“Wait, a theory was falsified by the invention of another theory?”
One that’s been confirmed against hundreds of predictions to over a dozen decimal places? Yes.
“Why put false in quotes?”
Because there’s a HUGE distinction between a theory like Newtonian gravitation and a “theory” like phlogiston, even if they’re both “false”.
“So you had a simple theory that explained some class of phenomena. Now there is some other phenomena that your theory fails to explain, why should simplicity be conserved here?”
I’m saying, “Successful theory of physics A was simple, and theory B was simple, and C was simple, and D, and E, and …. , but there’s a new class of phenomena which needs a new theory, so this theory will probably also be simple.”
“But the existence of competing approaches with varying degrees of popularity does not entail that.”
“Approach” here is a HUGE misnomer. “Approach” is a term commonly used in engineering to mean “different ways of accomplishing goal X”. You can build a machine in manner A to do X, or manner B. This ABSOLUTELY DOES NOT generalize to science, because there’s only ever one reality. If you have theory A and theory B both purporting to explain some phenomenon X, either A or B must be wrong, while in engineering sometimes there are ten different ways of attacking a problem, depending on what your goals are.
Is there a single example of this that you can think of? There are the different ways of computing classical mechanics (Newtonian, Lagrangian, Hamiltonian), but these were known to be just different ways of doing the same math at the time of discovery.
The original formulations of QM were famously shown to be equivalent, though I’m not sure they were ever expected to be incompatible. QFT and S-matrix theory were originally politically opposed (though QFT produces an S-matrix), but I have heard that recently people advocate widening QFT to the point that it appears to cover all of S-matrix theory.
Is there a single example of this that you can think of?
No, it’s just a theoretical property of the ‘A, B, X’ abstraction that you mention. In fact, it would not surprise me if the general problem of proving whether or not two theories are exactly equivalent is intractable in a similar way to the halting problem.
One that’s been confirmed against hundreds of predictions to over a dozen decimal places? Yes.
I was just clarifying, you said that it was ruled out “as soon as quantum mechanics was invented” which clearly isn’t right. Anyway, I had just meant LET versus SR, but what exactly is the experimental evidence against an aether? Obviously QM gets along pretty well without it and the vocabulary is incompatible… but the vocabulary of QM is incompatible with the vocabulary relativity as well.
Because there’s a HUGE distinction between a theory like Newtonian gravitation and a “theory” like phlogiston, even if they’re both “false”.
Maybe, but they’re still both false! What exactly is the distinction you have in mind?
I’m saying, “Successful theory of physics A was simple, and theory B was simple, and C was simple, and D, and E, and …. , but there’s a new class of phenomena which needs a new theory, so this theory will probably also be simple.”
This would be a lot more convincing if the most recent and most successful theory of physics weren’t such a glaring counter-example. We count on induction because we have no other option but this kind of thing is meta-induction (that changes in our understanding of regularities has regularities) hasn’t been justified enough to make it a tool in eliminating approaches.
“Approach” here is a HUGE misnomer. “Approach” is a term commonly used in engineering to mean “different ways of accomplishing goal X”. You can build a machine in manner A to do X, or manner B. This ABSOLUTELY DOES NOT generalize to science, because there’s only ever one reality. If you have theory A and theory B both purporting to explain some phenomenon X, either A or B must be wrong, while in engineering sometimes there are ten different ways of attacking a problem, depending on what your goals are.
I disagree. Even under the naive theory of truth that is popular here we don’t know in advance which theoretical apparatus will yield the right explanation for phenomenon X. While that is still an open question people might come to different conclusions about which way is most promising. As agreement here seems unlikely it makes sense to just have physicists work in the areas they think will be most fruitful. And then once you notice that theories of physics have this nasty habit of turning out false… well then I don’t even know what you’re using to declare A right and B wrong. Yeah “A or B must be wrong” but that is a seriously inclusive or. When we don’t know which theory is right, when they’re both probably wrong or if that turns out to be a nonsense question since they’re empirically equivalent it then makes a lot of sense to think about the benefits working under different sets of theoretical assumptions (i.e. approaches). That isn’t a prescription for anything goes. Some approaches are stupid and laughable. Others are powerful and clever. So I think “approach” nomes just fine.
“Maybe, but they’re still both false! What exactly is the distinction you have in mind?”
Yes, and me and the Pacific Ocean are both more than 50% water by mass. Newtonian gravitation successfully explained a huge number of phenomena. Phlogiston did not.
