Transcript round-table discussion

21 July 16.30
Round -table discussion
Panel:
David Wallace
David Z. Albert
Jeremy Butterfield

Floor speakers (in order of appearance):
Saunders
Deutsch
Myrvold
Albrecht
Valentini
Maudlin
Hartle
Geroch
Pitowsky
Hemmo

128

Wallace
1. It would be very churlish for any academic to mind the fact that they’ve had so many people criticising their ideas I don’t think I’ll get the chance to reply to them all. Nevertheless I’m very grateful to have this window to say few things. And I’m also very grateful for all of the various very constructive and very civil comments which have been made through discussion. I’ve been very pleased and pleasantly surprised that, though vigorous they were never heated in the discussions. Anyway, Simon, right at the beginning of the conference brought up the taxonomy of the issues of Everett, we should think about and basically broke them down into ontology, probability and evidence. And I think that’s a good breakdown and I’m just going to say a few varyingly-sized things on those topics.

2. I’m not going to say very much on the problem of ontology because I said most of what I wanted to say in my own talk two days ago and I think the issue’s been very thoroughly discussed at various points. I did just want to say I thought that I think perhaps unintentionally, perhaps not, Antony’s closing comment, albeit that it was intended as a criticism of Everett, is quite a nice way of seeing in some ways how the ontological situation developed because – I think it’s entirely right that for a long time we (people doing quantum mechanics) thought much too much about quantum mechanics in terms of eigenvalues, eigenvectors, operators, definiteness – you know, no-fact-of-the-matter-about, neither-one-nor-the-other and I think one of the things that’s really made it possible to get a real grip on what quantum mechanics is trying to tell us, at least Everett-style, has been successive moves away from that, it’s been quantative? operator value measurements and Wigner? function representations being played with a lot more and consistent history frameworks and decoherence and pointer bases and the whole story which has led us to understand that - if we do want to think about quantum mechanics realistically, thinking in the eigenvector-eigenvalue-operator way is simply flawed. So, I would say the Everett interpretation shows exactly how we should take the theory which we were bequeathed from … . classical considerations and read it properly. So, I entirely agree with what you wrote but drew slightly different conclusions from it.

3. Dealing with probability and evidence, I think I’ll take these the other way around. So, I think the evidence problem was brought up, I think very clearly, by David yesterday, but I think perhaps the shape of Everettian responses to this haven’t been made completely clear. And I think there have been two families of responses available, one sort of being driven by Simon and myself and one being driven by Hilary and more lately by Wayne and I think complicating matters is the fact that I think Simon and I broadly buy Hilary and Wayne’s strategy and I think Hilary at least broadly is at least sympathetic to Simon and mine or at least some of the premises of it, so I don’t think we should regard these as in conflict, and of course this is great because it ought to be the case in the world, when things are right one ought to be able to work out that they’re right by more than one means. It’s the things that can only be argued-for by a single tortuous route that you want to get worried about, whether they’re really getting at the truth. But just to say, I think, roughly how those two very different routes are supposed to go, here’s how Simon and mine is supposed to go: The reason we’re worried is that, and Tim put this very clearly, it looks as if the sort of framework for decision-making, for confirmation and for evaluation of theories just goes across hopelessly badly into the Everett framework, I mean in particular, vividly, that it is deterministic so we use of decision theory; in what is there in deterministic decision theory but transitivity? Also, how do we judge the valitidy of theories and collect data and evidence when all the data is there in the branching structure? Essentially the point of what Simon and I am saying and the response that we’re making to David’s worries about ‘look, since this is not metaphysics, how can it help?’ is essentially to say: By saying, look, our entire framework fits very badly into the Everett interpretation we’re conflating two things. We’re conflating our sort of ordinary pre-theoretic grip of how to think about theory and confirmation which scientists have been doing fine for ages and which people doing informal science have been doing for ages; we’re confusing those ideas (informal, intuitive statements of natural language) with the way those ideas translate and map and fit into the fundamental metaphysical picture , the book of the world as John called it. And the claim is that when we go from saying things about the way …. .we all know to mapping that onto some sort of statements about third-person accessible god’s-eye view facts, propositions as statements which take truth values on possible maximal states of the universe, all sorts of frameworks like that, when we make those moves we’re using a correct – we’ve got a correct intuitive theory of epistemology and so forth – we’re then applying – because, claim: we have the wrong metaphysics, we’re mapping that onto the world in a certain way which is more up for grabs than we realised. So our talk of: we confirm things by various ways like, for instance, Barry’s comment, that there’s no point carrying out an experiment when we know the answer, that’s inside our pre-theoretic framework. That’s just fine, we shouldn’t carry out experiments when we know the answer. It’s a mistake to move from that - so we shouldn’t carry out experiments when we’re in possession of a theory according to which the results of the experiment are predictable from a god’s-eye view with certainty – that move incorporates a lot of metaphysical assumptions about the single universe. If the single universe picture is right then those assumptions are just great, if it’s not they’re not great at all; it would beg the question to assume they’re right in the single universe picture and then therefore apply them in the wrong way to the many-universe picture. So that’s the response to the David criticism, it is true that this doesn’t affect our metaphysics, it only affects our semantics and non-metaphysical aspects; some of the aspects it affects are the rules of the game in our own theory of confirmation and decision. Anyway, that’s how our programme is supposed to go.

4. How does Hilary’s programme go? I think the way it goes is perhaps only partially visible in the presentation that Wayne gave on that project so I’d like to say a little bit more about it and Hilary can correct me if I get it wrong. Hilary’s framework is saying, as I see it, never mind these semantic roots, we’ll accept that agents need to accept the possibility that they should have non-zero credence in the Everett interpretation and non-zero credence in things other than the Everett interpretation and they should have that as their starting point and work out how they should update those credences when they receive data. So she’s assuming that Baysian epistemological framework, rightly or wrongly but very respectably and very mainstreamly and she’s saying there has to be a unifying framework that lets us work this way because if not it’s not a scandal for Everettians it’s a scandal for everyone and let’s be prepared to say the Everett interpretation is logically impossible, that we must somehow be prepared to give no credence to it at all, we have to have a manageable epistemic framework which includes in that framework somebody who gives a credence of ten to the minus 35, we’ve still got to give it some credence to handle that and, from that same framework, if our first stab at doing that says that all data we get massively confirms the Everett interpretation, that’s not a problem for the Everett interpretation, that’s an urgent problem for our theory of confirmation that needs to be dealt with, Everett or not-Everett. And I take it that what she does within that is construct a framework of that form which genuinely is unified and which includes as a sub-component in it how we handle these various branching situations but doesn’t presuppose those branching situations, and within that unified framework she establishes on what seem to very reasonable assumptions that we would indeed rationally increase our credence in the Everett interpretation given the sort of data that we do in fact receive from quantum mechanics. That’s how I see that programme. It’s been said at various points that it’s not at all obvious how it is, if the world is Everettian, then these statistics confirm it? Absolutely, on Hilary’s premises, not at all obvious, it’s hard work, here’s the work, here’s how it goes. That’s how the Greaves-Myrvold side of that goes.

