calm01

http://tinyurl.com/6vybxxv

http://tinyurl.com/89jdmbu

http://tinyurl.com/6o2vz2t The simulations that that GIB employed to determine that 4S is better than 3NT must be somewhat warped. What is going on in the program? Surely it is possible to trace its path to this unusual decision.

Apparently simulations do not include defeating the contract!

http://tinyurl.com/7rufkb2 This is a fundamental failure that occurs frequently and so is a high priority fix.

Barmar, Thanks for the prompt reply and the effort involved in looking again. Yes you did explain before about the Queen. Overall the answer is till unsatisfactory as: 1) the assumption that declarer plays double dummy is often false, 2) as you suggest there is more involved than just this false assumption, 3) it is a frequently recurring poor defence by GIB, 4) there is a bug which needs to be addressed if GIB is to stop frequently losing tricks when defending. Perhaps it is time to raise the priority of this issue in the list of bugs to be fixed - what do you feel?

Well found Bbradley62. GIB seems to like no-win defensive plays - GIB almost seems to specialise in the no-win 'technique'. GIB frequently employs another no-win defensive play. I call this particular unnecessary defensive play of King under dummies AQ "Group Captain" syndrome. This is because in the second world war apparently pilots of all nations flying for the Royal Air Force would see their death as nearly certain/imminent and deliberately fly their fragile plane into a mountain as this was certain death. In this way the continuing worry of when their life would be lost was resolved! When the King has gone, the unnecssary play of the Q under dummies AJ is a similar 'tactic' favoured by GIB. As a defender you rarely know if declarer just wants an entry to dummy and does not intend to take the finesse or cannot risk the loss of the finesse. Human player often entertain the Group Captain syndrome for the the pleasure of certain loss but I doubt whether GIB has similar psychological problems! Please fix this all too frequent bug in GIB defenses.

I appreciate the clear reply. In that case more simulations are required to enable the possibility of a trump promotion or over-ruff to emerge and to be considered.

GIB often appears to give partner a ruff. GIB often appears to abandon giving partner a ruff when one of the opponents can also ruff. If true this means GIB is unlikley to give partner a trump promotions or even an over-ruff. Does GIB try for over-ruffs and/or trump promotions?

I suspect 2S would be to play, 3S would be a six card suit, possibly non-forcing, 2NT followed by 3S may also show 5C and 4 spades, a direct 4S gives up on a possible club or spade slam so the North hand has a difficult bidding problem. With the bidding opponent likely to have the red suits, a black suit slam is a reasonable possibility if partner has few wasted points in diamonds but even then may we subject to bad breaks including a heart lead and a heart ruff. If I held the North hand I would try 3H to hear what partner has to say and go for 4S if partner does not mention a black suit. This must be a slam invitation as I could have bid it directly over 2H. But it is sometimes tough out there! Even a GIB with good bidding principles on board could not be reasonably expected to cope well with this difficult hand unless I have overlooked something obvious to others.

There seem to be three problems in the Robots bidding here: - when GIB mentions it use of 8421 rather than 4321 I always underbid my hand as it is appears to be the Robots way of saying "I do not know how to value this hand properly and I tend to overbid in these situations.", a) when GIB mentions total points in practice it seems that it is the Robots way of saying "I am in punt mode.", b) when a human partner raises a major to the five level it is meaningless unless there is a prior agreement - I always go with partners viewpoint which has included: * bid six if your trumps are better than you have promised, * I have the trumps covered - is your playing strength better than you promised, * grand slam enquiry - bid a side suit feature (as agreed), * if you have a lower suit void bid it or 5N for a higher suit void so I can judge if we have wasted values for the grand, * ... when GIB employs a raise of a major to the five level I have yet to devise any consistent pattern. But you may know better.

http://tinyurl.com/885vkwo What are the rules for GIB opening leads? Should they be modified to suggest, with no strong indication from the bidding to lead otherwise, an Ace from an AK side suit holding against a suit contract?