“Anyway, I had just meant LET versus SR, but what exactly is the experimental evidence against an aether?”
With quantum mechanics (and modern experimental technology), we can actually look down below the level of individual particles, and we have found that photons are actually their own particles, not patterns of vibration (or whatever) within other particles.
“This would be a lot more convincing if the most recent and most successful theory of physics weren’t such a glaring counter-example.”
“I disagree. Even under the naive theory of truth that is popular here”
Would you care to propose some alternative theory of truth?
“While that is still an open question people might come to different conclusions about which way is most promising.”
Obviously, but that disagreement should then be resolved by reference to experiment. There is no room for persistent disagreement. In engineering, you can have five different methods, each with their own advantages and disadvantages, and this is a stable state. In science, having five different theories is not a stable state; it needs to be resolved, rather than harden into different factions.
“And then once you notice that theories of physics have this nasty habit of turning out false… well then I don’t even know what you’re using to declare A right and B wrong.”
Experimental evidence?
“it then makes a lot of sense to think about the benefits working under different sets of theoretical assumptions (i.e. approaches).”
What does that even mean? How would you apply that to a theory of physics (past or present)? What “theoretical assumptions” are involved in, say, Special Relativity? Special Relativity makes the assertion that the speed of light is constant regardless of reference frame, but this isn’t just a mathematical axiom that you can pick up and discard at will; it is based on a huge pile of experimental evidence.
With quantum mechanics (and modern experimental technology), we can actually look down below the level of individual particles, and we have found that photons are actually their own particles, not patterns of vibration (or whatever) within other particles.
We knew this before quantum mechanics. Lorentz’s aether wasn’t matter.
I’ve read (indeed, I own) this book. I don’t know how to evaluate elegance, but the Standard Model particle zoo isn’t simple, at least not in the way Newton or special relativity is simple. I wish you would assume I have some idea what I’m talking about and that my concerns and questions might be well motivated, however alien they seem to you.
Would you care to propose some alternative theory of truth?
I would like too but I haven’t really figured out what I think. I’m not particularly radical, but I think the epistemology Eliezer has laid out on Less Wrong has some holes, leaves important questions unanswered etc. Maybe one day I’ll write something.
Obviously, but that disagreement should then be resolved by reference to experiment. There is no room for persistent disagreement. In engineering, you can have five different methods, each with their own advantages and disadvantages, and this is a stable state. In science, having five different theories is not a stable state; it needs to be resolved, rather than harden into different factions.
But they aren’t always resolved that way! Special relativity beat out Lorentzian Ether Theory even though they are empirically equivalent. Obviously if you have two theories that predict different outcomes of some feasible experiment you can run that experiment and resolve the difference. The question is, what we do when there is no experiment to run? Well we do some math and try to come up with testable hypotheses. Using different theories or different vocabulary seems to affect how easy it is to do the math and generate hypotheses.
“And then once you notice that theories of physics have this nasty habit of turning out false… well then I don’t even know what you’re using to declare A right and B wrong.”
Experimental evidence?
As the sentence immediately following this one was supposed to indicate, usually both theories are wrong in the long run.
What does that even mean? How would you apply that to a theory of physics (past or present)? What “theoretical assumptions” are involved in, say, Special Relativity? Special Relativity makes the assertion that the speed of light is constant regardless of reference frame, but this isn’t just a mathematical axiom that you can pick up and discard at will; it is based on a huge pile of experimental evidence.
It isn’t based on any experimental evidence that distinguishes it from a theory that says mass contracts in the direction it moves in. But it turns out that if you start from a theory which says’s light’s speed is constant you can come up with things like the theory of General Relativity. Alternately, you might have two theories that describe totally different phenomena without error but have the potential to describe things about other phenomena and eventually one might be subsumed under the other. But it won’t always be obvious which theory is the more fundamental one. I suspect one reason there is a lot of work done trying to incorporate gravity into quantum mechanics rather than the other three fundamental forces into General Relativity is that SR/GR just doesn’t have the vocabulary to make hypotheses about particle physics. The former is sort of obvious though “Oh there is another force, there must be this other wavicle: a graviton.” That doesn’t mean it will be an successful approach but that is part of the reason it is the popular one.
Special relativity beat out Lorentzian Ether Theory even though they are empirically equivalent. Obviously if you have two theories that predict different outcomes of some feasible experiment you can run that experiment and resolve the difference.