5. So much for evidence, let me now say something about probability. At various points there have been various allusions to, and in one case David’s interesting criticisms of, the decision-theoretic, representation-theoretic programme which David Deutsch instigated and which I and others at Oxford have worked on to some extent. But I think those discussions haven’t entirely made clear what’s going on in that framework and in particular there’s been a lot of talk about the assumption of Equivalence that’s going on in the way I presented those arguments. And I want to dispel one misconception about this. So Equivalence says if I’ve got two equally-weighted situations I should be indifferent between them. That is not intended to be an intuitively obvious primitive principle of rationality, it’s not intended to be a primitive principle of rationality at all. It’s intended to be something to which we argue, which we establish on the grounds of things which we do think are intuitively reasonable primitive principles of rationality. I think part of the reason why Equivalence seems to be a starting point rather than a middle point, apart from some presentational problems of mine in the paper I wrote on this, is that Equivalence is the point at which, at least in what we have so far, formalisation kicks in. From Equivalence it becomes a theorem, and, as Simon said, given Equivalence the game is over. But I wouldn’t want people to read that as saying Equivalence is a starting point, we make no argument for Equivalence. So I think that partly what that leads to is thinking, well, I can come up with lots of other rules which violate Equivalence, and David gave a very interesting example of this which makes vivid the point, this is David’s example of fatness. So the claim would be of course that violates Equivalence, David never claims that it doesn’t violate Equivalence. On my position, which I take as not being essentially different from David’s [Deutsch] position and at least others in this room who are pro the Everett interpretation or sympathetic to it, the problem with that position is not that it violates Equivalence per se it’s that it violates other things which we regard as in fact rationally compelling, which entail Equivalence. I’d like to flesh that out a little bit so, I’ve said this to various people at lunch, but I think it’s worth bringing out.

6. I started my presentation at this conference talking about Dan Dennett so I’ll finish be quoting Dennett again. …. .of Dennett’s that when you get an intuition pump, intuition pumps are fine if lots of people play with them, twiddle the dials, work out the details, see where things go, and I think that’s quite instructive when we think about David’s example. So, in David’s example let’s suppose I’m considering a situation where I’m considering for a large amount of money buying a quantum lottery ticket and if I win the ticket and let’s say the weight for me winning is 0.5 then I go on holiday in Mauritius and if I don’t win the ticket I don’t go on holiday. And I might do the calculation and think, well, it looks like, on the expected utilities, it’s not worth me buying the ticket. But, I know what I’ll do, when I’m in Mauritius I’ll eat lots. I’ll eat like a pig in Mauritius; my fatness on the Mauritius branch is going to shoot up and now it’s worth me going to Mauritius, and now it’s worth me buying a ticket. But hold on, I like to be fit and bronzed on holiday and I know that once I get to Mauritius I’m not going to want to put on weight at all but by David’s assumption we’re neutral about the value of our fatness, but I can enforce this so what I’ll do, I’ll pay some money to hire a minder and the minder will come to Mauritius with me and the minder will enforce the fact that I put on weight, put on fatness that is, and of course when I get there I’ll be annoyed about the minder, the minder will get in my way so I better brief the minder very carefully on my foibles to make damned sure that the minder doesn’t let me – that I’m not able to escape from the minder. Okay, so I get that far then I think some more. Then I think, oh dear, if I end up not getting the ticket I might go bankrupt, other bad financial things might happen and I know that my successor who doesn’t win the ticket is going to be very worried about that and he’s going to be thinking: well, when I get miserable I put on weight, I comfort eat when I get miserable and that’s going to lower the utility of all the things I’m going to do, so I’m going to hire another minder to make sure that actually what I do is put on lots and lots and lots of weight if good things happen to me in this branch and of course I have to brief this minder. So I now, before even buying the ticket, know this is going to happen if I don’t win the holiday so I could hire another minder and this other minder’s job is to stop the first minder. I know that my future self is going to hire a minder to make him go away I’ve got to stop that guy. So I’m committed to this massive game of conspiratorial action against my own future selves and I’m committed to this game of second-guessing all the things that I’m going to do in the future and this is after two instances of branching; branching is happening quintillions of times a second. So that’s the kind of way in which I think those sorts of objections, while they’re well worth having on the table to see what happens, why they ultimately don’t work. And it sort of makes vivid if you like what was in the theorems from (well, they’re not vigorous enough to be called theorems) the arguments with which I and earlier in different ways David [Deutsch] argued for Equivalence or an equivalent principle, and it’s kind of trying to show why, even if Equivalence is not a priori obvious as a rationality principle the other principle relying on our … like the requirement of temporal consistency, are kind of obvious.

7. As a last comment on that let me just say one response you sometimes can get from people worrying about these decision-theoretic arguments is, okay, that position’s irrational. Well, maybe everything’s irrational therefore in the Everett picture maybe this isn’t? .rational to do, and I think that really does miss the point because I think representation theorems, decision theory, Dutch book arguments, Savage’s position, Jeffrey’s position, Ramsey’s position; they’ve always been illimitive? in nature; they’ve always said, given certain rationality constraints I will rule out every possible strategy bar this one, so Savage establishes, in a context entirely independent of Everett, that, if you accept his axioms, any move other than maximising expected utility is irrational. You don’t respond to that at all by saying, well, maybe maximising expected utilities is irrational too. If you can provide the arguments to say it’s irrational then of course that’s fine, then there’s something profoundly wrong with Savage’s framework, but one can’t argue against Savage’s style of decision theory by simply just saying, well, maybe the only one ……… is also irrational. I don’t think you can argue against these positions by that move either. So I’ll put that as pre-emptive loophole closing, if you like. I’ll stop there.

