To compute or not to compute.

Is it possible to get a sensible answer from a computer without even running a program? Yes indeed, report Hosten et al. on page 949 of this issue— and they have the experimental proof. The thread that leads thus far has its origin in the early 1990s, with the introduction of a quantum paradox known as the Elitzur–Vaidman bomb thought-experiment. They showed that by sending a photon into a simple optical interferometer (Box 1), one can sometimes detect the presence of a light-triggered bomb without setting it off — in other words, without light interacting with it at all. ‘Sometimes’ — there’s the rub. At other times, you set the bomb off; at still other times, you get no information about the existence of the bomb at all. But then that’s the probabilistic nature of quantum mechanics. It might seem to defy common sense, but the result is a straightforward application of the quantum theory of light. A few years ago, it was pointed out that an all-optical quantum computer could replace Elitzur and Vaidman’s bomb. Rather than exploding or not, this computer either runs a program or doesn’t, depending on whether or not a single photon impinges on its ‘run’ switch. And just as we can obtain the answer to the question ‘is the bomb there?’ without exploding it, so we can obtain the right answer from the computer without running it. (The computer must, at least, still be turned on and properly programmed in order for the scheme to work.) However, although intriguing, this original ‘counterfactual computation’ scheme suffered from the same quantum-probabilistic issues as the bomb business: sometimes the device would actually run the quantum computer to get the answer; sometimes it gave no answer at all. The probability that it would give the correct answer while not running the computer was so small that one could always do better by guessing the answer randomly. Not much of a computer, that. Another spin-off of the bomb idea was Anton Zeilinger and colleagues’ elegant and real (as opposed to thought) experiment called ‘quantum seeing in the dark’. That experiment allowed safe, non-explosive objects such as a human hair to be imaged, sometimes using no photons at all. Again, the key word is ‘sometimes’: just as with the bomb, sometimes no photons were needed to image the hair; sometimes some photons were needed to image the hair; and sometimes the hair was not imaged at all. By sorting the individual imaging events carefully, however, an image of the hair could be constructed using only the events in which no photon had interacted with the hair. Better still, the probability of these photon-free imaging events could be increased using yet another paradox of quantum mechanics, the quantum Zeno effect. This effect is named in honour of Zeno of Elea, the philosopher who gave us paradoxes such as that of Achilles and the tortoise (you can run, but you can’t catch up). It is the quantum-physical equivalent of the saying “a watched pot never boils”. If you measure the state of a quantum system often enough and fast enough, the system will remain in that state and never evolve to another state — even if it would evolve if you were not measuring it. Zeilinger and colleagues applied this idea by using a sequence of polarization rotators as ‘Zeno boosters’ that forced their interferometer system to remain far more often in the state of ‘imaging the hair with no photons’, and far less often in the states of ‘imaging the hair