The Tao of Gaming

Boardgames and lesser pursuits

Posts Tagged ‘Bridge


Tired of working on my latest Factory, I fire up the robots for a few hands of bridge solitaire. I pick up the sort of hand you get in solitaire…

S:KQ7xxx H:A D:KQ62 C:xx 

(You get them because the computer deals out the hands, then swaps your hand with the best hand if you don’t happen to hold it).

Suprisingly, your robot partner opens 1 Heart. I bid 1 Spade and CHO bids 1NT. I bid 2 Diamonds (natural, forcing) and the CHObot admits to having two spades. I suspect four spades is cold and six spades is odds against, but I’m playing against robots and just goofing off. We’ll be light on points but maybe hearts will run (and I do have a sixth trump). Six spades it is.

I get the five of clubs lead and I don’t have time to wonder if I’ll be down before I won a trick because dummy is already tabled.

Dummy S:A9x H:KT9754 D:xx  C:Ax

Hand  S:KQ7xxx H:A D:KQ62 C:xx 

(Club 5 led by LHO)

That’s my type of opening … minimum. Well, I fly with the club ace and think. This is a decided underdog. Do I have any lines of play? In fact, I do see one. The nice thing about slam hands is that you usually don’t have a tangle of possible lines.

I win the club ace (naturally) and lead a heart to the ace. (Technically I should probably lead a spade the king and then cash the heart ace, but that’s a minor improvement and If hearts are 7-0 I think I would have heard something). I lead the spade king and then a spade to the ace, as spades break 2-2.

I needed that.

Next comes the heart king, pitching my losing club as LHO shows out. Uh, ok. I lead a diamond and it goes ten-queen-small. It looks like the diamond ace is onside.

I needed that, too.

But I also need RHO to have started with the ace and shortness (A, Ax, or Axx).

Nothing to do but lead a small diamond. LHO wins the 9 (RHO completing an echo!) and leads a club. I ruff the lub, ruff a diamond and … RHO plays the ace.

Score it up.

Once the diamond queen won, my odds were pretty good. RHO was known to have six hearts and two spades. Not much space for four diamonds and if he did have four, LHO had 8 clubs. I definitely think that would have been mentioned.

I also belatedly realize I missed another improvement.

If an honor had fallen on the second round of hearts, I would have had options. If RHO plays the honor I can play for him having QJx tight. If LHO shows up with an honor I could play for QJx tight or Hx. In the latter case I can run the HT for a ruffing finesse through RHO and (assuming it wins) get back to dummy with with a diamond ruff after losing the diamond ace and winning the other diamond, and pitch my last diamond on dummy’s good 9 of hearts.

I hadn’t considered it, but if the honor had popped up, I probably would have noticed. A minor demerit for not thinking about it ahead of time, but it turns out I had just enough luck on this hand.


Written by taogaming

September 24, 2017 at 9:25 am

Posted in Artificial Opponents, Bridge

Tagged with

On the same wavelength

Playing a club game with Hank, I am off my game (I blame anti-histamines and not having played in roughly a month). But due to the movement we skip the only difficult pair, and we’ve received lots of gifts (including the following auction (1D)-1H-(1S) – P – (6S) when I’m holding five spades to the ten.

So I’m in a good mood when I pick up the following collection:

S:QTx H:A D:KQT C:AK9742

It gets even better when Hank opens One Diamond in second seat. Our opponents are silent throughout.

I have an easy 2 Clubs bid, so I make it. (Despite playing 2/1, we don’t play this as game forcing).

Hank bids 3 Clubs. We have two ways to raise clubs, and this is the weaker one. I could cue bid, or even just sign off in Three NT, but slam would have play opposite some 7 point hands (Kxx xx Axxx xxxx). We may be off two fast spade tricks, and I could cue bid three hearts and then pull 3N to 4C, but I decide to not advertise my weakness and just check for key cards. I bid 4 Diamonds.