Albert
8. Let’s see, maybe I should start off not by what I was planning to say but responding to a few things David said. Of course, the choice of fatness in this example was partly to be funny. So, one can make up stories in which my life might become complicate if I were to adopt that caring measure. Two things to say about that. Good, I can make it depend on features of the world, physical features of the world not the amplitude, that are less malleable than that so as to make these compications a little less threatening but the bigger point seems to be: I’m not sure how the observation that it would be complicated to follow a certain decision procedure impacts on the worry that the coherence of such a procedure poses for a claim or an argument or a theorem to the effect that that’s the procedure – that some other procedure is the one a rational agent would have to adopt under those circumstances. Maybe that’s all I want to say about the fatness argument.

9. Maybe I can take a few minutes of my time just to try to learn more about the first of the evidential positions you were talking about, that is, the one based on ideas of ignorance. Maybe it will be helpful to other too. These are really just questions. If I understand your and Simon’s present position correctly one of the things that’ll turn out to be true is the following: I’m sitting here, there may be a couple of guys overlapping with me right now, I’m assessing probabilities about which one I am and those probabilities turn out curiously to depend branchings that may occur arbitrarily far in the future. That struck me as odd immediately; I tried to think about what’s going on there a little bit more and I guess it turns out to be related to the facts? .so call one of the overlapping guys David One and call another one of the overlapping guys David Two, so on and so forth. Of course locutions that you’re naively attempted to engage in like, you know, if I want to say wel David One is going to see spin-up, superficial appearances or grammatical appearances to the contrary, that doen’t turn out to be a contentful claim about the world, that’s just the meaning of ‘David One’. David One is the one who continues? .to spin-up, okay. So, it doesn’t amount to a real- it’s constitutive of which one David One is that he’s the one who ends up seeing spin-up

Wallace
10. Yeah, .. constitutive of your rules as to how the name ‘David One’ .., yeah, sure.

Albert
11. That’s right. So there’s something a little weird about speaking as if which David I am is some feature of how the world is now, of which I’m ignorant. It’s not like that at all really. In that case I’m not sure – this talk of ignorance begins – I don’t know what – maybe this kind of ignorance begins to lose a lot of its bite and a lot of its familiarity, that is, it just isn’t right and – this is the way I thought it was being portrayed: that’s going on now is I’m sitting here wondering which one of these I am, as if I’m wondering about some feature of the way the world is now. That’s not what I’m wondering about at all, okay. I don’t know – look,

Saunders
12. I’m wondering if it’s me or not

Albert
13. But the me here is just the one – that’s the trouble – it’s not as if – . Contrast it with the following situation, say, a single-mind situation, like Barry and I had, okay, there’s a guy here now - . Or, a multiple-minds situation. There’s a couple of minds, they’re different colours, okay, each of them is stochastically going to go one way or the other, or maybe even there are deterministic laws like in Bohm’s theory, that I’m wondering where my particle is, one of them is going to this way, one of them is going to go that way. In cases like that, wondering which one I am is in a clear and transparent and concrete sense wondering about some feature of the way the world is now, of which I take myself to be ignorant. That seems not ..

Wallace
14. Let me say something about that that may be helpful. Firstly, I think the disanalogy with the Albert-Loewer theory is less strong than this? In your theory minds didn’t, so far as I recall, come with colours or indeed serial numbers and of course they could, that would be a modified form of the theory, but in fact the theory as stated, as normally discussed, didn’t have that at all and it seems in that theory

Albert
15. Maybe not, but it did have - In the theory as stated we violating Leibniz’s identity of indiscernables, they were supposed to be different minds

Wallace
16. Sure, but what I mean is in that theory, again, if you want to name some of those minds I don’t see you’ve got any mechanism available to you better than just saying, well, David One or David .. One is the guy - I think that problem comes up again, it comes up just as much

Albert
17. I don’t think it’s coming up in the same way. These minds aren’t – There is such a thing – well, I don’t know how to say it – ther’re multiple minds here now, and there’s a fact about which one of them is going to go this way and which one is going to got that way

Wallace
18. So you want to say that the counterfactual I’m considering is the one where there’s a swerve

Albert
19. Right

Wallace
20. Okay, so that point I guess I take. Here’s the more general response to make: the uncertainty we’re dealing with here (I said this this morning) seems to me to be, if it’s not an indexical uncertainty I’m not quite sure why not. I mean, these very boring examples we use, that got talked-about a lot in philosophy in the seventies and eighties when I wasn’t really around. You have two qualitatively identical guys in boxes in distant parts of the universe, and this guy is wondering whether he’s in the Andromeda galaxy and so’s this guy. Now if we want to name those guys all we’ve got available is ‘Andromeda Guy’

Albert
21. I’m very confused by this and I’m really just thinking about it in the hour or so since I heard Simon this morning, but it seems, and I’ll abandon this here because I all I’m doing is expressing my confusion but it seems on the face of it like, even the indexical case you’re citing is a case of wondering, not how the objective third-person physical world is now but wondering where I am located now, wondering in that sense about a situation of the world now in a way that’s still not reproduced here.

Wallace
22. So I guess stands or falls on whether you buy the claim that underpins the semantics Simon was running that, in this picture, utterances or contentful mental states or whatever are attributed to these continuants

Albert
23. No. I understand that, but if you do buy that – and I have no idea, maybe this was a Lewisian reason for rejecting it, what sounds as if it’s a statement to the effect that there’s some feature of the world now about which you’re ignorant really turns out not to have that structure

Wallace
24. Okay, so a feature – and the word ‘now’ is what’s doing the crucial work here . There’s no feature of the world now .. Let me just add a little

Albert
25. But the whole appeal of this, the whole intuitive appeal here was presumably: you know how this is, there are a couple of tracks, which track are you on, but it’s not like that

Wallace
26. I don’t think that is .. the intention here. Certainly I’ve made lots of claims about the extent to which Everett? leaves things alone - I would not wish to claim that one of the things – where Everett preserves our intuitions is the correct god’s-eye theory of reference. …. .guessed it intuitive that that theory involved … common situation. The argument for that is not really and intuition-driven one, it’s going to be a sort of ..