These two theories are not equivalent at all; they predict different outcomes for the Michelson-Morley experiment, unless you patch up the ether by requiring it to be at rest relative to the Earth at all times.
Would the person or persons down voting my comments here mind at least explaining this one? Afaict I’m just correcting someone on a factual error and providing a citation. If wikipedia is wrong on this matter I’d like to know.
The problem with that solution is that you end up with a theory that makes the same predictions as SR, but contains an extra concept that plays no role in making predictions: a distinguished but unobservable reference frame of absolute rest, with respect to which all the other reference frames involve time dilations and length contractions. When you cut off that useless spinning cog you are left with SR.
Even under the naive theory of truth that is popular here
Which one is that? Do you mean naive in the sense that it is ‘unsophisticated’ or do you mean ‘actually wrong because it is too simple’? If the latter you could well be right but I’d like more information.
I do think “A true theory is a map that corresponds to the territory” is right as far as it goes. But there are going to be people who will ask things like “What the hell do you mean by territory?” and “How do you have any idea if your map corresponds to the territory?”. I don’t think those are wrong questions and I think answering them might require a little more work.
That said I might be missing where people are at on this because this line, from the wiki, is exactly right:
Since our predictions don’t always come true, we need different words to describe the thingy that generates our predictions and the thingy that generates our experimental results. The first thingy is called “belief”, the second thingy “reality”.
I think stopping there is about right but “reality” tends to get loaded with a bunch of additional properties. I think attributing features of your theory to reality other than it’s experimental predictions is a kind of map-territory confusion that nearly everyone still falls for, so I think looking at two empirically equivalent theories and saying one is true and the other is false can’t just mean “one matches the territory and the other doesn’t”. So either we say of these theories that they are both true but one is better for reasons other than truth or we say that truth involves something other than just corresponding to the territory.
A related issue is that it is unlikely the core concepts we need to state any theory themselves correspond exactly to an external world. A mathematical description is fine by itself but we always feel that merely stating the math is some how insufficient so we end up reifying our variables. At least when the object of the theory is significantly divorced from the environment that gave rise to the concepts used to state the theory it is extremely unlikely that these concepts map exactly to things in the the world. This doesn’t always change our predictions but it suggests that of two empirically identical theories neither is likely to conceptually correspond to an external world (the external world, if there is one, isn’t made up of concepts) but both will correspond empirically. So then how can we contrast these theories on the basis of their correspondence to reality?
Well I don’t know if “the external world is made of concepts” is a meaningful thing to say. It strikes me as a kind of category error. But yes, I can always be wrong. I actually had a hedge in there original but took it out for stylistic reasons since it was already a parenthetical.
Well I don’t know if “the external world is made of concepts” is a meaningful thing to say.
It probably depends on what we mean by ‘concept’. The laws of physics can probably be described as concepts and if we knew all the laws of physics the concepts would then be quite likely to include the entire universe as artifacts. But yes, parenthetical to the extreme.
Eliezer Yudkowsky stated that he “meant to restore a naive view of truth” in his essay The Simple Truth—I believe the more technical term is “correspondence theory of truth”.
“Wait, a theory was falsified by the invention of another theory?”
One that’s been confirmed against hundreds of predictions to over a dozen decimal places? Yes.
“Why put false in quotes?”
Because there’s a HUGE distinction between a theory like Newtonian gravitation and a “theory” like phlogiston, even if they’re both “false”.
“So you had a simple theory that explained some class of phenomena. Now there is some other phenomena that your theory fails to explain, why should simplicity be conserved here?”
I’m saying, “Successful theory of physics A was simple, and theory B was simple, and C was simple, and D, and E, and …. , but there’s a new class of phenomena which needs a new theory, so this theory will probably also be simple.”
“But the existence of competing approaches with varying degrees of popularity does not entail that.”
“Approach” here is a HUGE misnomer. “Approach” is a term commonly used in engineering to mean “different ways of accomplishing goal X”. You can build a machine in manner A to do X, or manner B. This ABSOLUTELY DOES NOT generalize to science, because there’s only ever one reality. If you have theory A and theory B both purporting to explain some phenomenon X, either A or B must be wrong, while in engineering sometimes there are ten different ways of attacking a problem, depending on what your goals are.
(Or equivalent in a way you haven’t understood yet.)
Is there a single example of this that you can think of? There are the different ways of computing classical mechanics (Newtonian, Lagrangian, Hamiltonian), but these were known to be just different ways of doing the same math at the time of discovery.