bbradley62, Thanks for the helpful update on the statistics of the hand. I usually agree with your assessments but not on this hand. A decision to lead partners suit comes before the result not after the result. So it is only a matter of post-mortem in the sense that the player that cost 2070 points for a failure to lead partners suit without a good reason might end up dead at the hands of the team-mates at the team review bridge table! While this particular failure was by a human player, GIB will never be a good partner until consideration is given by GIB to leading partners suit. This is a high priority for fixing because the issue has such a high frequency of occurrence. Recently 2C was opened on my left, partner passed and right hand opponent bid 2D. I doubled and the description of my double was showing rebiddable diamonds. GIB still failed to lead diamonds while having no good reason for any other lead. If GIB was human most of his/her partners who could not see the funny side would have even less hair left than I have! LOL. On the question of 6NT being superior to 6H in case partner was 4045: - if partner was 4045, partner would not bid 3NT, - if partner has a secondary stop in NT - say Q 10 9 x, 6NT would be one down when the likely Ace and then King of diamonds were led. My contention that 6H is better than 6NT is now perhaps clearer.

You have to have a good reason not to lead partners suit and South does not have a good reason and the failure to lead partners suit cost 2070. While this failure was by a human player, GIB will never be a good partner until consideration is given by GIB to leading partners suit. Recently I doubled a 2D response on my right to a 2C opener on my left. The description of my double was showing rebiddable diamonds. GIB still failed to lead diamonds while having no good reason for any other lead. If GIB was human most of his/her partners would have little hair left! LOL. On the question of 6NT being superior to 6H in case parter is 4045:

http://tinyurl.com/736k7rq This hand raises several issues: - the description of GIB East 3NT bid bears no relationship to its holding, - the actual GIB East holding bears no relationship to a hand wanting to play in 3NT, - based on the description of partners 3NT bid, why did GIB West bid 6N and not 6H, - six of seven GIB Easts seemed to bid the same way - only one found a better bid, - lead partners suit (even when partner is a robot) or it might cost a swing of 2070 points. As always still laughing - the alternative is crying.

What is the bridge principle/rules/lookup tables that underlies this bidding failure by GIB West? With 2S non-forcing responder has many fewer forward-going bids with which to locate the best suit/NT contract at game or slam level - and will sometimes pay a high price for failing to find the best suit and or level. In other words if 2S is non-forcing, a weak 2H opening can become a pre-empt of your side too. Just as it makes sense to support with support, it also makes sense to subside in a misfit - you want the opponents to declare in a misfit. So taking out a weak 2 is only done with tolerance for hearts or there is near game or better in responders hand despite a misfit. Either way 2S must be forcing. The choice of 2S is perhaps near the cusp in this hand - but perhaps it can just stand 3H non-vulnerable but game is unlikely. But openers subsequent pass makes no sense if 2S is forcing. A forcing 2S says if you like spades (3+) raise, or return to 3H as I can stand this or better. 2NT is also acceptable rebid showing precisely 2S (or one of AKQ of spades) and upper point range for the original 2H bid. 3 of a minor says I am upper range with a feature(/4 carder according to partnership agreement). You will note that all these bids are constructive and cooperative communications between bridge-savvy partners. Palookas often engage in unilateral bidding based upon "I can see my hand and I don't like your bid partner" or "I am minimum and ashamed of my bid and so will pass your forcing bid partner" and so treat this kind of bid to be non-forcing until they experience several expensive results in what are likely to be misfit hands. Some of course never learn. GIB East has its bid but GIB West seems to be the palooka of the partnership treating 2S as non-forcing.

Barmar, Thankyou for the effort you obviously to put in before developing your reply. The words "the bidding rule says to bid the longest suit, and the highest ranked of equals, so it prefers majors" is I feel sure a true reflection of GIBS code - I trust you. Just by merit of its truth does not make it anywhere approaching an optimal bidding rule in the situation GIB East finds itself. The existing rule applied by GIB, as described by you, clearly makes poor bridge sense. Please change it to based on something like "Partner, this is my longest 5+ suit or failing that my lowest ranking 4 carder. Feel free to bid another suit assuming I have at least 3 card support or perhaps just pass if happy or if not doubled or rescued by the opponents as we are in scramble mode here." A rule and an underlying rationale such as this is much more likely to get us out of trouble/be rescued on the majority of occasions. A better rule might be to bid 1NT requesting partner to choose the best landing spot (to include 1NT), and a optional redouble of 1NT for penalties with a good 19 if partner has chosen 1NT. But such a rule takes a level of bridge understanding between regular human partners it is unreasonable to expect from a computer program, even one partnering a copy of itself. calm01

