[iandayre]

Most of the worst robot bidding atrocities have been fixed so I rarely post here any more. But I could not resist here.

With no one vulnerable in a Daylong ACBL SYC game, N robot held KJ74, KJT84, J3. 97. The bidding went 1C by E, 1H by me (S), 3C weak raise by W. And somehow N decided that the excellent support and sound values were not worthy of a bid! PASS!!!!!. This is absolutely and utterly bizarre.

[zhasbeen]

 iandayre, on 2020-May-25, 13:19, said:

Most of the worst robot bidding atrocities have been fixed so I rarely post here any more. But I could not resist here.

With no one vulnerable in a Daylong ACBL SYC game, N robot held KJ74, KJT84, J3. 97. The bidding went 1C by E, 1H by me (S), 3C weak raise by W. And somehow N decided that the excellent support and sound values were not worthy of a bid! PASS!!!!!. This is absolutely and utterly bizarre.

You ain't seen nuthin' yet. You don't play that much, although I do see that you've stepped it up a little lately

"Most of the worst robot bidding atrocities have been fixed"

Not my finding at all. I will submit a few soon. I have hundreds, but a little pressed for time now. I still like to play but I see zero improvement in Gib's bidding.

[zhasbeen]

 iandayre, on 2020-May-25, 13:19, said:

Most of the worst robot bidding atrocities have been fixed so I rarely post here any more. But I could not resist here.

With no one vulnerable in a Daylong ACBL SYC game, N robot held KJ74, KJT84, J3. 97. The bidding went 1C by E, 1H by me (S), 3C weak raise by W. And somehow N decided that the excellent support and sound values were not worthy of a bid! PASS!!!!!. This is absolutely and utterly bizarre.

I know the protocol says 1 hand per thread, but with participation so light--these seem appropriate for the topic:

https://tinyurl.com/yabng2ld
Takeout double of 2 hearts?

https://tinyurl.com/ycqy62uj
Simple raise with 4 trump, 2 outside aces with one heading a good 5-card suit.

https://tinyurl.com/y9bunc44
Another simple raise. Says Mr.Gib, "I think I'll let them play 4 hearts rather than our 10-card spade fit". Just awful, what can we say...

https://tinyurl.com/y7bpqbd3
3NT??? Didn’t I bid clubs at 3-level? Where’s the heart stopper? Ah, a void; that should be helpful in NT.

https://tinyurl.com/ycle9pnl
Perfect balancing double for most beginners, but not for ol’ Mr. Gib. Geez, and he's not even vul.

Meanwhile the customers keep pouring in. Maybe BBO will get some needed competition before long.

[johnu]

 zhasbeen, on 2020-May-26, 15:03, said:

I know the protocol says 1 hand per thread, but with participation so light--these seem appropriate for the topic:

https://tinyurl.com/yabng2ld
Takeout double of 2 hearts?

https://tinyurl.com/ycqy62uj
Simple raise with 4 trump, 2 outside aces with one heading a good 5-card suit.

https://tinyurl.com/y9bunc44
Another simple raise. Says Mr.Gib, "I think I'll let them play 4 hearts rather than our 10-card spade fit". Just awful, what can we say...

https://tinyurl.com/y7bpqbd3
3NT??? Didn’t I bid clubs at 3-level? Where’s the heart stopper? Ah, a void; that should be helpful in NT.

https://tinyurl.com/ycle9pnl
Perfect balancing double for most beginners, but not for ol’ Mr. Gib. Geez, and he's not even vul.

Meanwhile the customers keep pouring in. Maybe BBO will get some needed competition before long.

I don't know if this qualifies as a defense of GIB, but your examples don't crack my top 200 worst bids and plays by GIB that have been posted on this forum.

[iandayre]

 zhasbeen, on 2020-May-26, 12:16, said:

You ain't seen nuthin' yet. You don't play that much, although I do see that you've stepped it up a little lately

"Most of the worst robot bidding atrocities have been fixed"

Not my finding at all. I will submit a few soon. I have hundreds, but a little pressed for time now. I still like to play but I see zero improvement in Gib's bidding.

I used to be one of the most active posters on this thread. I am proud to say I once got a vicious nastygram from Fred Gitelman himself! But you can't tell me GIB hasn't improved substantially. It used to be notorious for bidding 3 card suits at the 3 level in live auctions. Haven't seen that in ages.

[iandayre]

 johnu, on 2020-May-26, 15:52, said:

I don't know if this qualifies as a defense of GIB, but your examples don't crack my top 200 worst bids and plays by GIB that have been posted on this forum.

Maybe #1 does that is really bad AND then accepts the invitation GAG

[zhasbeen]

 johnu, on 2020-May-26, 15:52, said:

I don't know if this qualifies as a defense of GIB, but your examples don't crack my top 200 worst bids and plays by GIB that have been posted on this forum.

Ok, johnu, I'll find some that are a little more hard core, but as iandayre said, "#1 is really BAD"

Only #'s 2 and 5 are just poor bidding

#3 is a pure game destroyer and it's hard to imagine how it could be programed to make a simple raise with that hand. At the very least it is a serious error. Four spades has to be automatic.