Hank bids 4 Hearts, showing one ‘ace.’

Since I’m looking at the 5th ‘ace’ (the king of clubs), I know it is an actual ace.

I want to check on the club queen so I bid the next step 4 Spades.

Hank bids 5 Clubs, denying the queen. It occurs to me that in this one instance perhaps we should play 4 NT denies the queen (instead of returning to the suit) because I could have passed that.

If Hank doesn’t have the queen of clubs, how many does he have? I have an inference in that Hank could have also bid 2 diamonds over 2 Clubs, which just shows five diamonds (and is forcing one round) or 2 hearts, which shows a balanced hand. So if he only had 3 clubs to the jack, he had the option of not raising, immediately. As compared to that, 3 clubs shows a weak hand and two diamonds is wide.

As I consider this, I also realize that 5 Clubs is going to not be a great scoring contract. 3 N will probably make 4 for +630 (or +660 if it makes 5) and even if I play in 5 Clubs, I’m only getting +600. If Hank does only have three clubs without the queen I’ll need the clubs to break, but that also means that Hank is likely to have the Spade King. In short, Five clubs making exactly isn’t a great score, so even though I think I’m an underdog I bid 6 clubs.

(Funny auction. I bid 4S to check on the queen, but apparently I was going no matter what).

This gets passed around and LHO leads the diamond jack and Hank puts down:

S:Kx H:Kxx D:A8xxx C:Jxx

That is, in fact, a minimum, but at least it has the Spade King, so I have a play. I win the Diamond King and lay down the club ace. Do I get a 2-2 split? Do I get the stiff queen?

I get the rail.

RHO discards the three of spades. Looks like I’m off one, but worse contracts have made. Let’s put LHO to the test and force her to decide what to do. I lead a small club towards the jack and she … ducks. Huh. I win the jack, come back to the king and let LHO win her club.

RHO has been decidedly unhelpful in his pitches if he has the spade ace, playing up the line for the first three and then discarding a heart. If LHO is looking at the spade ace she’ll cash it, but for all LHO knows I may have the SA and be missing the HA.

LHO plays a small spade and RHO wins his ace and I claim the rest. Down one is not unjust, but I’m vaguely annoyed that she guessed right, because it was a guess as far as I can tell.

(Actually, later on I notice that RHO had four diamonds to the 9, so it was hopeless as long as he never pitched one).

The post-mortem is amusing. Hank’s first comment is “You know, 4N should deny the queen, but that’s not our system….” which is exactly what I had been thinking. He also was torn between 2 diamonds and raising to 3 clubs. We decide that raising to 3C can be done with only three clubs, but they should include the queen (or better).

I was still right to bid the slam (once I had decided to ask). It only took a 2-2 clubs split (which is over 40%) or the stiff queen (which is 1/4 of 50%, so 12.5%) so about 50/50 (dropping a few percent for a ruff on opening lead or ace of spades, ruff). As the only pair to bid it, we got a zero, but five clubs would only score a ‘1’.

I was getting 5:1 odds for an even money proposition. I would have been right if Hank’s clubs were Txx or worse, which then means I need the 2-2 break.

At least we were both on the same wavelength every step of the way.

Written by taogaming

August 28, 2016 at 9:36 pm

Posted in Bridge, Session Reports

Tagged with

Computer Bridge

There is currently a bridgewinners discussion on “When will computers beat human bridge experts?“. This is (unsurprisingly) triggered by the recent advances in Go playing computers, based on the deep learning system. The news from Google — taking time out of their military robotics schemes to focus on less Skynet-y ventures — was an interesting demonstration. My only expertise in this (apart from the fact that I’m not exactly a stranger to military robotics programs, but also medical robotics!) is that I’ve followed computer opponents in classic games somewhat.

There are three salient points to the system — the training method, the use of monte carlo systems in evaluation, and the hybrid engine.  For now, lets just consider a simplified bridge AI. It plays standard american, and expects its opponents to do the same. Teaching a program to handle multiple bidding systems is one of scale and scope, and not that different (in practice).

