[strings11]

It was with a sense of "at last - maybe..." that I read the recent update info that suggests that GIB may no longer decide to play me for 32-35 hcp if its deranged 'alert' description so indicates - haven't seen it work yet, but boy, it would sure be nice.

However, on the same general trend line is where I rank the preposterous hand evaluation parameters used by GIB. When GIB informs you through the alert system that it has say 10-12 "total" points, it really tells you nothing about the suitability of the hand for your decision to go on to game or slam. Stiffs or better yet voids in p's primary suit receive full value as though it was a void in the enemy suit. Not particularly helpful!

High level doubles are especially problematic: if you double the enemy in a pass or correct situation GIB will generally Pass in an OMG I (GIB) don't have enough hcps to bid, p must have a rockcrusher) tone, when it holds an undisclosed SIX card suit [precisely the time that it should bid on in Either of our fit suits (2 9-card fits for example)].

On the other hand, after a competitive auction with a flat hand and a hand relatively replete with 'kings and things' it chooses to bid on.

Effectively, GIB makes it unwise to treat it as anything remotely resembling a "partner". It is almost as if it acts as a 3rd opponent, insuring that it can find a way to create a mistake (clearly Yours) which you can be punished for, thus clarifying the inherent superiority of the GIBs.

IMHO, GIB takes a point of view on the value of its hand at the outset and NEVER adjust that value in either direction, during the course of the auction. Valuing shortness in p's suit is the most egregious example of this, but there are others. It's almost comical to reconstruct the means by which the robots come to the number listed for its 'total points'.

Not knowing how the robots are designed it's hard to know how to make them 'right', but might I suggest that the initial hand evaluation consist almost always of simply high card point count? Perhaps it is already so... I do find it frustrating when a robot refuses to get involved (as in responding) if it lacks at least six nominal hcps. I'd like to see a response with an Ace in the hand - but the robots are never going to do that! Perhaps this means that modified [where some Aces are worth 7 or even 8) and perhaps even fractional hcp values need to be instigated. Another possible improvement might be for the robot to listen to 'where' things are likely to lie, and adjust its perceptions of its values accordingly.

Simply put: there need to be more "loops" in the GIB thought process - It needs to adjust its preconceived notions based on developments.

The robots should be able to recognize that there are exactly the same 52 cards - and Only those 52 cards - in play, on any one hand. They should be able to draw inferences based on likely distributional configurations; winning, and losing, trick counts; etc., better than I can. If I am given an advantage beyond my skills, then perhaps I will improve my skills through observation of events? Would that be so bad a thing?

I should not find it necessary to learn of p's (GIB's) 6-card major during the play to opps, 4Mx +1 contract. If GIB could find a way to actually bidding its cards, rather than expecting p to infer them from opps bidding that would be most pleasant. It doesn't want to overbid its hand so it declines to compete, and then when p fights through the noise of opps bidding to give GIB notice of what is expected, it either assumes p started with 35+ hcp, or has 12 of the opps trumps... [Yes, that's meant to be hyperbole.]

I cannot avoid the perception that the robots are intended to insure that they do not 'give a player an advantage' by doing things of which he or she may not be capable. I suppose this is laudable in some view of the world, but I would argue that this nearly always means that the robot will reduce any player's skills to the least common denominator, since the robots induced stupidity eliminates its ability to rise to the current expected level of play.

The above notwithstanding, it would be most helpful if the 'alert' descriptions of GIB bids would clarify ALL of the implications of that bid. When a bid is potentially Stayman-ic, informing p that this either is, or, is not, part of the picture, rather than suggesting simply that the bid means 3+ cards in the suit called, would be helpful, and is unlikely to convey any advantage to p other than to allow him to actually understand the bid.

One of the less problematic examples of incomplete descriptions is where a Michaels bid by GIB is described as 5+ cards in, for example, the other major. Maybe some are given an advantage by being told that it implies an additional suit, but it hardly seems proper to withhold that information from the description. If having that information provides an untoward advantage, then the player will 'make up for it' in others ways.

Playing with robots is generally 'practice bridge'. It will never be for the "World's Championship". Why not try to maximize the value added by making the robots mimic an actual partner?

Finally: I availed myself of the 'export deal' mechanism to alert whomever to GIB concerns. However, of late this mechanism does not appear to present itself to me. I thought I seriously tried to use it, but could not find the route? I'll look again, but I would ask that you do so as well.


Thank you.

[Bbradley62]

strings11, on 2016-March-16, 10:51, said:

Finally: I availed myself of the 'export deal' mechanism to alert whomever to GIB concerns. However, of late this mechanism does not appear to present itself to me. I thought I seriously tried to use it, but could not find the route? I'll look again, but I would ask that you do so as well.

I also find this problem, in both Google Chrome and Microsoft Edge, and both after signing on and when using "Just Play Bridge" without signing on.