calm01

http://tinyurl.com/3kpc2r6 On this deal the first club lead was won in dummy, declarer playing low. Then the lowest diamond (2) was led on the second trick. GIB chose to play the the apparently no-win Jack of diamonds as East. There must be a fundamental programming failure there as no sensible simulation of hands can make this play reasonable as partner is unlikely to have both exactly Kx of diamonds (declarer has at least 3 on the bidding) and also the King of Spades as this would leave opener with just 8 points. Whatever happened to second hand low unless there is a very good reason such as an inspired play of the Ace of diamonds as if partner has the Spade King this play and a Spade switch defeats 3NT by one trick - also at little cost if partner does not have the King of Spades? Calm01

'Walking the dog' was losing bridge from the very first day it was invented. Opponents (sometimes even GIB) can also count their fit(s) if given the bidding space. Why - with a super fit (10+) - give the opponents space to describe their hands and show both strength as well as support and/or make lead directing bids or locate a double fit when you know where you want to be and have no intention of punishing opponents? Answer - you prefer losing as opposed to being able to bid more often. So why has a human psychological problem apparently been actively been programmed into GIB? Call in the vet quickly to humanely put down that electronic dog! calm01

GIB frequently falsecards as a defender when trumping in, preferring to ruff say with the Jack of trumps when the 4 or 7 of trumps will do nicely. More often than not there is no alternative way of playing trumps as declarer nor for declarer to assume the defender has no lower trumps to trump again 'knowing' that the Jack was the defenders lowest trump. So the false card normally fails to gain and sometimes loses by providing an otherwise impossible entry to dummy or just giving away a trump trick. A net losing strategy. I once made an impossible slam thanks to the trump 10 becoming an entry to dummy. The GIB logic appears flawed (if logic is what you call it). If done by a human player the behaviour may fall into the category of grandstanding. When down to two cards of equal value, GIB will usually discard the higher - again if human it would be showing off. Perhaps GIB is programmed to appear to be an emotionally challenged palooka to make us mere humans feel right at home! Calm01

Oh happy days! Thanks for the money. Opponents overcalling on 4 card suits at the one level (and 70%+ of the time paying the price) is what helped me get through college without worrying about where the rent or my next meal was coming from. But I admit the opponents standard of declarer play was generally not that hot, but then neither was mine. They often lost control in the play when their partner raised on 3 - especially when playing in a 3-3 fit after being forced as they did not have the declarer skills to handle even a 4-3 fit or just as often ended up declaring a hand when it was better to defend. If their declaring skills for 4-3 fits were up to the job, then their defending skills were also at a higher level and so would usually benefit from defending with defensive hands! If all you have to say is to use a 4 card suit overcall on a hand unsuited to double for takeout, you basically have a defensive hand unsuited to a 1NT overcall and can pass and wait for a decent partner to protect if it is your sides hand. Yes, I know being declarer gives you an significant edge at bridge but unless you are Ron Klinger or a member of the Italian Blue team with outsize declaring skills you often end up declaring more often with a hand better suited to defend with. Yes, I have played at a bridge club where some players just can't defend for toffee or just hate defending - so they overcall with abandon and so end up losing that way instead most of the time. Sputnik doubles existed even in my youth and so it was not really much disruption to overcall at the one level even with 1 spade. Overcalling on 4 card suits at the one level carries bidding frequency benefits as opposed to bridge value. For some the this carries a high psychological value and this seems to outweigh, for them, the actual bridge benefit. It is this longstanding habit of employing 4 card overcalls without the associated declarer skills that continues to fill my wallet when playing for profit - the happy days just go on and on. Thanks for the money. calm01

Bbradley62. Thanks for this - could you supply more detail? Also what is provisionally planned - no guarantees is understood - for the next release? calm01

Yes, BUT... only if GIB has built up a realistic view of Souths distribution. Since GIB does not always give normal preference at the 2,3,4 or 5 level, why should the 6 level be different. GIB does not appear build up a realistic picture of it partners distribution even when the bid descriptions of its partners bids make bridge sense. Until GIB computes the length messages of it partners bidding this kind of behaviour will continue at all bidding levels. Consider it amusing that such a basic concept, so easy to program is without GIBs current programming. calm01