Albert
27. But look, this goes back to, and I’m sorry, I know we’re going around in circles and I could be boring everyone here. I’ll say this and finish up. The real statement to make about this is that I have to think about this a lot more and it’s a very interesting claim and I make no bets on how likely it is I would say the same thing in two weeks’ time as I’m saying now. But, the worry is that once again the metaphysics at the bottom is remaining constant, that’s really where all the trouble is, okay. And one worries that there are various attempts going on at sugar-coating it, and at making it seem as if you can say standard sorts of things like ‘I don’t know which one I am’ although on further analysis that turns out to mean something very, very different from what it superficially appears to mean. But anyway, that’s a worry; I don’t know if I’ll say the same thing two weeks down the road.

28. Let me just say one other thing. As I mentioned, I had in mind to talk about something altogether different than what I actually talked about here and maybe I’ll just take two minutes to mention it. It’s on an altogether different subject, a subject which hasn’t come up here so much even thought it’s often discussed in connection with Everett, which is maybe worth mentioning. Much of the discussion here has focussed on the question of whether Everett is internally coherent and even minimally empirically adequate. If both of those of those questions were to be settled in the affirmative then other questions would arise. Good, how does it compare, what are its advantages and disadvantages put up against other solutions to the measurement problem and one of the advantages that is often cited in favour of Everett in those discussions, which as I say didn’t occur very much in this meeting, is a sort of compatibility with relativity that it’s thought none of the other attempts to solve the measurement problem can boast of and, just in connection with that, it’s recently begun to seem to me, and this also interestingly enough arises in part from discussions I’ve had with Wayne a couple of years ago and from some much older papers I wrote with …[name]…and Ahronov, all of which I somehow never really understood until recently. It’s struck me lately that there are reasons for worrying about how well quantum mechanics sits with special relativity that are quite independent of any attempt to solve the measurement problem, quite independent of collapse, quite independent of non-locality, things that will come up in the context of the dynamical equations of motion that have to do with the phenomenon of entanglement, that go like this and it’s a very trivial thing. It can be said very fast, but perhaps it’s worth thinking about in the context of these discussions.

29. So, consider the following situation, I have fours spin-half particles with no dynamics at all except possibly a small dynamics of interaction with one another that I’ll describe in a minute. In a certain frame of reference, here’s the world-line of particle 1 [going to board], here’s the world-line of particle 2, here’s the world-line of particle 3 and here’s the world-line of particle 4 and it’s arranged so that in the frame that we’re looking at things in these two crossings here are simultaneous with one another. Imagine that 1 and 2 start out in an entangled singlet state and that 3 and 4 start out in an entangled singlet state, so there are two EPR pairs here. And now I ask you to contrast the effect of the following two possible Hamiltonians under which this system might be imagined as evolving. Hamiltonian 1 is zero, a completely trivial Hamiltonian, everything just stays like this, these maintain their constant velocities, that’s it. The spin-state of the world from t equals minus infinity to t equals plus infinity is at all instants these two singlet states, this product of two singlet states.

30. Hamiltonian 2: there’s an interaction, a contact point interaction, the approximation of it as impulsive is not going to be important here, I can relax that approximation and everything I say will go through but let’s hang on to that approximation for the moment. So, Hamiltonian 2 is: there’s an interaction which, when two particles cross each other, when two particles coincide in space and time, they exchange their states, okay. So up-up goes to up-up, down-down goes to down-down, up-down goes to down-up, down-up goes to up-down. Once again, assume they do it impulsively. Note that if that interaction is on it will affect the total state history of this universe as viewed in this frame not at all. You will have a pair of singlet states from t equals minus infinity to t equals plus infinity because of the fact that these two exchanges occur simultaneously, so you’ll just be exchanging one singlet state for another between these two …..By the way, assume that 1 is distinguishable from 2 is distinguishable from 3 is distinguishable from 4, just to keep things simple. They’re four different colours. Note that in any other frame where these two events are non-simultaneous this won’t be true at all. Whether this Hamiltonian is on or off will make a huge difference to the state history. Because if the Hamiltonian is off you’ll still have two singlet product states from minus infinity to plus infinity. If the Hamiltonian is on there will be a period in between this interaction and this interaction when 1 is entangled not with 2 but with 4. And when 2 is entangled not with 1 but with 3. And then when the second interaction occurs 2 will become re-entangled with 1 and 3 will become re-entangled with 4. We’ve just proved that there can’t be any such thing as, as it were, a purely geometrical transformation that gets us from the full sequence of states on this set of hypersurfaces to the full sequence of states on this hypersurface.

Deutsch
31. In the Schrodinger picture.

Albert
32. Good, states will mean something different. Let’s stick with the Schrodinger picture for the moment. There can’t be any such transformation, that is, one cannot detect from the full set of state assignments?,,along this set of hyperplanes whether the Hamiltonian is on or off. But whether the Hamiltonian is on or off makes a huge difference to the sequence of states along that set of hyperplanes. There’s a lot to say about this. Let me just give you one punch line very, very briefly which is relevant to this. In cases where we’re dealing with entanglement, and in the Schrodinger picture, I guess I’m willing to say in the Schrodinger picture but - well, there’s a lot to say about this. There’s a prima facie sense in which I don’t understand the Heisenberg picture as an ontology, as opposed to a sort of instrumental kind of claim about how to make the calculations. But one will need to go into that, see what is necessary to call it a genuinely evolving ontology and see whether once that’s all there we can tell this story again or not. Anyway, in the Schrodinger picture, the story is that we’re faced with the following choice. Here’s a principle we never needed to name because it was never seriously called into question before because it has been true of all or our theories up until the relativistic quantum theories. That is, it’s true of classical relativistic theories, it’s true of non-relativistic quantum theories and of course it’s true of pre-relativistic, pre-quantum theories. All these theories have the property of portraying the world as what you might call narratable. What that means is, all these theories have the feature that they say that everything there is to say about the world can be told in the form of a story, by which I mean told by saying first this was the case, then this was the case, then this was the case, dot, dot, dot. That’s a principle to which one needs to consider how deep our commitments are, how it weighs against other principles and so on and so forth. What’s going to happen in relativistic quantum theories, at least in the Schrodinger picture and anything like it, is the demand for narratablity and the demand for Lorenz invariance are going to be in direct conflict with one another.

