Shuffling

As I’ve mentioned before, I play most of my bridge at the Manhattan Bridge Club. In response to customer requests for hand records after tournaments, the club has recently acquired a dealing machine. This is actually a pretty cool piece of equipment, when it’s working well. It’s hooked up to the PC, and the computer tells the machine which cards belong in which hand. An ordinary deck of cards is loaded in, and the dealing machine scans the face of each card, recognizes it, and feeds it into the correct (north, south, east or west) pile. You can either (manually) take the hands from the machine and put them in the board, or (this is the really cool way), if you have special boards that open up, you can put the board in the machine and then it deals the cards straight into the board. You just take out the board, close the lid, and move on to the next hand. Since the computer knows all the cards, it can produce the hand records, including double-dummy analysis by Deep Finesse.

So much to the good. But the introduction of this machine caused an immediate backlash against “computer-dealt” hands. The customer is always right, even when he’s wrong, so the management made a change to appease the new bunch of complainers. The hands are now shuffled and dealt by hand, so there is no basis for the “computer-dealt” complaint. The results of the manual shuffle are entered into the computer, and so (using one deck of cards really) a full set of 36 hand layouts are generated that can then be used by the dealing machine. Everybody is happy now, yes?

Well I’m not. Over the past few sessions, I’ve noticed a marked increase in the number of “interesting” deals. Misfits, slams on low point-count hands, singleton kings offside, etc suddenly seem to be cropping up all over the place. I have no evidence, of course, but it seems clear to me that at least one of the people doing the shuffling and dealing isn’t just entering whatever hand comes up. He is selecting for interesting or challenging hands, perhaps even adjusting things by moving a card or two around. This is understandable, in the sense of being a very human thing to do, when you are stuck with the tedious task of dealing and entering into the computer a set of thirty six hands. But I don’t need my games to be like a scenario from one of Hugh Kelsey’s books. I would be much happier with computer-dealt (ie truly random) hands.

The reason why people don’t like random deals is, as everyone knows, that they are used to seeing human-dealt hands that haven’t been shuffled properly. That typically results in hands that are noticeably too balanced, on average. So when they get a truly random set, where 4-3-3-3 is a rare hand pattern and most hands have a 5-card suit, they feel that things are “wild”. They aren’t: the usual hands are too tame.

Almost everyone uses some form of riffle shuffle. That is OK, as far as it goes, but most people shuffle three or four times and start dealing. If you really want to use a riffle shuffle, you should do it at least seven or eight times to get something approaching randomization. When’s the last time you saw anyone do that?

Many moons ago, when I was starting my duplicate career, there was a magazine article that outlined a very simple yet very random shuffle. You actually deal the cards, into a random number of piles, in a random order. To help avoid any unconscious bias, you can have someone else specify (occasionally) which pile a card should go onto, or whether it should start a new pile. (The piles don’t have to be the same size: if one has two cards and another fifteen, it doesn’t matter). I usually finish up with six or seven piles. These are then picked up in random order (and again, someone else can have input as to which pile goes on top next). The end result is a thoroughly randomized deck. And statistical analysis had been done to prove that the results are indeed more random than typical common shuffling techniques produce.

No particular skill is needed, because all you do is deal, mechanically speaking. This makes it a good method for children who haven’t mastered the necessary dexterity for other shuffles. The chief complaint I’ve heard (once I explained what I was doing) is that it takes too long. I think this complaint is bogus. For one thing, it doesn’t actually take all that long to do. For another thing, the complaint mostly comes from the people who don’t shuffle properly themselves. If they riffled nine times instead of three, they might rethink whether dealing the deck out takes too long.

This method gained some adherents back in the 1970s, but never became really widespread. I think that’s a shame, and if I could, I would try and make it better known.