GIB sometimes passes partners bids described as forcing even with silent opponents. GIB descriptions must be seen as just descriptions, nothing more. The descriptions are not a reflection of what GIB 'understands' let alone honours. So whether the 3C bid may be described as forcing or not forcing on you, GIBs partner, is not necessarily what GIB understands by its 3C bid. Your overcall of 2H is probably described as 5+ H, but GIB might not understand your 2H call as being a 5 card or longer suit despite the description. So the failure to raise to the clearly superior 3H may follow from a failure to recognise an 8 card or longer fit is held by the partnership. If you have ever taught palookas to learn bidding at bridge you will recognise this as classic palooka bidding: "I can see my hand, I cannot see yours.". When it comes to upgrading the bidding of GIB, perhaps the joke about an Irishman, on being asked the way to Tipperary, replies "I would'nt start from here."! Updating a fundamentally flawed bidding approach just makes for even more convoluted code - an exercise in futility perhaps. Maybe this is why a man-years effort every year (according to an earlier reply from Fred to one of my posts) goes on to yields such limited returns on BBOL investment. What appears to be required is to restart the GIB bidding approach from scratch: 1) update the descriptions of all bids made by GIB, or its partner, to make bridge sense in terms of length/distribution and strength and forcing implications promised..., 2) the descriptions need to consider any variations required according by the four types of relative vulnerability, 3) ensure GIB 'understands' the same as the descriptions in all four situations, 4) ensure GIB honours the decriptions in terms of supporting and raising partner and preempting opponents, using fast and slow arrival ... Only then will there be any hope of getting any real productivity out of the man-years work spent on GIB each year by BBOL and/or getting GIB past the palooka bidding stage it is clearly stuck in. Anyone who has taught bridge to beginners knows all this is just common sense - not rocket science. The trouble with accepting common sense is that it requires us to admit we have got it wrong all these years. Delusions about the quality of GIB's bidding held by even one or more BBOLers will most often be born of a refusal to admit that "we got it wrong all these years" in terms of the current bidding approach programmed into GIB. "Flogging a dead horse" rather than getting a new one is another way of saying the same thing. Sometimes we don't notice the horse is dead as we are too busy flogging it! This particularly true if it was our pony when we were younger. I continue to smile and occasionally belly laugh - human nature is so strange - and continues to make fools of us all - but where would humour be without our own crazy human nature to laugh at? calm01

I agree with xxhong. It seems GIB does not understand that bidding in 2/1 normally communicates length or is a waiting bid as he describes. Strength is also communicated, but is mostly implied by level of bidding of length except perhaps for 2C and no trump openers.. Most bids thus show length and usually imply some indication of strength. Rebiddable suits are just those that are rebid unless they are waiting bids as xxhong suggests. So GIB mentioning of various forms of rebiddable is mostly redundant. Often GIB does not always honour length information - e.g. GIB frequently normally does not give false preference and sometimes does not return to a 'known' 8 card fit, preferring to languish in a 4-2 or Moysian fit. Until GIB is programmed understand and remember and honour basic length information from its partner it will continue to perform, on occasion, like a palooka bidding partner. But that is perhaps part of the charm and attraction - the sheer randomness of playing with a bidding palooka - and it is a level playing field for everyone playing with GIB after all. I do agree with Fred that GIB can be an above average declarer and sometimes a decent defender. Shame about the bidding algorithms. Palookas usually begin understand what partners distribution information is after about 100 hands and to communicate back distribution after another 100 hands. GIB will require a reprogramming as it appears to have no learning algorithims related to what length partner has shown. Calm01

