xeno123

Now working again - guess it was the system crash.

Selecting Viewgraph deals as the deal source seems to be broken today. You get a message "Fetching Viewgraph Deals" then nothing happens. A couple of other suggestions: 1) Be really nice if when playing with 3 robots you could declare instead of GIB. 2) There are some pretty weird Viewgraph matches out there - how about if you add a 1-bit field indicating whether this is a serious match (Spingold etc.) and then we can ask for either random deals restricted to serious matches or can pick among the serious matches. A bit of manual work for you guys I know, but still would be very much appreciated. 3) Some sort of "intermediate" level robot rental - say $2/week for a some more simulations than the sometimes brain-dead basic GIB. Thanks again for all your work - it's really a great site and I know people whine more than they say thanks! Peter

http://tinyurl.com/9lc2y2m When South opened 2NT, GIB raised to 3NT. But when South (correctly in my view) opened 2♣ and rebid 2NT after GIB's 2♦, GIB then PASSED despite having 7 HCP and a 6-card suit. Clearly a bug.

Please!

Here GIB responds 3♠ to its partner's double of opener's pre-emptive 2♠. The bid is described as showing 13+ points. I'm not sure how best to describe the doubler's subsequent pass in response though. :) http://tinyurl.com/bmjql4a

My mistake - my commentary related to a different deal. In this hand, both expert Easts in fact bid over North's 1♥, and both E-W expert pairs ended up in 4♠ making.

I really appreciate the new feature that allows replay of viewgraph deals. At this point I've probably played about 200 deals with 3 (basic) robots as the other players. On average I'm clearly winning, which I guess means I must be (on average) better than basic GIB. (I view myself as a good intermediate player). It's interesting to examine the reason for the swings: 1. Most often a (positive for my side) swing happens because GIB doesn't overcall (or open weak hands) nearly as much as the experts do. So there are a lot more uncontested auctions. I think this is the biggest factor by far. Here is a typical recent example: http://tinyurl.com/6o538xg Both EW teams in the recent Euro teams bid and made 3NT over a 2♠ opening in one room and a 2♦ opening in the other. 2. GIB blunders in defense and less often in declarer play are the next reason. 3. On the negative swing side, the experts reach quite a few making slams that I and my GIB partner miss. GIB+GIB seems to synch a bit better at slam bidding than do I+GIB (but maybe not enough examples to know this for sure). Here's an example - maybe I (S) should have bid something other than 3NT or maybe GIB should have super-accepted? http://tinyurl.com/7975uqa (Actually only one of the other teams bid this laydown 6♥ (or 7 if no club lead), and seeing Fantoni and Nunes only bid to 5♥ after a long auction, maybe this slam is harder to reach than it looks). I'd be interested to know other (better) players' experience playing these viewgraph deals. Peter

GIB (advanced robot) as North twice declines to bid over opponents intervention when holding a 7 card ♥ suit and 7 HCP opposite partners 1♦ opening. http://tinyurl.com/85g5nod Probably I should have balanced here, but isn't GIB's bidding overly timid? (In European championship, both N/S teams ended up in a heart game over a 2♠ opening by East, making 5♥ and 4♥+1).

Yes, there are 11 top tricks at the outset, and it's trivial to always establish another ♣trick. It's basically a laydown. GIB just got its entries tangled up, but fixing that sort of error is much harder I suspect. (I assume the advanced robot would have no problem with this hand). Peter

I believe there is a simple heuristic that could avoid some of the basic robot's most egregious and irritating blunders. Here's an example of the problem - at trick 8 in a 6NT contract, having already lost one trick, GIB plays a guaranteed loser for down one, when what would have been a winning play (cashing the diamonds) was available: http://tinyurl.com/87zojnb I'm assuming what was going on here was that in all its simulations GIB saw one down, and so it picked one of its losing plays at random. A simple heuristic that says play a known winner in that situation instead of a known loser would have fixed this. An analogous (and even more irritating) situation happens in defense when GIB randomly discards a high card that would have been a winner. Here the heuristic should simply say that when choosing among discards with (apparently) equivalent outcomes in the simulation, discard the lowest of the available choices. If I am correct in my analysis here, these are two easy fixes that would reduce blood pressure among basic-robot users! Peter

FWIW, the two no-name viewgraph auctions both went: 1♦=P=1NT=2♥=P=P=3♣ (both down 3).

