard and then switched to a heart. Typically, however, when the bot is partner it waits until the second discard before ruffing in, or saves its high trump til the end.

It's almost as likely the declarer GIB would blunder if given a chance. I never concede tricks unless I know that either declarer or dummy is locked in hand with only 100% winners left.

jammen, on 2018-April-01, 11:21, said:

1) Why did it lead the 4♦ from J4?

Maybe GIB switched to 2nd and 4th best leads? Really, who knows. GIB makes many random bids and plays.

jammen, on 2018-April-01, 11:21, said:

2) Why did it false card the J♠? This can only fool partner.

In general, you should assume GIB is randomly tossing cards. In theory, GIB plays some count signals but it's not consistent with them.

[johnu]

Stephen Tu, on 2018-April-01, 12:32, said:

Besides when conceding don't you see all cards?

The hand is over when you concede the rest of the tricks.

[virgosrock]

jammen, on 2018-April-01, 11:21, said:

After the bot false carded me twice in the first four tricks I conceded the hand even though we could still have beaten the contract if North had ruffed the first discard and then switched to a heart. Typically, however, when the bot is partner it waits until the second discard before ruffing in, or saves its high trump til the end.

1) Why did it lead the 4♦ from J4?
2) Why did it false card the J♠? This can only fool partner.

---

GIB has led low from (T,J,Q)x many many times. there seems to be no correlation with bidding. rarely in the middle of a hand. mostly opening leads.
If has 3 cards x+1, x, low, will play low first then x+1. I tried same in human rubber bridge and partner was totally pissed off. some crazy kind of present count.
Its out-of-the-blue cuebids when no fit has been found drive you insane. If you are a passed hand you will survive GIB when you overcall. If not a passed hand it will punish you.

I would recommend to not quit GIB. It has become substantially better than 10 years ago. It is a genius with doubled contracts. It is a genius with disrupting communications when on defense. Ah! yes when on lead with HonorX in GIB's twice bid suit, lead low not high. for example LOL

---

vrock

[jammen]

Why doesn't the K♥, J♣, or a low trump stand out? I could understand an unusual lead if it is a problem hand, but this was a no-brainer.

[virgosrock]

jammen, on 2018-April-02, 17:11, said:

Why doesn't the K♥, J♣, or a low trump stand out? I could understand an unusual lead if it is a problem hand, but this was a no-brainer.

almost never opening leads from KQx.
don't know about why not cJ. usually will lead from a sequence down to 4+ cards.

re "no-brainer". does not have a brain as we know brains. was not fully tested before releasing bidding module. and the fixes are too many to make.
extreme hands, forget it.

vrock

[sfi]

virgosrock, on 2018-April-02, 17:20, said:

almost never opening leads from KQx.

That's just good play - leading from KQx is a poor default lead.

Not leading the CJ is harder to explain though.

[virgosrock]

sfi, on 2018-April-02, 18:27, said:

That's just good play - leading from KQx is a poor default lead.

Not leading the CJ is harder to explain though.

yes, opening lead from KQx is usually a disaster.

vrock