The original formulations of QM were famously shown to be equivalent, though I’m not sure they were ever expected to be incompatible. QFT and S-matrix theory were originally politically opposed (though QFT produces an S-matrix), but I have heard that recently people advocate widening QFT to the point that it appears to cover all of S-matrix theory.
No, it’s just a theoretical property of the ‘A, B, X’ abstraction that you mention. In fact, it would not surprise me if the general problem of proving whether or not two theories are exactly equivalent is intractable in a similar way to the halting problem.
I was just clarifying, you said that it was ruled out “as soon as quantum mechanics was invented” which clearly isn’t right. Anyway, I had just meant LET versus SR, but what exactly is the experimental evidence against an aether? Obviously QM gets along pretty well without it and the vocabulary is incompatible… but the vocabulary of QM is incompatible with the vocabulary relativity as well.
Maybe, but they’re still both false! What exactly is the distinction you have in mind?
This would be a lot more convincing if the most recent and most successful theory of physics weren’t such a glaring counter-example. We count on induction because we have no other option but this kind of thing is meta-induction (that changes in our understanding of regularities has regularities) hasn’t been justified enough to make it a tool in eliminating approaches.
I disagree. Even under the naive theory of truth that is popular here we don’t know in advance which theoretical apparatus will yield the right explanation for phenomenon X. While that is still an open question people might come to different conclusions about which way is most promising. As agreement here seems unlikely it makes sense to just have physicists work in the areas they think will be most fruitful. And then once you notice that theories of physics have this nasty habit of turning out false… well then I don’t even know what you’re using to declare A right and B wrong. Yeah “A or B must be wrong” but that is a seriously inclusive or. When we don’t know which theory is right, when they’re both probably wrong or if that turns out to be a nonsense question since they’re empirically equivalent it then makes a lot of sense to think about the benefits working under different sets of theoretical assumptions (i.e. approaches). That isn’t a prescription for anything goes. Some approaches are stupid and laughable. Others are powerful and clever. So I think “approach” nomes just fine.
(fyi, the downvote isn’t mine)
“Maybe, but they’re still both false! What exactly is the distinction you have in mind?”
Yes, and me and the Pacific Ocean are both more than 50% water by mass. Newtonian gravitation successfully explained a huge number of phenomena. Phlogiston did not.
“Anyway, I had just meant LET versus SR, but what exactly is the experimental evidence against an aether?”
With quantum mechanics (and modern experimental technology), we can actually look down below the level of individual particles, and we have found that photons are actually their own particles, not patterns of vibration (or whatever) within other particles.
“This would be a lot more convincing if the most recent and most successful theory of physics weren’t such a glaring counter-example.”
You mean quantum mechanics? Quantum mechanics is very elegant, it’s just usually explained badly. See http://www.amazon.com/QED-Strange-Theory-Light-Matter/dp/0691024170.
“I disagree. Even under the naive theory of truth that is popular here”
Would you care to propose some alternative theory of truth?
“While that is still an open question people might come to different conclusions about which way is most promising.”
Obviously, but that disagreement should then be resolved by reference to experiment. There is no room for persistent disagreement. In engineering, you can have five different methods, each with their own advantages and disadvantages, and this is a stable state. In science, having five different theories is not a stable state; it needs to be resolved, rather than harden into different factions.
“And then once you notice that theories of physics have this nasty habit of turning out false… well then I don’t even know what you’re using to declare A right and B wrong.”
Experimental evidence?
“it then makes a lot of sense to think about the benefits working under different sets of theoretical assumptions (i.e. approaches).”
What does that even mean? How would you apply that to a theory of physics (past or present)? What “theoretical assumptions” are involved in, say, Special Relativity? Special Relativity makes the assertion that the speed of light is constant regardless of reference frame, but this isn’t just a mathematical axiom that you can pick up and discard at will; it is based on a huge pile of experimental evidence.
I wouldn’t be so sure. (http://www.jimloy.com/physics/phlogstn.htm) But it certainly had other problems.
We knew this before quantum mechanics. Lorentz’s aether wasn’t matter.
I’ve read (indeed, I own) this book. I don’t know how to evaluate elegance, but the Standard Model particle zoo isn’t simple, at least not in the way Newton or special relativity is simple. I wish you would assume I have some idea what I’m talking about and that my concerns and questions might be well motivated, however alien they seem to you.