33. Once again, there’s a lot to say about this, I have a long paper about this and so on and so forth. Let me not try to …all that. The only thing to say is, this might be read as follows: there are reasons to worry about the principle of relativity in quantum mechanics way before you hit non-locality, way before you hit measurement, just as a result of entanglement. These will be worries that are inherited by an Everettian theory to the same extent as they’re inherited by any other quantum-mechanical theory. One could regard this in combination with all kinds of other things as saying that quantum mechanics generally is putting enormous pressure on special relativity and it may be more be in the balance than we have come to think and it’s not perhaps as obvious as it used to be that the relationship of an Everettian quantum theory with relativity is going to be so much more to be desired than the kinds of relations that exist in other ways of solving the measurement problem.

Deutsch
34. This is a great argument for how if you use the Schrodinger picture to analyse …[pouring water noise]

Albert
35. Well, that’s good. An additional argument is going to be needed. Prima facie, but this is the beginning of a discussion not intending to be shutting one down, prima facie I’ve just never understood, and this is not rhetorical this is just the straightforward truth, I’ve never understood what it would mean to be ontologically serious about the Heisenberg picture. That is, if somebody tells me the state, by which I usually mean the situation of the world, doesn’t change but operators do, whatever that means, my immediate reaction is going to be: gee, that can’t be right, stuff is changing all the time. It’s sort of as simple and as flat-footed as that and, like I say, that’s more of an invitation to teach me how to think about he Heisenberg picture than it is a declaration of what is or isn’t possible.

Deutsch
36. One quick point is that the “state” in the Heisenberg picture is is a misnomer. It’s a completely trivial concept in the theory and if the Heisenberg picture had been the only one ever in use then the observables would have been called the state.

Albert
37. But I’m not sure what that even means.

[new question]

Myrvold
38. I want to say something that David has heard me say before but I just want to get a clean response to what he said out on the table..[cross talk]…………it’s not news that going from a state history along one foliation to another one is not a purely geometrical thing. It involves knowing the dynamics. Gordon Fleming has a nice paper on the Philsci archive explaining all this and he was prompted to write this paper because he was talking with some philosophers and it surprised him to find that they
didn’t know this.

Albert
39. Ahronov and I said this in 1987. I just didn’t know what I was saying.

Myrvold
40. ……going from the state on one hypersurface to another involves evolving in the space in between so you need to know how to evolve the states

Albert
41. Hold on a second, that’s misleading. We’re not talking about going through this state to this state, we’re talking about going from this whole set of states that fills up the spacetime to this set [indicating on board]

Myrvold
42. Within each foliation evolving from one state in one to another state in another involves dynamics and that’s the reason why you need to develop?..a dynamics. Now,

Albert
43. I guess I disagree with that but anyway, go ahead.

Myrvold
44. Now is this a threat to narratability? Yes, on a Humean view that is non-realist about the dynamics. If you count whether or not that Hamiltonian is turned on as part of what happens and include it in your story along one foliation part of what happens is the dynamical evolution from one time to another and then you don’t have a threat to …[cross talk]…and that’s why I don’t think this indicates any kind of tension with special relativity, but we can continue this some other time.

[new question]

Albrecht
45. Just a brief comment which is: physicists when they talk about relativity in quantum mechanics….relativistic quantum field theory. That’s a very, very specific type of quantum mechanics with very strict limitations on what kind of operators you can have and what the Hamiltonian looks like and so I would expect, before I would get excited about any issues here, to see the whole thing, including interactions and whatever turns them on and off built out of relativistic quantum mechanics.

Albert
46. But that’s going to be what the problem is here. All I can do is - it would be a long story to build this as an explicitly field-theoretic model but that’s not going to be what the problem is.

[change of theme]

Albrecht
47. So I’m going to try to address a very small piece of all the wide-ranging conversation. I think earlier today there was a comment that Everettians don’t know how to get the macroscopic out that went unchallenged as far as I could tell. So I want to challenge it and one thing I’ve learned this last few days is that there’s many people here with many different ideas about what it takes to be happy about something or get convinced about something. So I don’t know how many Davids I’m going to convince about this but I can tell you why I’m convinced. I’m and Everettian who’s very happy about how the macroscopic world is to be understood. And the basis of that is actually a toy model. I know there’s a number of people who’ve done similar things so I’m just going to talk about what I did but there’s people in the room who’ve thought along similar lines and done similar exercises and I’m not going to give details unless people ask me to.

48. So basically one exercise I did [going to board] was write down a toy model where I computed numerically the Schrodinger equation for a spin coupled with a system which I called the apparatus. It was small enough to model on a computer therefore nothing like a real apparatus but I specifically engineered the Hamiltonian of the system to have this form with self-Hamiltonians for each of these and an interaction Hamiltonian. And I constructed these pieces in a way that would allow the following evolution to take place. So the initial wave function was a combination of up and down, times some state for the apparatus and if I evolved it over a period of time, I would get at a later time psi-primed ..[writing]..So I was able to make this apparatus entangled with the spin, in this particular basis, the nature of the interaction Hamiltonian chose a basis for me, and this had a property - and these continued to evolve over time but the inner product of these two states over time was approximately zero. Over a long period of time, it was a finite system so in fact it would eventually cycle back to - so in principle it would come back to the state but I was able to engineer it to make this stable as long as I wanted and to have it copy the coefficients into the entangled state. So I was able to do this, I was able to fiddle with the Hamiltonian to make it a good or bad apparatus and see which properties it had to have and then I was able to reflect on real physical systems and say whatever I had to do here, they?..can clearly do it better than this. And so that’s a small piece. And there’s other aspects of the classical world as well, this is a small piece of the story and it’s why I feel very comfortable thinking about the emergence of macroscopic behaviour in the Everett picture.

Valentini
49. ..start off with the state of the system….say one basis is up-plus-down and another is up-minus-down..

