As AlephNeil points out, theory unification in physics doesn’t invalidate everything that came before; quite the contrary. In the passage from classical to relativistic mechanics, you lose absolute simultaneity but you retain conservation laws. In that case, it may also be said that the feature of classical physics that was retained—conservation principles—was something for which there was ample empirical evidence, whereas the feature which was lost—the existence of absolute time—was an ontological presupposition which proved to be dispensable.
I consider your argument from the supposed incompatibility of QM and GR to be a bad argument precisely because it has no evident connection to the aspect of QM which is at stake here. In other posts, Eliezer advocates for Julian Barbour’s approach to quantum gravity. There is a small change in the nature of QM here: in Barbour’s theory, there is no time evolution. So the conflict is resolved in a way which does not touch Eliezer’s argument in this post.
You also mention string theory as a unification of QM and GR from the particle physics side, implying, I suppose, that it is GR which has been modified. That would be a very debatable assertion. It is a commonplace of the string versus loop debate to present strings as an outgrowth of particle physics culture, and loop quantum gravity as an outgrowth of the culture of gravitational physics (relativists), and to say that the string theorists neglect general covariance (or “background independence” as it is usually termed in such discussions). This appears to be a historical contingency; it is now a common belief in string theory that the real observables all exist only on the boundary, precisely because the bulk has diffeomorphism invariance; and meanwhile, for practical purposes, diffeomorphism invariance is just another symmetry, which you break by gauge-fixing for the purposes of calculation, but which will still be there in the predictions (i.e. the effects of gauge-fixing must disappear by the end of the calculation).
It would be very unusual to argue that the unification of QM and GR requires a change in how we think about particle statistics. If you actually had such an argument, it would be worth hearing, but you don’t; you just have an argument which bundles the whole of quantum mechanics into one proposition, and the whole of general relativity into another proposition, an argument which claims that the conjunction of these propositions is a contradiction, and that therefore one or the other is false.
Incidentally, how do you get from that to QM is “very probably false”? “The conjunction of A and B is false” does not imply “A is very probably false”! It seems clear that your problem with QM has nothing to do with the alleged incompatibility with GR—this incompatibility (which perhaps you only believe in because of what various authorities say) merely gives you leverage in debate against quantum dogmatists, or hope for a concrete alternative.
Unless you actually have an argument which makes a connection between particle statistics and quantum gravity, you should just stick to particle statistics, and not bring gravity into the picture.
I’m hardly claiming that if we find some true future unified theory of physics, every physical proposition we current believe is false. To assert that would be absurd. The interesting question is which of the propositions that are part of the current packages (at least one of which is false) are in fact false.
If you want to pick some of those propositions and rely upon them, you’d better have independent evidence for their truth (the accurate predictions made by the package isn’t going to count). So rely on that as your evidence, and not on the false package. This is what less wrong people would do.
Barbour is engage more in philosophy than in hard science, and his work is published in either published in poor journals or book form. I could care less whether Eliezer endorse Barbour’s views, if he does. And if he does, then maybe he’d do better to rely upon them than upon QM. If Eliezer does advocate for Barbour’s views, one must wonder why. On the basis of some nice hard scientific evidence? Or on the basis of lots of wishful thinking?
I’m not very interested in getting into a debate about how to properly taxonomize physical theories. That’s irrelevant to any of the points I’ve made, and the debate would be even more irrelevant. Two paragraphs of your response are on this irrelevant subject matter.
It is perfectly acceptable for me to bundle the whole of QM into one proposition, if Eliezer is baselessly relying upon QM. If Eliezer wants to rely on something OTHER than QM, then he can. Something OTHER than QM would be a part of the theory (but not the whole of the theory), or something else like a theory of quantum gravity. If he wants to rely on those things, fine. Then we’ll see what the evidence is for what he relies upon. What, for example, is the evidence for treating Barbour’s odd-ball theories to be more likely true than, say, a fixed-foliation quantum gravity? Well, there isn’t really much evidence. Bad Eliezer.
Your last comment is well worth remarking on. By “very probably false”, I roughly meant that there is a 50% probability that it is false. If there is a 50% epistemic probability that P is false, and your belief that Q is based solely upon your belief that P, then you are irrational if you believe that Q.
You know, I made a mistake at the beginning of this discussion, by conceding your premise that QM and GR are inconsistent. I should have just asked to see the proof.
Usually when people say this, they mean that applying the techniques of perturbative quantum field theory (the techniques which produce Feynman diagams) to general relativity gives rise to an unpredictive theory, because there are infinitely many undetermined quantities associated with higher-order terms in the expansion. The philosophy of renormalization is that you measure such quantities and use them to define the renormalized theory, but that this only makes sense when there are finitely many such quantities.
However, there’s still no proof that this is true for general relativity. There is a research program, “asymptotic safety”, which hopes to find evidence that only finitely many quantities are needed after all. I am skeptical, even though a physicist as good as Steven Weinberg is interested, and as I said, no proof exists. Meanwhile, it does look as if d=4 N=8 supergravity is perturbatively finite. That is a general-relativistic theory—it contains Einstein gravity, coupled to other fields. Although it is perturbatively finite term by term, the Feynman expansion probably doesn’t converge, in which case there need to be other contributions, and most likely N=8 supergravity should be considered a limit of string theory.
Anyway, the bad behavior of general relativity when it is treated using the simplest methods of quantum field theory is by far the main reason that people have for talking about QM and GR as inconsistent, and it’s a theoretical opinion from several decades ago, that is not borne out by more recent developments. If that’s your reason for asserting that QM and GR are inconsistent, I am prepared to rebut that point in as much detail as you wish. If you have some other reason for asserting that they are inconsistent, let’s hear it.
