Ken Wharton
@kenwharton.bsky.social
Physics professor at San Jose State. Quantum Foundations. Big fan of space and time, and also many things therein.
In some ways your concern here is analogous to analyzing the heat equation without any knowledge that there’s an underlying particle-based explanation of heat. That equation also predicts faster-than-light influences. Almost makes you wonder if we’re missing an underlying story in QFT, too. ;-)
November 11, 2025 at 12:44 AM
In some ways your concern here is analogous to analyzing the heat equation without any knowledge that there’s an underlying particle-based explanation of heat. That equation also predicts faster-than-light influences. Almost makes you wonder if we’re missing an underlying story in QFT, too. ;-)
The one story I heard was that he was embarrassed for SLAC to have to pay the page proof expenses, so he submitted it to this brand new PPF journal (with waived fees) instead of something more established. But I don't know anything about the Madison connection.
November 7, 2025 at 10:41 PM
The one story I heard was that he was embarrassed for SLAC to have to pay the page proof expenses, so he submitted it to this brand new PPF journal (with waived fees) instead of something more established. But I don't know anything about the Madison connection.
Bell was officially at SLAC that year, on Sabbatical from CERN, but travelled around a bit from there. Do you know how long he was in Madison?
November 7, 2025 at 10:33 PM
Bell was officially at SLAC that year, on Sabbatical from CERN, but travelled around a bit from there. Do you know how long he was in Madison?
Or at least with tenure / job security... :-)
November 4, 2025 at 12:38 AM
Or at least with tenure / job security... :-)
Didn't Emily explain it well in her book? I noticed you didn't talk much about that section in your long review. Anyways, I'm more than happy to try to help clarify the terminology... just ask! :-)
October 25, 2025 at 8:58 PM
Didn't Emily explain it well in her book? I noticed you didn't talk much about that section in your long review. Anyways, I'm more than happy to try to help clarify the terminology... just ask! :-)
There are several classes of models that would get rid of an ontic configuration space. After all, such spaces are precisely how one would formalize incomplete knowledge of 3D classical states. Here’s a link to my 2020 Rev Mod Phys piece which categorizes such “spacetime-based” reformulations.
Bell's Theorem and Locally-Mediated Reformulations of Quantum Mechanics
Bell's Theorem rules out many potential reformulations of quantum mechanics, but within a generalized framework, it does not exclude all "locally-mediated" models. Such models describe the correlation...
arxiv.org
October 23, 2025 at 8:36 PM
There are several classes of models that would get rid of an ontic configuration space. After all, such spaces are precisely how one would formalize incomplete knowledge of 3D classical states. Here’s a link to my 2020 Rev Mod Phys piece which categorizes such “spacetime-based” reformulations.
Maybe every “interpretation” of the wavefunction has this problem (depending how you define that) but not every possible “reformulation”. There’s no configuration space in the path integral formulation, for instance. Models can aim to explain quantum phenomena without interpreting wavefunctions.
October 23, 2025 at 8:34 PM
Maybe every “interpretation” of the wavefunction has this problem (depending how you define that) but not every possible “reformulation”. There’s no configuration space in the path integral formulation, for instance. Models can aim to explain quantum phenomena without interpreting wavefunctions.
That’s a good case for MWI. I’m in full agreement with this observation that the “main weirdness” is that one needs an exponentially large configuration space (plus time) instead of something that fits in our observed 4D universe. But it’s not fair to say that every approach has this problem.
October 23, 2025 at 8:34 PM
That’s a good case for MWI. I’m in full agreement with this observation that the “main weirdness” is that one needs an exponentially large configuration space (plus time) instead of something that fits in our observed 4D universe. But it’s not fair to say that every approach has this problem.
Interesting question! As I see it, many different causal models are consistent with E&M. (The causation isn't in the bare equations, those are just correlations.) Different cases evidently call for different causal models (controlling particles with lasers, controlling fields with charges, etc.)
October 14, 2025 at 3:40 AM
Interesting question! As I see it, many different causal models are consistent with E&M. (The causation isn't in the bare equations, those are just correlations.) Different cases evidently call for different causal models (controlling particles with lasers, controlling fields with charges, etc.)
Who knows? No one knows the right ontological model of what is "really" happening when we're not looking. My point is that if you are just looking at absorption phenomena, and trying to model them using classical E&M, there is motivation to consider causal models with future inputs/constraints.
October 14, 2025 at 3:36 AM
Who knows? No one knows the right ontological model of what is "really" happening when we're not looking. My point is that if you are just looking at absorption phenomena, and trying to model them using classical E&M, there is motivation to consider causal models with future inputs/constraints.
So, if I'm consistent in my application of causal logic, every time I see a photoelectric effect I should infer that I should use final-field-inputs (at least in part), to explain the observations. Using final-field causal inputs is actually a lot simpler than ditching classical E&M entirely.
October 14, 2025 at 12:08 AM
So, if I'm consistent in my application of causal logic, every time I see a photoelectric effect I should infer that I should use final-field-inputs (at least in part), to explain the observations. Using final-field causal inputs is actually a lot simpler than ditching classical E&M entirely.