Admittedly my 1♦ opening was very aggressive, but then things really get away from me with my partner GIB's exceedingly weird 3♦ bid, which according to the description means 4+♦ over W's 2♦ balance. In fact we both had only 3 cards in diamonds, and needless to say the resulting contract was dismal seeing West's 2♦ bid showed at least 5 cards in that suit. I knew we were headed for disaster, so figured I'd pass and hope we weren't doubled. http://tinyurl.com/6ls8fnh Surely my partner should pass over West's 2♦ bid, or at very least try 3♣ if GIB was determined to bid.

East's 4♠ is described as a forcing cue-bid, but apparently nobody told West about that. http://tinyurl.com/85y97pv

But I really like the new "Viewgraph Games" ability. If I just want to play against a strong bot on my own computer, I can use Bridge Baron, whose play and bidding is very good. Peter

Given that the basic robot makes some really dumb plays on occasions (like randomly discarding the highest instead of lowest card in some suit at the cost of a trick), and the "advanced" robot is simply too expensive for those who want to play only a few hands a day, how about an "intermediate" robot that would cost say $2/week? It could be the basic robot that simply ran more simulations, so presumably a trivial change to make. I'm guessing that the really dumb plays are the result of some unlucky set of simulations, so upping the count would generally take care of that. Peter

In this deal, why did GIB bid 3♣ rather than 3♦ as a game try suit? GIB has a strong club holding and a weak diamond holding. I assumed this bid meant GIB needed help in the bid suit, and so my club void would be useful rather than duplication. http://tinyurl.com/8axf489 Or am I not understanding the meaning of this bid? Peter

Basic.

Thanks. Here's the diagram: [hv=sn=xeno123&s=SAK764H5DJ63CT865&wn=Robot&w=SQ2HQJT432DT75C42&nn=Robot&n=SJ9HK8DAKQ94CAQJ7&en=Robot&e=ST853HA976D82CK93&d=e&v=n&b=2&a=PP2H(Weak%20two%20bid%20--%206+%20H%3B%205-11%20HCP)D(3-5%20C%3B%203-5%20D%3B%202-%20H%3B%203-4%20S%3B%2013+%20total%20points)3H(3+%20H%3B%203-8%20total%20points)3S(Invitational%20--%204+%20S%3B%2011-%20HCP%3B%2010-12%20tot)P4D(rebiddable%20D%3B%2019+%20total%20points)P5D(3+%20D%3B%204+%20S%3B%2010-11%20total%20points)PPP&p=H9H5H4HKH8H6CTHTC4C7C9C8S5SAS2S9D6DTDAD8SJS3SKSQD3D7DKD2D4STDJD5C6C2CQCKH7S6H3DQD9C3S7HQCAS8C5HJCJHAS4H2]400|300|[/hv]

Can anyone explain GIB's play to tricks 2 and 3? http://tinyurl.com/6vppl3t If GIB ruffs the heart on trick 2 (instead of the mysterious club discard), then 12 or 13 tricks (depending on the club finesse) are there off the top. Instead GIB comes away with 10 tricks and down one. Thanks. (BTW, how do I post the diagram rather than a link? I tried posting the expanded url, but the line length seems too long for the forum to cope and so only parts of the full hand get posted)

If Attitude signals are turned on, does the program understand the notion of an encouraging or discouraging card from its partner? Peter

I see that the basic robot signals Count rather than Attitude. I submit that for the vast majority of robot users, Attitude would be more helpful. At least with an Attitude signal you can perhaps nudge the robot in the right direction - I have no great faith that it ends up doing anything useful with a Count signal. Peter

GIB has a beautiful hand, opens 1♠, and then passes my supposedly forcing 1NT rebid! (I was not a passed hand). I could have 3 spades and 11 points, right? Not sure if 1NT or 2♦ was the right bid by me, but I'm pretty sure GIB's pass was wrong: http://tinyurl.com/8x9y4yp

Ah - wasn't aware of that, being new here, and not seeing it discussed anywhere obvious in these forums. Maybe it should be in the FAQ or pinned to the top to avoid further grief from Newbies like me. (But I remain unconvinced that is that hard a change given the obvious expertise displayed in the system interface here).

When playing with three robots, how about an option for the human to be declarer for all contracts declared by N as well as contracts by S? There's not a lot of fun (or really anything to be learned) watching a robot play (or sometimes massacre) your contract. This option would certainly reduce the luck involved in playing with 3 robots. Most bridge programs I know of rotate hands to allow this, but that wouldn't be necessary.

At trick 7 in the following hand, GIB elected to finesse in spades (going down 3) instead of just going up with the ace and simply taking the rest of the tricks (AS, QC and the 5 high diamonds), so making 3NT irrespective of the distribution: http://tinyurl.com/44nlucd (I realize the contract is unmakeable if W holds up the diamond King so the end result is not that unfair, but that is a separate issue). Peter