Albrecht
50. I can do that with aphas and betas, right

Valentini
51. Yeah, I mean you’re going to get new thing, it’s going to be alpha-primed 1 plus beta-primed times 2 and then

Albrecht
52. But I rewrite - If you’re going to let me have a pure state for the spin I can always write it this way if I want. But the reason I write it this way is because ..interaction Hamiltonian which we might as well put down. This has the form ..[writing]..So the interaction Hamiltonian chooses this basis, and these are different Hamiltonians. So basically what happens is that if the spin is up this guy pushes the apparatus one way..

Albert
53. Look, I don’t think that any of the people who were worrying about how you get from the many-worlds interpretation to our everyday language about the way the world behaves were doubting for a millionth of a second that something like this was true.

Albrecht
54. It’s just a question of what makes you happy. This makes me very happy…

Albert
55. Maybe it shouldn’t. I’m speaking here I guess mostly for Tim who was the one who raised the doubts explicitly about this and they’re doubts that I may think are more addressable than he thinks but, I mean, the doubt was just about, you know - What the hell is this about? [gesturing at board] This is some wave function in an extremely high-dimensional configuration space, it’s some blob up there, it doesn’t look like an apparatus, it doesn’t smell like an apparatus…..[cross talk] ….

Albrecht
56. I’ll also go on the record as saying that doesn’t sound worrying at all to me. I can make this - I can iterate this and make it smell and look more and more like an apparatus without …

Albert
57. You’re not going to iterate it out of a high-dimensional space into the prosaic low-dimensional space, okay. No matter how many other apparatuses you add, or something like that. There’s going to have to be some argument. Now, like I said, I’m more optimistic about the possibility of providing such and argument than Tim may be but Tim’s point, which doesn’t seem to be addressed here at all is that - I mean it seems to be exemplified here - is that people don’t seem to grant that there’s something to be discussed there.

Maudlin
58. From this thing all I can do is repeat - this of course is no surprise to me even vaguely. So there’s just some complete misunderstanding what we were concerned about if you’re under the that pointing at this….[cross talk]…this is sociological, you’re not on the right wavelength at all….

Albrecht
59. One of us is not on the right wavelength.

Maudlin
60. You’re not on the right wavelength for what our problem was if you thought we didn’t know this. We knew this.

Hartle
61. I’ll just say that I didn’t completely understand Tim’s talk but we have certainly a view of what it means to get out the macroscopic world which consists of thinking in terms of some large-dimensional configuration space…..[faint]…..describe it…..rather what we think it consists of are the histories that describe the evolution of say fields or..[noise]..in time and that are constructed in such a way that we look for histories that first decohere so that we can assign probabilities, second that have within them the descriptions of all the possible objects, tables, chairs and so forth that we might see typically as represented by densities, appropriate distributions of particle number and so forth and of energy within the world and that these evolve, approximately, according to classical laws. I’m not sure whether that would satisfy Tim but that the way we would get out the usual description of classical physics; that’s how we get out a classical physics from quantum mechanics.

Maudlin
62. When Jim gave his talk he was talking about a very substantial ..[noise]..picture

Hartle
63. I’m not talking about that now

Maudlin
64. No, no, let me make the comment, there’s one where you say the histories are somehow associated with projections and then when we go to quantum cosmology the histories aren’t projections, you have things that you might think of as a particular evolution of spacetime and …..and those are what the histories are and if, in addition, because you asked the question, does it really help you to think that one of those actually occurred, I would say yeah, ……one of them actually occurred then I know what the probabilities are for…[noise]..the one around me is the one that actually occurred, then I’m perfectly happy, so

Hartle
65. I’ll just say – I didn’t completely understand Tim’s talk, but we have certainly a view on what it means to get out the macroscopic world, which is that it consists of – we don’t think in terms of some large-dimensional … some large-dimensional configuration space in trying to describe it . So … for example, .. not in phase space, rather what we think it? .consists of, are the histories that describe the evolution of, say fields, or particles .. in time, right, and that are constructed in such a way that we look for histories that, first, decohere so we can assign probabilities, second, that are – have within them the descriptions of all the possible objects, tables, chairs and so forth that we might conceive of typically as represented by densities, with distributions of particle numbers and so forth and energy within the world and that these evolve approximately, according to classical laws. I’m not sure that would satisfy Tim but that’s the way we would get out the usual description of classical – that’s how we’d get out classical physics from quantum mechanics.

Maudlin
66. When Jim gave his talk he was talking about .. very substantially …. ..picture … There’s one where you say, well, the histories are somehow associated with projections and then with your quantum cosmology the histories aren’t? .projections, right, you have things that you might think are .. …. Spacetime and stuff … those … what the histories are and if, in addition, you ask the question does it really help you to think that one of those .. had actually occurred I would say yeah, one of them’s actually occurred and I know what the probabilities are for .. … the world around me is the one which actually occurred, that I’m perfectly happy with. So,

Hartle
67. We’re in agreement at some level

Wallace
68. Quite a shallow level actually

[Wallace invited to comment]

Wallace
69. It’s really just to see, if you like, what the Albrecht-Maudlin gap - I think once one’s got on the table something is agreed is common ground, these various specific models of how the decoherence does certain things to certain environment .. divisions? .of various sorts there’re two more steps between that and the classical macroworld. The first step is that you might start out worrying about the arbitrariness of that decomposition, all the many ways it could be done – you might think that actually you’re seeing ghosts when you see this phenomenological stuff and partly that’s .. Antony’s concern .. do things in different bases. And it seems to me that problem, at a technical level, is very much the concern of the particulars of the Gell-Mann and Hartle strategy in decoherent histories in the nineties, at a conceptual level it’s very much the focus of the sort of structuralist ideas that I’ve been shouting that Everettians ought to be talking about and by now are talking about. And I think most physicists who are sympathetic to Everett agree that step needed to be done and an awful lot of that has been done. Then there is the second step which says once we’ve got something that’s structurally the real world, looks structurally like the world, can be matched to the world in some natural way, is that enough or is there more to do? And that, I think is more the sort of problem I guess Tim is saying and people … have said . That may or may not be a good? problem; I think it’s a philosophers’ problem, it doesn’t bother physicists. I think they’re right not to be bothered actually, but, as a comment on the structure of the debate, that seems to be what the gap is, from here to the end.

