There is an important question in physics and philosophy which can be traced all the way back to Einstein. It was never really laid to rest, and in recent years it has become, if anything, still more pressing. The question is this: does pure quantum mechanics (without any state reduction or hidden variables) describe any actual events? In particular, does it describe ordinary, actual, observable events? The orthodox response was always yes to the first question, so long as we are talking of the microscopic world; and no to the second, so long as we are talking of observed experimental events; for on making an observation, on the orthodox interpretation, the wave-packet collapses, and selects out one or another of the possible experimental outcomes.
Strange as that sounds, it was more than enough of a correspondence between formalism and experiment to be able to make predictions and tests. For physics it seemed it was business as usual. There remained only the problem of explaining how “observation” could have such extraordinary effects - the problem of measurement.
The problem was supposed to be a philosophical one. In philosophy it festered. It was refreshing to learn, much later in the history of the subject, that the problem could be solved by changing the equations instead. In fact we now have an embarrassment of solutions: the deterministic hidden-variable theory, due to de Broglie and Bohm, and a spectrum of stochastic state-reduction theories, for example the one due to Ghirardi, Rimini, and Weber. All these theories represent actual events uniformly, microscopic and macroscopic, observed and unobserved, as making up a single world (as a single space-time whole). There is no special role for “observation”: this is very much the sort of solution which most of us desire. Alas there is a difficulty: it turns out to be extremely difficult to carry over these methods to the case of Lorentz symmetries. There is to this date no non-trivial model, neither in hidden variables theories nor state-reduction theories, for even the simplest of genuinely relativstic quantum effects (such as pair-creation or annihilation in quantum electrodynamics).
Since manifestly not a tool of research in particle physics, none of these deviant theories were ever been taken seriously by the mainstream in fundamental physics. And meanwhile, beginning with Mott and continuing with Everett, Zeh, Zurek and others, a number of methods have been found for talking about definite events within quantum mechanics without any additional mechanisms. If there is any state reduction involved, so goes the jargon, it is “effective”, and not part of the fundamental equations. At the latter, deeper, level relativity is preserved. The wheel doesn't need reinventing after all, the equations of relativistic particle physics are perfectly good just as they stand. Or so it is claimed.
The price, however, is high - too high for most. The way to do without state-reduction, whilst insisting that the state alone represents the world that we see, is if it represents worlds we don't see as well - an idea first intimated by Schrodinger but only properly explored by Everett. It is committed to version of what philosophers call modal realism: the thesis that all (physically) possible states of affairs exist.
I have defended a formulation of the Everett interpretation, based on the decoherent histories formalism and assuming that decoherence alone, an effective dynamics, determines the branching structure. That further a notion of uncertainty is engendered with branching, and an agent is right and proper to be uncertain of which branch to expect in the face of branching. Chancing turns out to be branching, according to Everettian quantum mechanics, somewhat as time turns out to be the lengths of timelike lines, in special and general relativity, determinism, respectively changlessness, notwithstanding. I have further argued that probability otherwise has a similar status as it does in other physical theories, and suffers from some of the same conceptual difficulties - but in a particularly vivid way. Thus there are branches with deviant records of statistics, albeit of infinitesimally small quantum mechanical weight. They still exist. (But so, in a stochastic theory over sufficiently large times, do there exist stretches of records of deviant statistics.) In these senses I have defended the claim that the Everett interpretation is the only conservative interpretation of quantum theory (which perserves the equations as is).
Is there no alternative? - but of course: there remain hidden-variable and state-reduction theories. The majority of philosophers of physics champion one or other or more of these theories. But they mostly consider only the non-relativsitic formalism: my work on alternatives to quantum mechanics has focused on the meaning of probability in these theories and on the difficulties facing these theories when it comes to explaining - predicting - the phenomenology of relativsitic particle physics.
Is it really true that there is no other conservative interpretation of quantum mechanics? But then how has it been interpreted all these years? The answer is not so much that it was interpreted, as that it was seen (by Bohr at any rate, whose writings are the nearest to an "orthodoxy") as a symbolic calculus: the real work of interpretation concerned experiments rather than equations. And here Bohr did, in my view, and contrary to most if not all commentors on Bohr today, give an interpretation largely free of metaphysics and philosophical principles - in fact a broadly realist interpretation. For the details, see my "Complementarity and Scientific Rationality".
That does not, however, mean we have an interpretation of the equations. The equations seem to tell us something very different (and Bohr's view, maintaining as it did the priority of classical theory, is of little use when it comes to quantum cosmology). I return to the Everett interpretation. The two problems - of relativsitic covariance, and of the treatment of probability - have a very different status in the Everett interpretation: there there is no obvious problem with relativity, but extremely vivid problems with the interpretation of probability (it is the other way round with pilot-wave and state-reduction theories). Very few philosophers of physics have been prepared to advocate Everett's ideeas; mostly, they find them unbelievable. But this is not the objection usually made (although it is a reasonable objection): more commonly, the Everett interpretation is either held to be ill-posed, or incoherent, or both, increasingly on the grounds that it offers no intelligible interpretation of probability. But here there have been significant recent developments, due in particular to David Deutsch and David Wallace, which I believe, and contrary to my own earlier expectations, do genuinely yield a derivation of the probability rule in Everttian quantum mechanics, of a kind that has never been possible before in a physical theory.
I arrive at the suprising conclusion, that whether or not Everettian quantum mechanics survives as the correct theory of the world (at least up to the regimes of energy and scale so far explored), it will remain the paradigm of a theory of the world in which probability is fully explained in terms of the physical properties of things. <
Copyright Simon Saunders 2001. Last updated: 8 Dec, 2007.