[aleaxit]

Happened again: the robots' "system" makes a certain common hand impossible to bid without outright, complete lies (i.e. making bids that directly violate a description). This time: robot partner opened 4♥; with a good hand, I considered whether to control-bid or blast slam. So I hovered on the descriptions of various bids available to me: 4♠ is natural (long strong ♠); 5♣ shows the A of ♣ and denies the A of ♠...!-(

So if you have both black Aces, as I did, you cannot control-bid without outright lying, i.e., outright direct contradiction of a key aspect of the description of the bid you're making (for example, bid 5♣, mendaciously denying the A of ♠ which you do have).

Another more common example: you open 1NT, partner transfers to, say, ♥, on your 2♥ they bid 3♦, and you fit both red suits. What should normally be a joyous occasion is turned to ashes by the terrible descriptions of your various bids now available: every bid showing ♦ supports categorically shows 2 cards in ♥, while you have 3; every bid showing 3+ hearts categorically denies the 4 cards in ♦ which you also have. So with a double fit what are you supposed to do -- give up the game and turn to tiddlywinks?!

If BBO cannot devote staff time to correct such glaring mistakes (surely some bid should be available for every reasonably common hand, easily obtained in this case by having some cheap bid, say 3 ♥, promise 3+ ♥ without constraints on the length in ♦), I suggest they crowd-source the corrections!

To be specific, I suggest...: Publish the marked-up (whatever markup language) text (and whatever machine-readable description of bids the robots must consume, or whatever structured hand-describing language can be easily compiled into such a MR description) on, say, github; selectively accept PRs from volunteers to fix the text and MR description for some sequence or groups of sequences.

Surely there can be no commercial or other disadvantage to BBO from such an act, as the robots' 2/1 is hardly a trade secret or source of competitive advantage to BBO, right? Many of us would happily volunteer to express our frustration with logically inconsistent bid descriptions by offering PRs for them, and once a volunteer has proven their worth and dedication by giving enough contributions of high-enough quality BBO might decide to promote them to "core committer", able to review and authorize other contributors' PRs, further reducing the workload on BBO staff if that's an important consideration.

Comments, feedback, kudos, barbs...?

[hrothgar]

aleaxit, on 2020-June-01, 17:14, said:

Comments, feedback, kudos, barbs...?

The code base that GIB uses for bidding is a nightmare
BBO developers have posted about this on numerous occasions.

The best chance that you have to get something that actually bids decently is if BBO tosses GIB and replaces it with WinBridge or Jack or whatever...

Its cute that you think that cuebidding opposite GIB is ever a good idea

[TylerE]

What I don't understand is why, now, over a year after the buyout, GIB is still in use. The Funbridge robot is 10000000x better and actually bids pretty well. The system is even configurable, to a point.

[hrothgar]

TylerE, on 2020-June-01, 17:56, said:

What I don't understand is why, now, over a year after the buyout, GIB is still in use. The Funbridge robot is 10000000x better and actually bids pretty well. The system is even configurable, to a point.

Because the GIB code base is a mess and the new owners need to maximize their Return on Investment

[smerriman]

My favourite was the time GIBs description promised 5 spades, 5 hearts, and 5 clubs.

Open sourcing GIB has been something I've wished for for a long time, but BBO have sadly said they won't do it, despite the fact it would no doubt result in huge (and free) improvements. Perhaps when it gets replaced with Argine, they'll reconsider for old time's sake.

[TylerE]

hrothgar, on 2020-June-01, 18:00, said:

Because the GIB code base is a mess and the new owners need to maximize their Return on Investment

I'm confused as to what you mean. The new owners already owned Funbridge and whatever robot that uses. If anything it would save money as they are only supporting one robot, not two.

[johnu]

aleaxit, on 2020-June-01, 17:14, said:

So if you have both black Aces, as I did, you cannot control-bid without outright lying, i.e., outright direct contradiction of a key aspect of the description of the bid you're making (for example, bid 5♣, mendaciously denying the A of ♠ which you do have).