:rolleyes: :rolleyes: :rolleyes: reply completed after connection failure Bbradley62/Barmar, Both of your are correct in my experience. In particlar well-defined sequences, GIB seems to notice what you have not bid - such the class of situations mentioned by Barmar. In general situations GIB appears to have little or no actual memory or partners bids and their meanings let alone and failure to bid at all - such as the recent post where a 20 point assessment was made by GIB of a human that opened and immediately passed at the next opportunity but bid later to the 4 level with a ten card fit. The two core basic bridge bridge principles could be said to exist within the framework of any agreed bidding system: a1) Every bid made communicates a meaning (strength and/or length) to partner, a2) every bid NOT made communicates a (strength and/or length) meaning to partner, a3) you can add to the meanings previously communicated to partner (extra strength and/or length), a4) you CANNOT UNcommunicate what you have already told partner. b1) You can tell the truth, b2) you can misrepresent your holding just for fun or for upsetting/confusing the opponents, b3) you can be pessimistic or optimistic or downright crazy, cognisant of the vulnerability or trust to some extent the opponents, b4) you must live or die by what you say or do not say and partners responses. For these princincles to operate it is necessary for GIB to be programmed to: i) notice and take account of strength and length meanings of partners bids both made and not made, ii) use discrimination. To a fair extent both can be programmed and provide a core to be built upon. GIB is not yet very strong on either of these core bidding principles. The evidence of this is clear to anyone who actually plays with GIB over hundreds and thousands of hands as many on this site have clearly done. One cannot judge without seeing any GIB design specifications, let alone the code implemented, but there is little evidence of these basic bidding principles being substantively employed by GIB. The use of sheer computing power to differentiate between $1 a day and $1 a week versions of GIB suggests that the principles used by GIB are of a completely different order as neither memory or discrimination require much computing power to implement. To he extent this assessment is even partly true, our expectations of substantive improvement in GIB bidding may be destined to be unsatisfied. Regards, Calm01

Bbradley62/Barmar, Both true in my experience. In particlar well-defined sequences GIB seems to notice what you have not bid - such the class of situations mentioned by Barmar. In general situations GIB appears to have little or no memory at all - such as the recently mentioned post where a 20 point assessment was made by GIB of a hand that opened and and passed at the next opportunity but bid to the 4 level with a ten card fit. The bridge principle that: .

Barmar, What a good idea. When users of GIB know fuller and usable details of each improvement, then we can commence to fully take advantage of any development effort that has been expended on our behalf by BBOL. Enjoyment increases and the developers gain more satisfaction that their efforts are fully appreciated. At the moment it seems 'information is power'. Maybe not intentionally, but BBOL appears to collectively take the stance that if does not tell us the detail of enhancements then it knows more than we do and so has the upper hand. Often, but not always, I found when a KPMG management consultant, that this all too common behaviour usually reflected the unspoken but clearly communicated attitude of senior management in the business. Once the management 'opened up to be criticised' - their perspective not mine - much of the perceived criticism that made them so sensitve in the first place magically reduced. What a surprise! We are all so strange: we often unwittingly generate by our own behaviour that which we most fear. But that what makes life such an interesting game to play - much like in bridge when so many, holding little or nothing, choose to pass partners clear takeout double through fear of being doubled for penalties as declarer. Laughing as always at my own foolishness reflected in others, a very amused Calm01

Fred, Tanks for your prompt and informative reply to one of the three queries. Feedback on bug fixes successfully implemented is not a lot to ask for as it costs little and provides substantial benefits - a good return on investment. Examples for popular software products are: http://msdn.microsoft.com/en-us/ie/ff959805 http://www.mozilla.org/en-US/firefox/releases/ If a user is not aware of new features/bug fixes he/she is unlikely to fully benefit from them and thus some of the development effort is effectively wasted as it is neither totally visible nor can it be fully appreciated. You site advertises new features such as bridge bingo - why not advertise GIB enhancements in the Robot World section of the site? Regards, calm01

Hi, I currently have limited belief that using this forum for reporting bugs (big or small) actually results in GIB being substantially improved. The main function of the GIB part of the forum appears to be to annoy those who work for BBOL and are already frustrated that more resources have not been allocated to improving GIB. So I have three questions including one for Fred G.: 1) Is there a list of GIB releases/versions and dates implemented with detail of bug fixes included in each release/version? 2) Is such a list made accessible to us users? 3) Could those from BBOL who kindly reply to real or apparent bug reports please do so without giving the impression that Fred G. allocates little or no resources to improving GIB? Regards, calm01

