The Tuned Deck

In Sweet Dreams: Philosophical Objections to a Science of Consciousness, Daniel Dennett recounts the story of “The Tuned Deck,” a mysterious magic trick that’s more — or less? — than it seems. Conjurer Ralph Hull performed it over and over throughout his life, challenging his fellow magicians to figure out how it was done. Here’s Dennett’s account of how Hull’s act went:

“Boys, I have a new trick to show you. It’s called ‘The Tuned Deck’. This deck of cards is magically tuned [Hull holds the deck to his ear and riffles the cards, listening carefully to the buzz of the cards]. By their finely tuned vibrations, I can hear and feel the location of any card. Pick a card, any card… [The deck is then fanned or otherwise offered for the audience, and a card is taken by a spectator, noted, and returned to the deck by one route or another.] Now I listen to the Tuned Deck, and what does it tell me? I hear the telltale vibrations, … [buzz, buzz, the cards are riffled by Hull’s ear and various manipulations and rituals are enacted, after which, with a flourish, the spectator’s card is presented].

Nobody ever figured out how it was done, and Hull only released the secret posthumously. Dennett explains:

Like much great magic, the trick is over before you even realize the trick has begun. The trick, in its entirety, is in the name of the trick, “The Tuned Deck”, and more specifically, in one word “The”! As soon as Hull had announced his new trick and given its name to his eager audience, the trick was over. Having set up his audience in this simple way, and having passed the time with some obviously phony and misdirecting chatter about vibrations and buzz-buzz-buzz, Hull would do a relatively simple and familiar card presentation trick of type A (at this point I will draw the traditional curtain of secrecy; the further mechanical details of legerdemain, as you will see, do not matter).

His audience, savvy magicians, would see that he might possibly be performing a type A trick, a hypothesis they could test by being stubborn and uncooperative spectators in a way that would thwart any attempt at a type A trick. When they then adopted the appropriate recalcitrance to test the hypothesis, Hull would ‘repeat’ the trick, this time executing a type B card presentation trick. The spectators would then huddle and compare notes: might he be doing a type B trick? They test that hypothesis by adopting the recalcitrance appropriate to preventing a type B trick and still he does “the” trick – using method C, of course. When they test the hypothesis that he’s pulling a type C trick on them, he switches to method D – or perhaps he goes back to method A or B, since his audience has ‘refuted’ the hypothesis that he’s using method A or B.

And so it would go, for dozens of repetitions, with Hull staying one step ahead of his hypothesis-testers, exploiting his realization that he could always do some trick or other from the pool of tricks they all knew, and concealing the fact that he was doing a grab bag of different tricks by the simple expedient of the definite article: The Tuned Deck.

It’s a cool story, but it’s also a relevant lesson for scientific research, or any situation in which you’re testing  theories about what’s causing some phenomenon. If you test each possible cause in isolation, you could end up concluding that none of them are responsible for the phenomenon — but in fact it could be that two or more of them are at work, either simultaneously or at different times, and so when you remove one at a time you’re not changing anything. You’d have to remove both at once to see the effect.

I was reminded of the Tuned Deck recently when my friend and I were having trouble with our Skype video chats getting dropped or becoming pixelated and distorted. We came up with a list of possible reasons it was happening: (1) Maybe the connection is weak in this room? (2) Maybe the time of day is a busy one for the Skype servers? (3) Maybe our computer has too many applications running in the background? etc. And we tested each reason in succession, ruling each one out in turn.

But what if there were different reasons Skype was failing at different times? Maybe the time we tested hypothesis (3) by closing all the background applications, it didn’t fix our video chats because we were chatting at a busy time of day. And then the time we tested hypothesis (1) by moving our laptop closer to the router, it didn’t help because we still had too many applications running in the background. So we ruled out both hypotheses, unjustifiably. (Until eventually I exclaimed “Wait a minute! The Tuned Deck!” and we had to overhaul our entire hypothesis-testing protocol.)