But here's the thing: this 'time-reversed movie' I mentioned doesn't just happen when I play an emission event backwards. The same situation appears in real life, every time an array of atoms actually absorbs a photon! That's the photoelectric effect, an empirical fact.
October 14, 2025 at 12:06 AM
But here's the thing: this 'time-reversed movie' I mentioned doesn't just happen when I play an emission event backwards. The same situation appears in real life, every time an array of atoms actually absorbs a photon! That's the photoelectric effect, an empirical fact.
If I took a movie of this and played it in reverse, despite the apparent advanced field, I'd reach the same conclusion. The way to explain a convergence of the field (onto one particle) would be to use a *final* field input. (Final in the time-reversed movie, meaning "initial" originally.)
October 14, 2025 at 12:01 AM
If I took a movie of this and played it in reverse, despite the apparent advanced field, I'd reach the same conclusion. The way to explain a convergence of the field (onto one particle) would be to use a *final* field input. (Final in the time-reversed movie, meaning "initial" originally.)
Okay, now, replace the charges with a bunch of excited atoms, in some metastable state. One atom decays. A classical E&M model is still pretty good here. A similar initial-field-input causal model is needed to explain the pattern of where the field can eventually be detected.
October 13, 2025 at 11:58 PM
Okay, now, replace the charges with a bunch of excited atoms, in some metastable state. One atom decays. A classical E&M model is still pretty good here. A similar initial-field-input causal model is needed to explain the pattern of where the field can eventually be detected.
Yes, exactly. If you use a causal model with a different field input, you won't get the retarded solution. And notice it's not a Cauchy problem! The particle input isn't all at the beginning, it's a full worldline. (And if I told you I shook the particle with a laser, that's another deal entirely.)
October 13, 2025 at 11:53 PM
Yes, exactly. If you use a causal model with a different field input, you won't get the retarded solution. And notice it's not a Cauchy problem! The particle input isn't all at the beginning, it's a full worldline. (And if I told you I shook the particle with a laser, that's another deal entirely.)
Now, I want to causally model this experiment, in this region of spacetime (without zooming out to the whole universe!), recovering the observed retarded fields from my model output. What model “inputs” should I therefore use? (Think about both field inputs and 4-current inputs.)
October 13, 2025 at 11:15 PM
Now, I want to causally model this experiment, in this region of spacetime (without zooming out to the whole universe!), recovering the observed retarded fields from my model output. What model “inputs” should I therefore use? (Think about both field inputs and 4-current inputs.)
This will be fun! :-) First, a warm-up exercise. A bunch of charged particles sit on a plane. We decide to shake particle X. There are many solutions to the eqns. of E&M which are consistent with the worldline of X. But we generally see the EM fields of one particular solution (the retarded field).
October 13, 2025 at 11:14 PM
This will be fun! :-) First, a warm-up exercise. A bunch of charged particles sit on a plane. We decide to shake particle X. There are many solutions to the eqns. of E&M which are consistent with the worldline of X. But we generally see the EM fields of one particular solution (the retarded field).
If you decide my argument works, start a new thread and ask me about the analogous case of microscopic classical E&M. The empirical evidence for causation in that case sometimes points in a very different direction!
October 10, 2025 at 3:14 PM
If you decide my argument works, start a new thread and ask me about the analogous case of microscopic classical E&M. The empirical evidence for causation in that case sometimes points in a very different direction!
That’s not circular logic: that’s using empirical observations to draw conclusions. Sure, it doesn’t go through if you think that there’s no causation, just bare correlations. But then, wouldn’t the empirical success of causal models in this context be evidence that we should use them? ;-)
October 10, 2025 at 12:36 AM
That’s not circular logic: that’s using empirical observations to draw conclusions. Sure, it doesn’t go through if you think that there’s no causation, just bare correlations. But then, wouldn’t the empirical success of causal models in this context be evidence that we should use them? ;-)
So your questions really come down to this: Why do classical causal models with past inputs work, and why do models with future inputs fail? If our universe can be causally modeled, I can see only one possible answer: some relevant external “inputs” to our universe really do lie in our past.
October 10, 2025 at 12:35 AM
So your questions really come down to this: Why do classical causal models with past inputs work, and why do models with future inputs fail? If our universe can be causally modeled, I can see only one possible answer: some relevant external “inputs” to our universe really do lie in our past.
When we model thermodynamical processes in some region, we set the inputs at the beginning, and almost without trying we find successful models which generally explain the observations. Sure, you could try different casual models, putting the inputs at the end. But those models fail, empirically.
October 10, 2025 at 12:34 AM
When we model thermodynamical processes in some region, we set the inputs at the beginning, and almost without trying we find successful models which generally explain the observations. Sure, you could try different casual models, putting the inputs at the end. But those models fail, empirically.
Causal models are asymmetric, by definition. In any causal model, we treat the causes as special inputs, imagining that we can set them to anything we want. Then we compute the result for everything else, calculating the “effects” of the causes as we counterfactually tweak the inputs.
October 10, 2025 at 12:34 AM
Causal models are asymmetric, by definition. In any causal model, we treat the causes as special inputs, imagining that we can set them to anything we want. Then we compute the result for everything else, calculating the “effects” of the causes as we counterfactually tweak the inputs.