I would like too but I haven’t really figured out what I think. I’m not particularly radical, but I think the epistemology Eliezer has laid out on Less Wrong has some holes, leaves important questions unanswered etc. Maybe one day I’ll write something.
But they aren’t always resolved that way! Special relativity beat out Lorentzian Ether Theory even though they are empirically equivalent. Obviously if you have two theories that predict different outcomes of some feasible experiment you can run that experiment and resolve the difference. The question is, what we do when there is no experiment to run? Well we do some math and try to come up with testable hypotheses. Using different theories or different vocabulary seems to affect how easy it is to do the math and generate hypotheses.
As the sentence immediately following this one was supposed to indicate, usually both theories are wrong in the long run.
It isn’t based on any experimental evidence that distinguishes it from a theory that says mass contracts in the direction it moves in. But it turns out that if you start from a theory which says’s light’s speed is constant you can come up with things like the theory of General Relativity. Alternately, you might have two theories that describe totally different phenomena without error but have the potential to describe things about other phenomena and eventually one might be subsumed under the other. But it won’t always be obvious which theory is the more fundamental one. I suspect one reason there is a lot of work done trying to incorporate gravity into quantum mechanics rather than the other three fundamental forces into General Relativity is that SR/GR just doesn’t have the vocabulary to make hypotheses about particle physics. The former is sort of obvious though “Oh there is another force, there must be this other wavicle: a graviton.” That doesn’t mean it will be an successful approach but that is part of the reason it is the popular one.
These two theories are not equivalent at all; they predict different outcomes for the Michelson-Morley experiment, unless you patch up the ether by requiring it to be at rest relative to the Earth at all times.
Could you be missing the “Lorentzian” there?
What FAWS said. The problem is solved with length contraction.
Would the person or persons down voting my comments here mind at least explaining this one? Afaict I’m just correcting someone on a factual error and providing a citation. If wikipedia is wrong on this matter I’d like to know.
The problem with that solution is that you end up with a theory that makes the same predictions as SR, but contains an extra concept that plays no role in making predictions: a distinguished but unobservable reference frame of absolute rest, with respect to which all the other reference frames involve time dilations and length contractions. When you cut off that useless spinning cog you are left with SR.
Which one is that? Do you mean naive in the sense that it is ‘unsophisticated’ or do you mean ‘actually wrong because it is too simple’? If the latter you could well be right but I’d like more information.
Naive as in “This essay is meant to restore a naive view of truth.”.
I do think “A true theory is a map that corresponds to the territory” is right as far as it goes. But there are going to be people who will ask things like “What the hell do you mean by territory?” and “How do you have any idea if your map corresponds to the territory?”. I don’t think those are wrong questions and I think answering them might require a little more work.
That said I might be missing where people are at on this because this line, from the wiki, is exactly right:
I think stopping there is about right but “reality” tends to get loaded with a bunch of additional properties. I think attributing features of your theory to reality other than it’s experimental predictions is a kind of map-territory confusion that nearly everyone still falls for, so I think looking at two empirically equivalent theories and saying one is true and the other is false can’t just mean “one matches the territory and the other doesn’t”. So either we say of these theories that they are both true but one is better for reasons other than truth or we say that truth involves something other than just corresponding to the territory.
A related issue is that it is unlikely the core concepts we need to state any theory themselves correspond exactly to an external world. A mathematical description is fine by itself but we always feel that merely stating the math is some how insufficient so we end up reifying our variables. At least when the object of the theory is significantly divorced from the environment that gave rise to the concepts used to state the theory it is extremely unlikely that these concepts map exactly to things in the the world. This doesn’t always change our predictions but it suggests that of two empirically identical theories neither is likely to conceptually correspond to an external world (the external world, if there is one, isn’t made up of concepts) but both will correspond empirically. So then how can we contrast these theories on the basis of their correspondence to reality?
… well, maybe...
Well I don’t know if “the external world is made of concepts” is a meaningful thing to say. It strikes me as a kind of category error. But yes, I can always be wrong. I actually had a hedge in there original but took it out for stylistic reasons since it was already a parenthetical.
It probably depends on what we mean by ‘concept’. The laws of physics can probably be described as concepts and if we knew all the laws of physics the concepts would then be quite likely to include the entire universe as artifacts. But yes, parenthetical to the extreme.
Eliezer Yudkowsky stated that he “meant to restore a naive view of truth” in his essay The Simple Truth—I believe the more technical term is “correspondence theory of truth”.