Training — The Go program was trained with 30 million expert positions, then played against itself to bootstrap. This method could be used with bridge, assuming a large enough corpus of expert deals exists. However, there are some issues.

Every go (and chess) program starts from the same board position, a fact that isn’t true of Bridge. To counter balance that the search space for an individual deal is much much smaller. Still, it’s not clear that 30 million deals is enough. Presumably you could use some non-expert deals for bidding (take random BBO hands and if enough people bid them the same way, that’s probably good enough). Top level deals can be entered, especially those with auctions duplicated at two tables.

Card play could use a similar method — for a hand and auction, if the opening lead is standard, you could assume (absent further training) that it is right. A clever AI programmer could have a program running on BBO playing hands, and then comparing it’s results (already scored, no less!) against others. Your scoring system may want to account for weird results (getting to good slams that fail on hideous breaks, etc), but that’s pretty simple.

So, there may be a problem getting enough expert deals, but there should be enough to get a large corpus of good deals (particularly if the engine weights others and then uses better players as a benchmark).

Randomness — Some people on the BW thread are saying that randomness will stop an AI.


The news out of Google is ahead of schedule, but it didn’t surprise me as much as Crazy Stone (the precursor to Alpha Go). Crazy Stone’s innovation was that if it couldn’t decide between two moves (because they were strategic, not tactical, or if the search depth got too great), it would simply play a few hundred random games from each position, and pick the move that scores better. Adding randomness to the evaluation function (of a non-random game!) greatly improved the structure, so much so that I believe I commented on it at the time. (Sadly, that was before the move, so I don’t have a tagged post. See my posts tagged go for some tangential comments.

Randomizing bridge hands would present different challenges, but the idea of just saying, “I don’t know, let’s just try each lead a few hundred times against random hands (that match with what we expect” is obvious, as well as using randomness (to decide whether to continue or shift suits). Because bridge doesn’t have Go’s massive search depth, you could also drop each hand into a single dummy solver for each position, or have it play randomly only until breaks are none (so plays randomly but not with a known position).

The thing about random play is that it’s fast. So you’ve won the opening lead, what to play? Whip up 100 random deals (not hard since you can see two hands plus a few other cards, plus all your bidding inferences) and try them out.

Hybridization — The trick is that you only resort to randomness if your trained algorithm isn’t confident of its training. This happens quite a bit in Go. (Go is amazingly frustrating in that expert or even master level players will be unable to communicate why a play is correct. I remember a lecture at the Pittsburgh Go Association and the lecturer, an amatuer 3 dan or so, was reviewing a game between two pros. And someone asked “Why did so-and-so play that move on that spot. Isn’t one space to the right better?”

Neither move had a tactical flaw, and the lecturer stumbled, then called out to a late arrival (a graduate student from Japan and — I believe — soon to turn Pro after getting his degree). The arrival went up to the big magnetic board, stared, said “Ah! It’s because of” and then laid out 10 moves for each side. Then reset, shifted the stone, and laid out ten different moves for each side then walked the few people who could understand the differences through it.

The point of my story? Go is hard. Go is hard enough so that the professional players routinely make moves that  amateur experts cannot reasonably understand. Go experts can look much farther ahead than bridge players (and computers) — yet random simulation coupled with deep learning can handle it.

The Go playing program might very well have learned to play the move on the correct spot, and not one-to-the-right, in our example. How did it learn this? Because the experts did it. It gained a feel for what to do in those situations. But even assuming that it hadn’t learned, and was sitting in the back of the room (like a 20 year old me) and couldn’t see a difference between the two. It might still grope its way to the correct move using a Monte Carlo simulation on both moves. (This is assuming that it’s near term tactical engine couldn’t find both sequences and judge one obviously better).