BBO continues to innovate successfully in so many areas of its already impressive site. Why would BBO not want to similarly build on its achievements in Artificial Intelligence (AI) at low cost and low/medium risk? Initial low cost enhancements to GIB with a high return on investment in AI could include: - add many times more simulations, - include passes in the process of building a picture of partners hand, - have different rules for suit selection for initial lead against NT and suit contracts, - particularly add a rule about preferring long suit leads at NT, - add opening/subsequent lead look-up tables, - add simple rule about considering leading partners suit unless having a better lead, - publish full details for each new release/version of what has been added/removed/fixed, - publish wish lists sourced by both BBO/users, - replace any separate expensive to maintain bid descriptions with what GIB actually understands by the bids so avoiding conflict and confusion and maintenance costs, - note discrepancies between the new generated descriptions and bridge sense to highlight patterns of errors and thus promising areas for further enhancement with a high return on investment, - ... In a years time BBO will owna a steadily improvig GIB to be even more proud of, knowing even more significant improvements are just a low investment away. And as the underlying performance is enhanced, the simulation count could be reduced. calm01

Before fixing GIB so it will bid hearts on this hand... There are so many bidding sequences possible in a bridge auction, let alone the huge combination of bridge hands held with any one auction, a bridge software program needs to be based on either: - sound bridge principles within which each of the near unlimited combination of auctions and hand holdings must be assessed, or - resort to brute force - put so much computing power behind simulations and/or look-up procedures so that bidding principles are not required. At the moment we seem to be caught between these two approaches with the results we all see in mostly wasted BBO effort and a poor GIB. It is clear from opening leads that few effective principles are employed (as opposed to exist) and lead look-up tables are rarely used (as opposed to exist). Please go for one of the two approaches not a poorly implemented fudge of both as we appear to have now. Yes. you may have only one horse to flog, but try dramatically increasing the simulation numbers - it may solve several classes of GIBs problems some of which you are wasting programming effort to partially fix and causing other issues in the process. But BBO will have to admit you have been wrong all these years. Most substantial moves forward require getting past denial that change is both desirable and indeed possible. calm01

Perhaps 3 time the current number of situations will give a decent base bridge robot that is worth improving and will actually bid hearts on this hand. There are so many bidding sequences possible in an auction, a bridge software program needs to be based on either: sound bridge principles within each of the near unlimited combination of auctions and hand holdings must be assessed, The altrnative need to brute force - put so much computing power behind simulations and/or lookup procedures that bidding priciples are not required. At the moment we seem to be caught between two approaches with the results we all see. calm01

http://tinyurl.com/corl7t8 In this hand the bidding by GIB East really scrapes the barrel. One of the joys of bridge is that it is about competing evaluations. And my evaluation could be wrong in the long term but work on this hand or vice versa. But sometimes a bid is so bad and so unlikely be the best bid, it lies clearly in the category of ugly. The double of 1!S by GIB East is fine but the self-rescue into hearts fits the ugly category. What is wrong with 2!C - it gives partner more choice at the 2 level and the opponents more chance of rescuing you. 1NT also gives the opponents an increased the chance of rescuing you by bidding on and you are no worse if they double you in INT as partner has another chance and a third chance when you redouble to find the best suit. 2!H is just a terrible choice - only 3!H or higher would be worse. Not even a palooka would make this mistake - more likely leaving the contract in 1S XX. This is not a bug it is a complete failure of bridge logic which is likely to cost so much in one deal as probably ensuring the loss of a tournament. Please fix this bug and any underlying basis for the logic (if any) that was used by GIB. If it is a case of simulation error - just increase the number of hands simulated by a factor of 3 or more for lower versions of GIB. A small delay in GIB responding is a small price to pay for sensible bidding. And with servers so cheap, BBO can readily afford an extra server or two if required as this is so much cheaper than wasting programming effort tinkering with such a poor base product. calm01

barmar said: "Unfortunately, none of us are AI programmers. We're going to have to tread very carefully if we're thinking of changing basic design of GIB. Also, there's an enormous rulebase that's tied to its current evaluation methods. If we make a radical change to the GIB engine, we'll have to go through thousands of rules, adjusting them to work with the new evaluation method." Then it is perhaps simpler and cheaper in the long run to purchase/licence another bridge playing and bidding software program that works on sensible principles not simulations. This would be easier and less expensive to enhance and fix as a result. And the starting point would be better than where we are now. May God have mercy on the soul of GIB/BBOL while you continue to flog a dead horse. calm01