Albrecht
70. That sounds right

Geroch
71. On the theory that confession is good for the soul, let me tell you a misimpression that I had up to about three days ago. What I always thought was, that the way classical and things came out all had to do with people and it was a special feature of the kinds of interactions that people have, and it was going to be a very difficult problem because people are so difficult. Now what I’ve come to see - it’s not a problem of people interactions it’s property of the various interactions that go on in the world that even don’t have people involved at all- the electromagnetic interaction of the weak interaction and so on; it’s that the entire possible list of all the interactions you could imagine, somehow nature doesn’t seem to pursue every one of those, but it’s a very special kind of list that gets pursued and it’s that feature which gives rise to the classical appearances so, I’m thinking not now of classical things as really being a property of physics whereas before I thought of them as being properties of people. And to me that’s a very encouraging kind of thing because it’s my sense that what interactions are available is a really important thing in how physics goes.

72. Let me just give you one example. Let’s take classical mechanics. Well, the most naïve version of classical mechanics would have a space, a manifold of states of the system, which is roughly phase space, and a particle is described by a curve in phase space, and now we can ask: we don’t do it that way though, we have a configuration space and a momentum space, and you could ask where does the information come, just given the world as – or just given a system as a point in a space of states, where does the information about what are the configuration variables and what are the momentum variables, where does that come from? And, where it comes from is what interactions the system has available to it, it’s easy to show that the character of interactions of the external system within this system is such that the interaction changes the dynamcal vector field on the space of states and the allowed changes in the dynamical vector field by virtue of the interactions is the mechanism by which we decide which of the configuration variables and which are the momentum variables, so this is a kind of another example of something that we think of as having a lot to do with people but in fact it’s some general feature of nature in selecting from among all possible interactions some particular low class. So to me this has really been eye-opening to realise that somehow classical – so my enthusiasm for classical things has soared.

Hartle
73. I just want to stress one thing. We think of the so-called quasi-classical realm as something that is much stronger than I think people here are envisaging; that it’s a particular feature of our universe that doesn’t just exist at a few places of our choosing it extends, as far as we can tell, over the entire spatial slice much after the Planck scale. That is, when we take David’s NCG 2258

Wallace
74. 2293

Hartle
75. .and we equip ourselves with a – or we get enough money to make a journey – we don’t think that we’re going to have to take along apparatus that is sensitive to some sort of coherent superposition; the same [noise] will decohere to the same approximation roughly over there and would exhibit the same pattern of classical correlations over there. It’s not something which - and the level of fine-graining - is not something which is contained by – is described by the limitations of our ability to make measurements, that is, which is contingent on our particular level of ability. But it is something which is maximally defined, consistent with decoherence. So it’s not something of our choice but a feature of the universe in its entirety and the classical equations of motion are not things which follow in some elementary way from the underlying Hamiltonian, which exists in simple particle models; we’re talking after all about classical equations – a classical description, say in the context of heterotic superstring theory where the basic variables are quite distantly related to the underlying classical ones and we have to give an explanation of why those particular variables, which we intended to do? in terms of … and so forth, persist in the face of .... mechanisms of decoherence produces, and yet the nineteenth century classical law is a much stronger feature of our particular universe, less than I think the discussion has indicated, dependent on particular human choices that we might happen to make on this planet, and ….

Wallace
76. And so do I, incidentally

[change of subject]

Pitowsky
77. Let me be very brief, I’m speaking now about the Deutsch-Wallace derivation and let’s look at the derivations that the – there’s deriviations, right, de Finetti, Savage and so on - there’s a very clear distinction there between what they call rules of rationality which are based on things like the Dutch book argument and so on and so forth and what you consider in say in de Finetti’s case is a prior, or you consider in the Savage case is small? world models and so on and so forth in which you just set certain possibilities at zero, probability zero, in your prior and therefore they’re not going to get higher probability later on and I think some of the assumptions that are coming into your derivations are actually part of what traditionally within this derivation are going to be in the prior. Just one example, a simple example, nothing to do with …. or anything like that, suppose that your there’s a person who believes in quantum mechanics and the Born rule is sets for one - and this is a particular experiment at MIT, which in the normal circumstances .. [noise] .. spin half , you know, the wave function just gives half and half. But he thinks, for some reason, that the outcome … is somehow higher because of the stock exchange in New York and that’s his prior, so he violates Born’s rule and he’s violating non-contextuality of probability and he’s certainly not being rational in the general sense of what we mean by rationality but he’s not violating any rule of rationality in probability theory and therefore I think – or at least what – in the derivations you should be clear what you take as part of the rationality principles and what you take as some kind of empirical judgement of the kind that is usually made. I’m not saying that these empirical judgement are not justified, I think they are justified but I think that their justification without without a prior notion of probability, because after all we’re talking about priors, may be circular. So I think this is the challenge here.

Wallace
78. I think your being unduly kind by saying it unclear in my paper, I think it’s crystal clear, it might be completely wrong but I think it’s clear they’re all supposed to be rationality, they’re not supposed to be empirical at all. And I think you’ll just say that’s wrong, they should not be counted as rationality principles, but I’m interested in what – I always talk about these things but David came up with these ideas and I think it’s-

Deutsch
79. First of all there’s decision-theoretic rationality which doesn’t extend as far as ruling out that the stock exchange could affect …

Pitowsky
80. Right, exactly, that’s what my point is –

Deutsch
81. But then there’s a wider conception that one might call scientific rationality where you – physicalism, let’s say- so you want to say that your view of what is actually happening in a particular place only depends on the physical variables at that place, in the Heisenberg picture, and not on the variables in other places. Now that – I think we probably do need to make several assumptions like that to me, I don’t care how many assumptions we make so long as they’re not probabilistic assumptions. So long as we start off with something that isn’t talking about probability and end up saying that the rational agent behaves as if there were probabilities. So long as that happens I don’t mind those additional assumptions and in the case that you’ve mentioned they are, I would call them, scientific rationality not-

Albert
82. But, David, if you mean that literally why not just assume that the rational agent will behave as if the square amplitudes are probabilities and get it over with?

Deutsch
83. Well that is a probabilistic assumption.

Albert
84. No, no. As if. The rational agent will maximise utility by the following formula including the squared amplitudes, period.