Even bridge beginners would be unlikely to find this bid made by GIB partnering GIB with silent opponents The bidding was: 1D 1S 2D 3S 4H pass The 4H bid was described as "4+ H; 3-S; rebiddable D; 14-16 total points" as did sort of correspond as the opening GIB was holding S 6 H K J 10 7 D A K Q 7 6 5 C 9 7 Partner is unlikely to hold a heart fit on this bidding, and you cannot really stand 5D even if partner has one or two. Of the 16 hands played with these cards 4H was made once, was down 2/3/4 on 14 other occasions and 3S was made once. Partner is asking if you have 2S (bid 4S) or if holding fewer spades but many extra points to compensate (bid 3N) or else pass. Not many beginners would consider 4H as an option even if the above was understood. 1X 1 Major 2X 3 repeated Major is a fairly common bidding sequence and so GIb requires to be programmed to cope effectively. What do you feel?

Hi Cloa513, I have compassion for all those that appear to possess such a despairing approach to (working) life. Life (which includes work) can be fun rather than frustrating if you choose to make it so. It might pay you to see the place/state your words appear to come from. "Probably not little effect with programming like it is" - Fred will fail to leverage the full value of the GIB potential and will be eventually overtaken by others. Such can be business and is not your problem as such but all of us users and workers alike are impacted by this unfortunate approach. Perhaps trying to change Freds business priorities would be more fulfilling for you and all the BBO staff. Or you will have to remain mostly in defensive mode and repeatedly note (elsewhere) that this problem or that problem keeps recurring until programming resources change. Has the change in programming resources kept pace with even 30% of the last 2 years growth in user numbers? Are users involved in setting priorities? "3- which means 3 or less" - it rarely pays to assume stupidity in others when conversing. "There are many silly descriptions which don't respresent the hands or what GIB thinks the opponent's hand is." - the concept of there being lots of mistaken descriptions is not a reason to despair and do little but to engage with enthusiasm to eat the elephant one slice at a time. Look to the future not reasons to continue behaving as you done in the last two years. Cheer up!

All Robot bid descriptions could be computer scanned to see where there are more than 13 cards in one hand! A recent example is 1H 1NT 2C 2D 2S all pass where the 2S bid was described as "4+C; 3-D; 5+H; 4+S; 14-18 total points" The bidding does clearly suggest a 4,5,0,4 pattern. Why are diamonds "3-" and as 13 cards are accounted for perhaps the "+" signs could be removed? Perhaps after the auto scan finds all strange descriptions for subsequent automated correction a second pass could represent the length holdings in the international accepted and standard sequence of S,H,D,C and remove all "+" signs where 13 cards are accounted for. This would represent a big improvement in the bid descriptions. If the Robot bidding algorithms then employed the bid newly improved and structured bid descriptions this might significantly and further improve accuracy in Robot bidding with relatively little additional programming effort. In business. ideas with little effort for potentially high reward are always worth investigating. What do you feel?

http://www.bridgebase.com/tools/handviewer.html?sn=calm01&s=SQJ54H85DAT2CKJ32&wn=Robot&w=SAT86HQ96D983CT86&nn=Robot&n=S9HKJT74DKQJ7CA95&en=Robot&e=SK732HA32D654CQ74&d=n&v=n&b=5&a=1H(Major%20suit%20opening%20--%205+%20H%3B%2011-21%20HCP%3B%2012-22%20total%20points)P1S(One%20over%20one%20--%204+%20S%3B%206+%20total%20points)P2D(Opener%20two%20rebid%20--%204+%20D%3B%205+%20H%3B%203-%20S%3B%2011+%20HCP%3B%2012-19%20total%20points)P2N(Balanced%20invite%20--%204-5%20S%3B%2010-12%20HCP%3B%20lik)P3S(4+%20D%3B%205+%20H%3B%203%20S%3B%2015-19%20total%20points)P3N(5-%20H%3B%204-5%20S%3B%2010-12%20HCP%3B%20likely%20stop%20in%20C)PPP&p=C6C5C4CJH5HQHKHAS2SJSAS9C8C9CQCKC2CTCAC7D7D4DTD3C3S6H4D5H8H9HJH2HTH3S4H6DJD6DAD8D2D9DQS3DKS7S5STH7SKSQS8 The !S3 bid is descibed as having 3S - patently untrue. When is the Robot going to be upgraded to speak something approximating its actual holding?