cloa513 said: "Its favourite is total points and also uses HCP." If true, there is little hope for GIB's bidding being substantially improved until its favourite is changed to a better system. Any programming effort would be tinkering with a poor evaluation system until total points is dropped. My experience teaching beginners is that they soon grasp and love HCP and get so attached it is difficult to get them to downgrade Quacky hands (Queen and Jacks with not Aces and Kings). Onece HCP is understood and employed sensibly, losing trick count seems the simplest approach that leads to the fewest poor evaluation errors by beginners. Yes I know adjustment for no Aces or 3/4 Aces is ideally required but this is often too sophisticated for most beginners and only confuses them. Please, unless you want to continue to waste much of the BBOL programming effort on GIB, replace its favourite total points system with another evaluation system. Note that even a simplistic losing trick count would give (on the assumption of an average 2 spades in partners hand) a losing trick count of zero and 7 spades would be easy to find even by a software program. Please stop flogging a dead horse (total points). Yes, I, know admit that BBOL would have to admit it has been wrong all these years, but the alternative of an expensive to maintain GIB (in terms of return on investment) is the alternate approach. Losing trick count is not for Christmas (freaky hands) - it is an everyday tool for 8 card fit hands that works here too. Calm01

An error of hand evaluation of one trick between partners could be put down to a matter of opinion. A three trick error suggests that hand evaluation has just one wheel left with the stagecoach missing. Total points can be a poor evaluation method in the hands of a reasonable human player and perhaps readily becomes almost meaningless in a computer program such as GIB. Most beginner (human) players would realise that AKQJx has better length points than 98653 but does GIB? If you ever experienced unilateral biddng when playing with a beginner, you can relate to this. One definition (I feel sure it is not the only one) for total points that can be found on the internet is: "Total Points The sum of high card points (HCP) and one of the shape points (long suit or short suit point) is called total points (TP). When you are opening a bid or bidding a new suit you must evaluate your hand with HCP & long suit points. So, TOTAL POINTS= HIGHCARD POINTS + LONG SUIT POINTS" This definition is incomplete (how are long suit or short suit points counted?) and is not self-consistent as short suit points suddenly get overlooked at the conclusion. Almost any recognised method will be better than the existing GIBs use of total points. If you doubt this just look at when GIB raises a major from 4 to the 5 level and (ignore its description) its variations for the same total points (zero to 2 extra tricks range compared to prior promises and holding between 0 to 4 key cards). Why not abandon total points which GIB cannot seem to handle effectively and begin to rely on a more accurate and self-consistent technique of hand evaluation. There are so many better techniques to select from such as: high card points especially for opening, simple responses and NT contracts, fit/misfit adjustments, consideration of likely blockages or good communications between a possible dummy and declarer, playing strength adjustments, entry considerations on weak hands with one very long weak suit, losing trick count for known 8+ card fits where a suit contract is anticipated, adjustment for hands holding 3 or 4 aces, ... We will all use one more and have our own favourites. Simply give GIB one 'favourite' and apply it sensibly and consistently in the framework of prior promised values, steadily making improvements where appropriate, and slowly, GIB will bid better and better. Only then add a secondary evaluation string to its bow. But please, if you wish GIB to ever bid well, abandon its misuse of total points. Calm01

http://tinyurl.com/3kpc2r6 The same hand - this time a query on GIB leads. Surely the best lead from this club holding is the 10 not the 6 both at trumps and NT. It cannot be difficult to program standard leads based on a lead table having selected the best suit to lead - so maybe the standard lead tables are amiss or just missing. This naturally raises the suit selected by GIB as the suit to lead: - all too frequently GIB leads its shortest suit at NT even when it holds a likely majority of the likely defensive points. - GIB rarely leads partners bid suit even when it has no decent lead of its own. What rules, if any does GIB have for opening leads? If just a random selection of hands is generated to determine the best lead, then this would explain the poor choices GIB repeatedly makes. Even if this random selection of hands must be the starting point, surely some allowance for long suit leads at NT might get a look in, using standard lead tables would not go amiss, and leading partners suit might be given some consideration. Still, when improvements ensure GIB leads at last approach palooka status, the fun of discerning GIB's unfolding holdings from it opening leads would disappear. Calm01