#4? Check again. Partner bid diamonds and clubs. If south bid the no trump it wouldn't quite as bad, but north bidding it with a heart void and known 10 or 11-card club fit would have to make any "Horrific" list.

[smerriman]

 zhasbeen, on 2020-May-26, 22:17, said:

Ok, johnu, I'll find some that are a little more hard core, but as iandayre said, "#1 is really BAD"

GIB's book bid is to pass 2♥. So there's really no bug to fix here; if GIB simulated a number of hands and the result for those hands was that dbl and 4S worked out better on average, then it worked out better on average. Those sims can always lead to unusual situations; if probability states that a finesse works 55% of the time and a drop works 45% of the time, then a large number of times a sim will tell GIB to play for a drop.

Given the simplistic nature of GIBs points-based bidding rules the reasons for its other bids seem pretty obvious and easy to imagine programming wise, even if they're wrong.

[zhasbeen]

 smerriman, on 2020-May-26, 22:43, said:

GIB's book bid is to pass 2♥. So there's really no bug to fix here; if GIB simulated a number of hands and the result for those hands was that dbl and 4S worked out better on average, then it worked out better on average. Those sims can always lead to unusual situations; if probability states that a finesse works 55% of the time and a drop works 45% of the time, then a large number of times a sim will tell GIB to play for a drop.

Given the simplistic nature of GIBs points-based bidding rules the reasons for its other bids seem pretty obvious and easy to imagine programming wise, even if they're wrong.

The closest I ever come to programming is Excel macros, but even that gives me an appreciation of how difficult must be to program a robot to play bridge. However, when one of them makes bad bids with the frequency that Gib does it really tests our patience. When you are playing you don't care if it was because of a simulation, bug, or whatever. Then end result is frustration and disappointment. You just hope it will get better. Some continue to play frequently, some cut back, and some leave for good.

We know there are several better robots out there, and would hope that BBO would finally purchase one that will make its customers happier. Either that, or pay the extra money it would take to hire programmers having all the hidden documentation that BBO's current staff apparently does not have access to. I was hoping that with all the extra money they have been taking in since covid-19 struck, something might be in the works. I would suspect that nothing will done unless there is some pressure in the form of new competition or something else that affects their bottom line. If/when covid-19 lifts I'm guessing the traffic will be less than now, but still more players than before, and could even be close to the same due to new players that never played much in offline clubs.

[pescetom]

 zhasbeen, on 2020-May-27, 12:34, said:

We know there are several better robots out there, and would hope that BBO would finally purchase one that will make its customers happier. Either that, or pay the extra money it would take to hire programmers having all the hidden documentation that BBO's current staff apparently does not have access to.

Throwing extra money at GIB does not make sense and would be perverse for a company that already struggles to recreate the core business client in HTML5.

What makes it worse is that the owners of BBO already own a better and more flexible robot, Argine - it uses the same outdated approach as GIB, but is evidently better coded and designed and being a more recent and internal project it would be relatively simple to port and support.

[zhasbeen]

 pescetom, on 2020-May-27, 13:07, said:

Throwing extra money at GIB does not make sense and would be perverse for a company that already struggles to recreate the core business client in HTML5.

What makes it worse is that the owners of BBO already own a better and more flexible robot, Argine - it uses the same outdated approach as GIB, but is evidently better coded and designed and being a more recent and internal project it would be relatively simple to port and support.

Do you have any idea why they wouldn't try Argine? All about money?

[smerriman]

 zhasbeen, on 2020-May-27, 13:56, said:

Do you have any idea why they wouldn't try Argine? All about money?

 barmar, on 2020-May-08, 16:28, said:

We've discussed internally switching to Argine, FunBridge's robot.

But adding a new bot is not trivial, there isn't a standard bot interface that allows them to be interchanged. So we'd need to develop code to make Argine look like GIB to all our bot-related software.

[pescetom]

 zhasbeen, on 2020-May-27, 13:56, said:

Do you have any idea why they wouldn't try Argine? All about money?

Money? They could only gain.
Lack of vision or grip by corporate management is a much more likely explanation.
Vision to me is working with others on an open standard interface that allows bots to be interchanged across all online platforms.
An internal modification to replace GIB with Argine is just a question of grip.

[johnu]

 pescetom, on 2020-May-27, 14:27, said:

Money? They could only gain.
Lack of vision or grip by corporate management is a much more likely explanation.
Vision to me is working with others on an open standard interface that allows bots to be interchanged across all online platforms.
An internal modification to replace GIB with Argine is just a question of grip.

IMHO, it makes no sense to maintain 2 separate and entirely different bridge platforms. Take the best features of each and combine them into a single product. That will mean sunsetting one of the bridge playing programs, presumably GIB since there doesn't seem to be anybody who can modify the bidding and play functions beyond basic or standalone features. Then, you only need 1 set of programmers and designers since there is only 1 program.