As AlephNeil points out, theory unification in physics doesn’t invalidate everything that came before; quite the contrary. In the passage from classical to relativistic mechanics, you lose absolute simultaneity but you retain conservation laws. In that case, it may also be said that the feature of classical physics that was retained—conservation principles—was something for which there was ample empirical evidence, whereas the feature which was lost—the existence of absolute time—was an ontological presupposition which proved to be dispensable.
I consider your argument from the supposed incompatibility of QM and GR to be a bad argument precisely because it has no evident connection to the aspect of QM which is at stake here. In other posts, Eliezer advocates for Julian Barbour’s approach to quantum gravity. There is a small change in the nature of QM here: in Barbour’s theory, there is no time evolution. So the conflict is resolved in a way which does not touch Eliezer’s argument in this post.
You also mention string theory as a unification of QM and GR from the particle physics side, implying, I suppose, that it is GR which has been modified. That would be a very debatable assertion. It is a commonplace of the string versus loop debate to present strings as an outgrowth of particle physics culture, and loop quantum gravity as an outgrowth of the culture of gravitational physics (relativists), and to say that the string theorists neglect general covariance (or “background independence” as it is usually termed in such discussions). This appears to be a historical contingency; it is now a common belief in string theory that the real observables all exist only on the boundary, precisely because the bulk has diffeomorphism invariance; and meanwhile, for practical purposes, diffeomorphism invariance is just another symmetry, which you break by gauge-fixing for the purposes of calculation, but which will still be there in the predictions (i.e. the effects of gauge-fixing must disappear by the end of the calculation).
It would be very unusual to argue that the unification of QM and GR requires a change in how we think about particle statistics. If you actually had such an argument, it would be worth hearing, but you don’t; you just have an argument which bundles the whole of quantum mechanics into one proposition, and the whole of general relativity into another proposition, an argument which claims that the conjunction of these propositions is a contradiction, and that therefore one or the other is false.
Incidentally, how do you get from that to QM is “very probably false”? “The conjunction of A and B is false” does not imply “A is very probably false”! It seems clear that your problem with QM has nothing to do with the alleged incompatibility with GR—this incompatibility (which perhaps you only believe in because of what various authorities say) merely gives you leverage in debate against quantum dogmatists, or hope for a concrete alternative.
Unless you actually have an argument which makes a connection between particle statistics and quantum gravity, you should just stick to particle statistics, and not bring gravity into the picture.
I’m hardly claiming that if we find some true future unified theory of physics, every physical proposition we current believe is false. To assert that would be absurd. The interesting question is which of the propositions that are part of the current packages (at least one of which is false) are in fact false.
If you want to pick some of those propositions and rely upon them, you’d better have independent evidence for their truth (the accurate predictions made by the package isn’t going to count). So rely on that as your evidence, and not on the false package. This is what less wrong people would do.
Barbour is engage more in philosophy than in hard science, and his work is published in either published in poor journals or book form. I could care less whether Eliezer endorse Barbour’s views, if he does. And if he does, then maybe he’d do better to rely upon them than upon QM. If Eliezer does advocate for Barbour’s views, one must wonder why. On the basis of some nice hard scientific evidence? Or on the basis of lots of wishful thinking?
I’m not very interested in getting into a debate about how to properly taxonomize physical theories. That’s irrelevant to any of the points I’ve made, and the debate would be even more irrelevant. Two paragraphs of your response are on this irrelevant subject matter.
It is perfectly acceptable for me to bundle the whole of QM into one proposition, if Eliezer is baselessly relying upon QM. If Eliezer wants to rely on something OTHER than QM, then he can. Something OTHER than QM would be a part of the theory (but not the whole of the theory), or something else like a theory of quantum gravity. If he wants to rely on those things, fine. Then we’ll see what the evidence is for what he relies upon. What, for example, is the evidence for treating Barbour’s odd-ball theories to be more likely true than, say, a fixed-foliation quantum gravity? Well, there isn’t really much evidence. Bad Eliezer.
Your last comment is well worth remarking on. By “very probably false”, I roughly meant that there is a 50% probability that it is false. If there is a 50% epistemic probability that P is false, and your belief that Q is based solely upon your belief that P, then you are irrational if you believe that Q.
You know, I made a mistake at the beginning of this discussion, by conceding your premise that QM and GR are inconsistent. I should have just asked to see the proof.
Usually when people say this, they mean that applying the techniques of perturbative quantum field theory (the techniques which produce Feynman diagams) to general relativity gives rise to an unpredictive theory, because there are infinitely many undetermined quantities associated with higher-order terms in the expansion. The philosophy of renormalization is that you measure such quantities and use them to define the renormalized theory, but that this only makes sense when there are finitely many such quantities.
However, there’s still no proof that this is true for general relativity. There is a research program, “asymptotic safety”, which hopes to find evidence that only finitely many quantities are needed after all. I am skeptical, even though a physicist as good as Steven Weinberg is interested, and as I said, no proof exists. Meanwhile, it does look as if d=4 N=8 supergravity is perturbatively finite. That is a general-relativistic theory—it contains Einstein gravity, coupled to other fields. Although it is perturbatively finite term by term, the Feynman expansion probably doesn’t converge, in which case there need to be other contributions, and most likely N=8 supergravity should be considered a limit of string theory.
Anyway, the bad behavior of general relativity when it is treated using the simplest methods of quantum field theory is by far the main reason that people have for talking about QM and GR as inconsistent, and it’s a theoretical opinion from several decades ago, that is not borne out by more recent developments. If that’s your reason for asserting that QM and GR are inconsistent, I am prepared to rebut that point in as much detail as you wish. If you have some other reason for asserting that they are inconsistent, let’s hear it.
(edited to add: undetermined quantities)