Deutsch
85. The game is, you start of with assumptions, as many assumptions as you like, so long as you can make them – if you didn’t know about observers bifurcating then if – by the way, it doesn’t cover in my opinion other types of probability which don’t involve the observer bifurcating ……. like ….. principle. So, then the question is do those principles, and you can have as many of them as you like so long as they are

Pitowsky
86. But I? think they’re probabilistic, that’s my point, that’s what

Albert
87. I don’t know what David was going to say – let him finish

Deutsch
88. I’d almost finished - If you can have as many of those as you like so long as they apply – so long as they can be rationally justified in the case where observers don’t bifurcate, and then you add the additional fact that observers do bifurcate according to these laws and then use the same principles and see what they then imply; that was the game.

Pitowsky
89. I think – that was my point, that these are probabilistic assumptions, because these are assumptions about the prior; what you are giving initial probability zero. Initially the probability is zero, so they are assumptions about probability. Within this framework of rational decision theory, these are exactly the assumptions about the prior probabilities that you set up before the game begins. You can stand aside and say, well, I mean, these are just general scientific assumptions; I wouldn’t disagree with that, I’m just saying that, within this framework, they have probabilistic content so… And actually you sound as if, because of Gleason’s theorem, technical details apart, because of Gleason’s theorem, your are actually ..justifies? ..assuming that the square of the amplitude is the probability.

Hemmo
90. I want to continue what Itamar just said now, and that will relate to Hilary and Wayne’s paper. Given that what we think, that the assumptions we are talking about are assumptions about probability in the quantum picture the question is can you really argue – get to these assumptions – from empirical data? Namely, if you have some assumption about equal probabilities – just consider in a branching, basically a many-worlds theory, the situation where you live on an anomalous branch, I mean you’re going to come up with assumptions about priors which are not quantum mechanical. It seems to me there’s no sense which one could say that you – and then of course you’re going to reject quantum mechanics – and there’s no sense in which your objectively wrong in such a picture. I mean, in a sense which is related to the alternative features of the world

Pitowsky
91. There’s no sense in which you’re unlucky, that’s the point.

Hemmo
92. Yeah

Albrecht
93. Are you saying that, sort of at the classical level you have to have priors to make a sensible statement that probabilities are leaking in before you’ve gone through the exercise that Deutsch and Wallace ..

Pitowsky
94. I’m just saying that there’s a great freedom on the priors. This is just classical decision theory and probability theory …. there is freedom, actually a complete freedom on the priors if you don’t know anything, right. And if you know something that’s a source … it’s a part of decision theory that you make this distinction ……… The priors are judgements but they’re not judgements of rationality, they are based on something like knowledge

Albrecht
95. But you don’t think David addressed that just now?

Pitowsky
96. No, because I think he’s making probabilistic assumptions because the prior is an assignment of probability to certain possibilities.

Wallace
97. Just in terms of thinking about this project in the light of these objections, and I buy this way of thinking ultimately, it’s nice to think in terms of: ah, we Everettians solved probability. If you like a more conciliatory starting-point route. What do you need to put in and what of what nature do you need to put in to get probability out in Everett in contrast to say, what do you need to put in to get it out in a non-Everettian framework. In a non-Everettian framework it looks like you put in probability, maybe, maybe, maybe you can do better but I don’t think you can do better. Great if that’s the case, that would be worth knowing. Doesn’t look that way at the moment. What do you need to put into Everett? Things that seem to be much more plausible as axioms of rationality than the ….. itself. Are they logically true, probably not; are they reasonable assumptions about rationality, I’d be inclined to say yes. Are they probabilistic? I’d say no. I don’t think it’s entirely fair, certainly in the version of the argument I presented, for Itamar to feel that the assumptions going in are probabilistic. Maybe they’re unreasonable, but I don’t think they’re probabilistic. In …. .particular case the assumptions going in for instance, is that agent – that agent is assumed to only care about the state of the universe that is generated – of the multiverse that is generated by this process. So, for instance, unless he has independent reasons for caring about the New York stock exchange, he should be indifferent about the transformation in the change in the state of the stock market as well. And notice one of the assumptions going in there, and it is an assumption, is that the details of the short-period history that get us from before the bets are taken till after the result of the bet, including, for instance, in this fanciful example, the quite dramatic ….. of the entire New York stock exchange. That events of that small history are events the agent ought to be rationally indifferent to. Is that a logically true assumption, of course not. Is it a reasonable assumption, I’d say yes. Is it weaker than what’s going on in classical claims, I’d say yes. Is it probabilistic, I’d say no.

Myrvold
98. .. to say something that I think might help with our understand in that maybe ……………………. from the two Davids as we might think. ……. a moment ago said … the principle of scientific rationality … but it’s something like your probabilities only depend on whatever physical variables there are, right. Okay I just take that as a principle of rationality. To actually get something like Equivalence out we also need some kind of substantive physical claim about what kind of physical variables there are. And I don’t think you have any problem with getting probability assignments out of symmetry considerations – I’m going to require my probability assignment to depend only on certain variables … assumption about priors I would agree – and let me just make the point that it can’t be the case that Equivalence follows from … of pure rationality, it follows from – and I this sometimes gets obscured, but I don’t think either David is claiming that it does, it’s – it follows from traditions of rationality plus certain claims about what is and isn’t in the ontology of the world. And, let me just give an example, Antony is going to go look for evidence of non-equilibrium matter and if Everettian quantum mechanics is true there will be branches on which Anthony finds stuff that everyone is going to take as evidence about non-equilibrium matter and we’re going to award Antony the Nobel prize, we’re going to all think that quantum mechanics isn’t quite right, that some other theory is wrong [right?] and we’d be rational to do so, and I think that, I know David Wallace would say we’d be rational to do so, even if in fact the case is that we’re misled on that branch and that Everettian quantum mechanics is right.

Hemmo?
99. In what sense are we disputing this? …. .except that we assume by the postulates and … quantum mechanics is true

Myrvold
100. That’s why I made it conditional. If Everettian quantum mechanics is true and we see things that lead us to believe that it’s false then that’s that. That’s …… necessary .

Wallace
101. ……

Myrvold
102. But, David, you would agree that even if Everettian quantum mechanics is true we would be rational to reject it on those grounds.

Wallace
103. Oh yes. Rational - unlucky

Hemmo
104. …….. even if you find evidence in the physical world which refutes Everett you still can say what you just said and Everett is true and we are misled so

Wallace
105. That’s not true, that’s not the case. And that’s not what Wayne’s saying either.