Right now bridge computers have many advantages, and can play perfectly once enough is known about the hand. You’d never use a random engine at that point. This hybridized strategy would be for your master solver’s club type things where experts disagree.

And, if you are deciding between those two things, you are (by definition) an expert.

So, I stand of the opinion that Bridge hasn’t been solved because nobody has thought to attack it. Or perhaps there is not a large enough body of expert deals that can be conveniently fed into a computer. A clever programmer (which I am not) could probably have a system learn just by having it log onto BBO, assuming that it could learn which players to trust and which to not (and which ones to use as bidding examples). 30 Million deals, each played 4 times by experts may not be enough, but it’s probably in the ballpark.

Why hasn’t this been done? Probably nobody cares. Go is (by far) the sexiest game right now because it’s search space is unfathomably deep. Go players routinely scoff at the simplicity (by comparison) of chess. In terms of search space (for a single hand) bridge doesn’t compare. If Google put its money behind it, I think a Bridge computer would do well in a match against a top team. Also, there were prizes offered for Go programs that could play at a high enough level, which spurred on development over the last 20 years.

Written by taogaming

February 7, 2016 at 12:05 am

Posted in Artificial Opponents, Bridge

Tagged with , ,

Why I’ve got little to say

I’m playing the same games over and over. Right now I have five Quarters for the year, and only one of them (Coup) is short (and I have almost 50 plays of that). Netrunner is fairly fast (~20 minutes) but  I’m over 80 games so far. The others are Bridge (2-3 hours per session), Sentinels of the Multiverse & Mage Knight,

At my current rate, The City is likely to make it to 25+ games but that’s it.

And none of the newer games have really captured my attention (except Federation Commander).

I suppose at some point I could write a few thousand words about The City, but honestly I’m not sure that anything I say would be non-obvious. I do believe that fountains win slightly more than any of the alternate strategies, possibly because of the number of synergy cards.

But that’s all I’ve got right now.

Actually, while I’m rambling I’ve slightly burned out on Bridge, and have cut back my number of club games in the last month or two. It was bound to happen after ~5 years of pretty constant play. Having a terrible regional didn’t help.

Also, while I’m thinking about it. Congrats to David Grainger on his first (US) National Victory. (He’s won a Canadian National event). I met David at the Gathering this spring, and he was at the San Antonio Regional, where his team unsurprisingly destroyed mine in the evening Knock Outs.

Written by taogaming

August 17, 2013 at 11:13 am

Musings on Bridge Bidding

(This is probably not of interest to non-bridge players. Tough. Learn to play).

Over the course of the years, I’ve read quite a few bidding systems. This week I read Marshall Miles book “The Unbalanced Diamond.” It’s interesting, particularly the aforementioned diamond bids.

Standard American (at least, the 2/1 variety that most experienced tournament players use) actually handles things quite well, but it’s interesting in that there is only one strong forcing opening (2 Clubs).  Precision (and other Big Club systems) also only have a single strong forcing opening, but it has to handle moderately strong hands (16+) as well as the huge hands that 2 clubs openings cover.

[I’m ignoring the 2 No Trump opening which shows 20-21 in most systems, and any openings above 3 Spades. Some tournament players tack on some openings to show a long major suit that can almost make game (The Namyats convention) OR 3NT to show a long running minor. And some hands can just open Blackwood, to ask about aces].

So what’s interesting to me is seeing systems that have multiple ‘big opening’ hands.  The Unbalanced Diamond actually has three big opening hands, 1C usually shows 15-19 (strong, but not overwhelming hands) but can include a few well defined hands that show more. (24-27 point balanced hands, and game forcing hands with both majors), 2 Clubs shows 20+ hands with a five card major (or two), 2 Diamonds shows 20+ hands without a five card major (excpet hands that are exactly 22-23, with 5332 distribution).

In the old days, of course, any 2 bid was strong. Four strong opening bids (for 20+ point hands). The problem is that these hands are pretty rare. Better to condense all strong bids into 2C (and use the rest for something else, typically pre-empting).