Why do you have both black aces? Obviously you are at fault for having the wrong cards for the system.

aleaxit, on 2020-June-01, 17:14, said:

Another more common example: you open 1NT, partner transfers to, say, ♥, on your 2♥ they bid 3♦, and you fit both red suits. What should normally be a joyous occasion is turned to ashes by the terrible descriptions of your various bids now available: every bid showing ♦ supports categorically shows 2 cards in ♥, while you have 3; every bid showing 3+ hearts categorically denies the 4 cards in ♦ which you also have. So with a double fit what are you supposed to do -- give up the game and turn to tiddlywinks?!

A good partner has the right cards for the system, such as it is. Please stop holding the wrong cards

[johnu]

TylerE, on 2020-June-01, 20:38, said:

I'm confused as to what you mean. The new owners already owned Funbridge and whatever robot that uses. If anything it would save money as they are only supporting one robot, not two.

Assuming they decide the Funbridge robot is their best option, that requires completely rewriting the interface of the Funbridge program to work with the BBO interface, assuming they decide to keep the BBO interface. In the long run it should save money as you only have to support 1 robot program, but in the short run, you can milk more money from users by not doing so.

[Vampyr]

johnu, on 2020-June-01, 20:43, said:

Assuming they decide the Funbridge robot is their best option, that requires completely rewriting the interface of the Funbridge program to work with the BBO interface, assuming they decide to keep the BBO interface. In the long run it should save money as you only have to support 1 robot program, but in the short run, you can milk more money from users by not doing so.

Maybe BBO can simply be replaced by Funbridge. Does anyone know what the interface is like?

[pilowsky]

There is absolutely nothing wrong with the GIB system.

[TylerE]

There is plenty wrong with the GIB system, even it actually played what it claimed to play, which it doesn't. There are hands that, in either "variant" are 100% impossible to bid - and that's not even getting into competitive auctions.

[zhasbeen]

pilowsky, on 2020-June-02, 05:06, said:

There is absolutely nothing wrong with the GIB system.

Are you serious? Are you really, really, serious??

[pilowsky]

Yes

[zhasbeen]

pilowsky, on 2020-June-02, 19:41, said:

Yes

Whatever you think, pilow. Gib was testing me this morning and I came here to vent. He could use a tuneup, though.

[pilowsky]

No problem hasbeen, you can think whatever you want. In my still limited experience, GIB plays better than me but then so do most people. Leo LaSota consistently plays better than almost everyone else. You don't see him on the Forum complaining about GIB. When I do, then I might pay attention. In the meantime, I'll just keep trying to learn how to play better Bridge.

[TylerE]

pilowsky, on 2020-June-02, 23:06, said:

Leo LaSota consistently plays better than almost everyone else. Y

LaSota is consistenly better at expoliting GIB>

[hrothgar]

TylerE, on 2020-June-03, 15:34, said:

LaSota is consistenly better at expoliting GIB>

He also has a very strong record playing against people

[zhasbeen]

smerriman, on 2020-June-01, 18:14, said:

My favourite was the time GIBs description promised 5 spades, 5 hearts, and 5 clubs.

Open sourcing GIB has been something I've wished for for a long time, but BBO have sadly said they won't do it, despite the fact it would no doubt result in huge (and free) improvements. Perhaps when it gets replaced with Argine, they'll reconsider for old time's sake.

"Perhaps when it gets replaced with Argine..."

By "when", do this mean that you expect that it eventually WILL happen? How does Argine compare with some of the other top robots such as Microbridge or Jack 6?

[johnu]

pilowsky, on 2020-June-02, 23:06, said:

No problem hasbeen, you can think whatever you want. In my still limited experience, GIB plays better than me but then so do most people. Leo LaSota consistently plays better than almost everyone else. You don't see him on the Forum complaining about GIB. When I do, then I might pay attention.

That's Zhasbeen you were responding to.

As far as Leo LaSota goes, he hasn't posted to these forums in years so you will have a long wait before you see a complaint from him. And many other well respected players play robot tournaments and are not even members of the forum.

[pilowsky]

That's a logical fallacy called a complete non sequitir johnu, I'm sure rumours of Leo's demise, who is the number one player in the world at the moment are exaggerated. Since zhasbeen referred to me as pilow, I thought s/he was having a bit of fun so I reciprocated. Kind of you to leap to his/her defence.

As for the question of which is the better robot. I have pitted GIB against itself in the Prime club. You can guess who won/lost; every time. The question makes no sense at all to me. Bridge is a game played by humans. Even when we compete against three robots. I actually think that the challenge format is one of the purest forms of the game.