Wednesday, August 16, 2017

Teleportation redux

I’ve had an illuminating discussion with Mateus Araujo after he blogged about my piece for Nature on quantum teleportation. Mateus pointed out my error in suggesting that the protocol requires (rather than merely permits) the teleported state to be unknown, for which I’m very grateful. I’ve amended the story to put this right.

This doesn’t mean Mateus is now happy about all of the article – and given our (friendly and useful) exchange, I didn’t think he would be! He was also concerned that I was perpetuating the notion of some kind of magical, superluminal transmission of information between the original and target particles in teleportation. I was puzzled by that, because part of the point of the piece was to call out that very misconception. But then it transpired that Mateus’s concern stemmed from this paragraph:
“A common view is that quantum teleportation is a new way of transmitting information: a kind of high-speed quantum Wi-Fi. What’s amazing about it is that the quantum ‘information’ is ‘sent’ instantaneously — faster than light — because that is how two entangled particles communicate.”

I explained to Mateus that I had imagined this statement would be obviously intended ironically: that “amazing” refers to the breathless visions of the “common view”. It seems Mateus still feels I’m claiming that “something unobservable is going on faster than light, and that there is some kind of conspiracy by Nature to cover that up.” I’m not sure if Mateus thinks I still believe this, or simply that my piece still implies it regardless. So to be clear: I don’t think anything of the sort is happening at all. It seems to me that the causal language in which something here (Alice’s measurement) influences something there (the state of Bob’s particle) is precisely the wrong one to describe quantum entanglement. (I agree with David Mermin’s comment that entanglement presents us with correlations for which there is no “explanation”.)

This does raise the issue of whether rhetorical devices like irony should be used in science writing, where on general one tries to be as clear as possible (often to an audience whose first language is not English). It’s a question I do ponder over, and I’ll be writing about it soon for Chemistry World. I’ll say here only that I have been surprised (and not a little alarmed) before to discover that some scientists (not Mateus, however) seem unaware that such devices even exist. A little slice of Two Cultures Pie, perhaps?

Mateus originally stated that I also invoked in my article the “discredited notion of wavefunction collapse”. I made no reference to that at all, but he argues that it was implicit in the discussion: the idea of Alice’s actions having any kind of instantaneous consequence – perhaps better to say, instantaneous implication – for Bob’s particle requires wavefunction collapse. It was not my intention to do so, and perhaps by disavowing any sort of “instantaneous information transfer” in a physical sense I clear that up. But at any rate, to say that the notion of wavefunction collapse is now “discredited” (Mateus has suggested that no serious scientist now talks in those terms) is simply wrong. Indeed, the book chapter on which my article on the “measurement problem” (which Mateus liked) was based was taken seriously to task by another expert who decided that (my actual text to the contrary) it claimed to do away with collapse altogether.

The idea that decoherence completely “replaces” wavefunction collapse is not mainstream at all. Even some of those who think that it can – such as Roland Omnès, who says collapse can now be seen as “a convenience, not a necessity” – are careful to point out that decoherence doesn’t clear up everything. For it doesn’t explain the uniqueness of measurement outcomes. That, says Omnès, still needs to be added in what is essentially an axiomatic manner: unique facts exist.

So on the one hand I think her Mateus’s comment is an example of a more general tendency I’ve noticed among folks who think deeply about quantum foundations to suggest that, not only is their preferred interpretation the only one that makes sense, but it is the only one taken seriously at all.

On the other hand, this suggestion that “wavefunction collapse” is an obsolete idea raises the interesting question of what we mean by it in the first place. Some, of course – like Roger Penrose – think that this collapse is an actual physical process, just as decoherence is. And it is, moreover, a process that breaks the unitary behaviour rigorously observed by the Schrödinger equation (and which poses such dilemmas in connecting the theory to experience). This is not a mainstream view either, but it is a respectable one, pursued by some leading scientists – for it has the advantage, perhaps uniquely among quantum “interpretations”, of being empirically testable in principle.

For the likes of Bohr, wavefunction collapse was indeed a vague and problematic notion (or so it seems to me, though I’m wary of saying anything about what Bohr meant). But most physicists who talk about it don’t see it as some physical process; rather, it is a part of the mathematical formulation of quantum theory. It is, moreover, a necessary part: it amounts to applying the Born rule in order to make probabilistic predictions about the outcomes of experiments, and so it is needed to make any kind of connection at all between the theory and empirical experience. In other words, it’s a mathematical process. Some don’t think it is very helpful to talk about that process in the “physicalist” language of collapse; others have no problem in doing so. Everettians claim that the Many Worlds view does away with collapse at all, but then they face the problem of explaining why we need the Born rule to actually use quantum mechanics to make predictions. (Some, like Sean Carroll here, have claimed to derive the Born rule within the Many Worlds framework using choice theory, but this requires one to invoke the notion of a rational observer with a well-defined conscious experience that is continuous in time, which seems to be precisely what the Everettian view renders incoherent, unless it is simply imposed by fiat.)

So wavefunction collapse hasn’t gone away – though how nice it would be if we could make it do so.