Bidding requires space — consider it an auction from 1-35, where each number affects play (and scoring). There’s an argument to bidding “1” say “I have a great hand” like precision does, but then your opponents will typically use their bids to quickly reach the highest safe number possible without worrying that they could have a bonus available. (Your bid has told them they probably can’t). Which may not be that high, but it’s often around 9-13. So precision’s advantage isn’t the big club. It’s when you open something else and get all those negative inferences. When a precision player opens 1 Spade, you often don’t have to worry about whether you can make a slam and bid slowly, you just bid your game. (Sure, you may rarely miss a slam with a perfect fit, but you’ll probably wind up making enough games because the defenders have no information that a revealing auction would have). Also, when a precision player opens one spade, the opponents compete, and then he bids at the three or four level, his partner knows that he’s doing it on offensive strength, not just high card points.

So in some sense, the strength of precision is every time you don’t use the big opening. I’m intrigued by the Polish Club systems, (where 1C shows all the big hands, but also shows the “balanced, barely enough to bid” hand, which puts some uncertainty for the opponents.  If they have a “compete quickly based on shape” system, often they could miss a reasonable game, or sacrifice when the opener can’t make much of anything. Or they could bid reasonably and then discover opener does have a huge hand.

Miles system works the opposite way. The 1 Club is usually moderately big, so the partner of opener will often be able to quickly determine that the opponents have gotten too high, but there’s no shot at slam (or game) and be able to punish them. Opener’s tightly constrained hand (almost all opening bids in this system have a 5 point range, except for the few exceptions on 1C (which are rare) and the 2C/2D bids technically have a 17 point range, but effectively 5 points.  (Only 1.5% of the hands have 20+ HCP, and only 0.02% have 26+)

Anyway, there’s very little discussion online about this system (at least, not that google can see), although I know that sometimes commenter JeffG plays with Marshall Miles (and in fact they were on the same team for the Vanderbilt’s a few weeks ago). The ACBL regulates bidding at tournaments somewhat draconianly (although I understand the reason behind it, it does annoy me), but this system is (apparently) legal. I guess I enjoy tinkering with systems, just like as in other games.

Anyway, in my mind, the unbalanced diamond compares with Precision (since they are both “Big Club” systems) and seems to compare reasonably well. It’s certainly interesting.


Written by taogaming

March 28, 2010 at 5:08 pm

Posted in Bridge

Tagged with

Recent Rumblings

Regarding El Capitan, the graphics are slightly awkward, the game play is reasonable. I’ll need to try it with the expansions. No major change in my opinions.

I got to try Ad Astra; with very low expectations. It was fast … (3 player game), I’m told that it dawdles with more, but with fewer players the score cards cycle more (and each player is more likely to win the “first place” bonus), so the game flew by. Honestly, a bit too quickly to make a judgement, other than “Don’t let someone get in the lead in two categories, as he’ll play all his scoring cards each turn.”

Silk Road — Meh.

Airships — Inoffensive. Not quite willing to purchase a copy, even at 50% off. Makes me wish that the To Court the King expansion saw the light of day…

Campaign Manager — Broke the dime mark. The “we’re waiting for more people” game of choice these days. I need to work on my expansion cards.

Bridge — Won a few small club games recently with a good partner, and got a good team for the next sectional’s Knock Outs. Which is important (my first tournament I’ve had a reasonable team in a decade or so). Almost won the self-declared 8 board championship of the Universe last night (self-sponsored), but our touchy slam had a bad trump break. Thinking a lot about bridge ….

Written by taogaming

March 25, 2010 at 7:48 pm

Some tips for beginners

This is not part of my series, in that it doesn’t deal with general principles. But Vhojha Moi the howlers I just witnessed. Your bridge tip of the day:

Do not overcall your three card suits. No, really. Don’t.

Written by taogaming

March 14, 2010 at 9:34 pm

Posted in Bridge

Tagged with