Yes, I know that GIB is said to bid like a good beginner, but ... When will GIB be taught to devalue hands with a clear misfit. When you look at others results to find out why you did so well on what seemed to be destined to be a poor result for you, you find that the Robot and several other humans have repeatedly outbid each other, sometimes to the 7 level, on a clear misfit. It is noticeable that this rarely happens with GIB partnering GIB. Surely GIB can be readily programmed to stop earlier when partnering human opponents on misfit hands. Regards, Ric.

GIB is consistently passing to leave opener in a 4-3 fit when a known 5-2 fit is available. For example opener opens 1S and rebids 2H over a 1NT response by GIB. GIB holding a hand with 6-9 point passes holding 2S and 3H. This is so easy to fix and is a critical skill GIB is currently missing that one wonders why it has not been fixed before.

If the fast GIB was free to compensate for its problems, I for one would pay more - namely a $1 a week for the slower better version.

To the GIB support team: I think that GIB is a great facility but is not without significant failings. When will GIB me programed to understand a GOSH? Support for understanding a good one suited hand is a critical omission and has been raised by more than one forum member. I feel the forum members deserve a clearly defined implementation date or a statement that it is not scheduled at all. A vague suggestion that it is on a long list is not an acceptable answer. A statement that little or no support is allocated is an acceptable answer as it lets us know where we stand. If removal of weak jump overcalls and replacement by strong jump overcalls is an acceptable compromise to forum members, I still feel we deserve an implementation date. Supporting a GOSH is not sufficient - we must all know that such support exists. Please actually do something to fix this critical omission in the GIB bidding. If you do not agree with my assessment of the priority for GOSH support, then poll the forum users for their priorities from a list of 6 or more or the more common requests I will be pleased to participate and abide by the majority views - but don't bother to poll us if you are not going to implement at least two of the top three suggestions. You will be judged by the forum participants, I feel sure, by your actions or non-actions and the quality of your follow-up and communication skills. Regards, calm01

Your Robot partner opens 1D and you respond 2C. Your partner rebids 2D with the description "5+D; 11-21 HCP; 12-22 HCP". The actual holding is: KJ86 9743 KQ5 QJ Fortunately the contract ended in Hearts, bit surely at least 4 good diamonds is required for the 2D rebod. What do you think?

Uday, Thanks. Please let us know what bugs are fixed in what release and when. The standard communication with users - like you get with a browser such as chrome or firefox ... Without knowing what has happened and when we forum users do not know if we can employ cue bids with safety. Calm01

I enjoy playing with 3 robots because I can take a phonecall or make a sandwich in the middle of a hand. Also learning to cope with its foibles is a good mental exercise and particularly useful for exercising my bridge logic muscles. There is one area that has stopped being much of a learning opportunity and become a source of much amusement - GIBs opening leads. It is often difficult to guess what GIB has led from and so what declarer has got because: - at NT GIB often likes to lead short suits in preference to a good 5 or 6 card suit,, - in a high level contract (5 or above) it likes to underlead side suit Aces, - based on probably my biased perception, GIB seems to choose suits to lead without any reference to partners bidding, - it will lead x from J10x to block partners suit! I accept all this experience of GIB choices of opening lead but can see several opportunities to dramatically improve GIB defending skills with little coding effort. For example - the lead from J10x - new bridge players are taught some kind of standard lead system to minimise damage from too much thinking in their early bridge career. GIB can readily be given lookup tables for standard opening leads - differentiating between NT and suit contracts. Lookup tables are mothers milk to computer systems. GIB desperately needs to start weaning in respect of opening leads.

Bbradley62, Thanks for correction - you are right.