I What tests can falsify general relativity?

[francis20520]

TL;DR Summary

falsification

I know that general relativity fails in singularities like the center of a black hole or the big bang.

GR also fails at fundamental particle levels like electrons, protons and neutrons etc. I.e. GR cannot explain interactions of various fundamental particles?? (Am I correct?)

But these does not make GR wrong because it works every where else. Above things don't falsify GR, right?

But my question is what test will falsify GR?

Like, what experimental setup can be done (theoretically) that if successful will prove GR wrong?

I.e. What are the falsification tests people have come up for GR?

 

[Ibix]

Loads of tests have been done. Gravitational deflection of light. Gravitational waves. Motions of planets. Redshift of distant galaxies. Tick rates of clocks at different altitudes. Frame dragging. Probably others that haven't occurred to me off the top of my head. Any one of these could have falsified general relativity, but has not done so.

On a point of language, I would say that what you seem to be talking about isn't falsification, precisely. It's finding the limits of applicability of GR as an approximation to a more fundamental theory. Falsification would be finding something right in the middle of GR's "known valid" regime that didn't behave as predicted.

GR keeps passing the tests we can do, in that its predictions match measurement to available precision. We are pretty sure that it's not quite right, that it's an approximation to a more general theory, but we can't reach a regime where it is measurably inaccurate, so we cannot gather data to help develop it.

 

[Ibix]

I see I haven't quite answered your question.

Better gravitational wave astronomy might show differences between the detected waves and the predicted ones in the final stages of a black hole merger, which is one way of probing near black hole event horizons. I believe that there are studies in the works to see how particles in superposed states fall under gravity, which might yield some information about gravity and quantum states.

 

[pervect]

Summary:: falsification

I know that general relativity fails in singularities like the center of a black hole or the big bang.

GR also fails at fundamental particle levels like electrons, protons and neutrons etc. I.e. GR cannot explain interactions of various fundamental particles?? (Am I correct?)

But these does not make GR wrong because it works every where else. Above things don't falsify GR, right?

But my question is what test will falsify GR?

Like, what experimental setup can be done (theoretically) that if successful will prove GR wrong?

I.e. What are the falsification tests people have come up for GR?

I'd suggest Will's review paper, "The confrontation between General Relativity and Experiment", https://arxiv.org/abs/1403.7377

There's too many tests to summarize them all, but I'll give a few.

The weak equivalence principle can be detected by testing that objects of different compositions all "fall at the same rate", and by Eotovos type experiments to measure the gravitational attraction of materials of different composition. If materials of different composition but identical masses attracted each other differently, GR would be falsified.

Tests that special relativity holds locally also test GR - for instance, the Michelson Morely experiment. Wills describes this as "tests of local Lorentz invariance". So if any of the standard SR tests of the speed of light failed, GR would also be falsified.

Gravitational redshift is another test of GR, though many other theories predict this phenomenon. Wills calsifies Pound-Rebka experiments in the same general category here.

To compare General Relativity with other metric theories of gravity, various tests using the PPN formalism have been done. The commonly known tests here are the deflection of light, and the "Shapiro time delay" tests.

The perihelion shift of mecury's orbit is another test that fits into the PPN framework.

The general idea is that metric theories of gravity have a set of parameters that can be measured (beta, gamma, and others), measuring the values of these parameters distinguishes GR from other metric theories such as Branse-Dicke gravity.

Gravity probe B tested for gravitomagnetic effects predicted by GR.

Moving onto strong field tests, the Ligo experiment detecting gravitational waves that fit with the predictions of GR is a good one, and we have some inspiral measurements that also indirectly indicate gravitational radiation of the expected amount.

 

[atyy]

I know that general relativity fails in singularities like the center of a black hole or the big bang.

Yes.

GR also fails at fundamental particle levels like electrons, protons and neutrons etc. I.e. GR cannot explain interactions of various fundamental particles?? (Am I correct?)

A quantum generalization of GR works for fundamental particles like electrons, protons, neutrons etc. These are all still at relatively low energies compared to where we think the quantum generalization of GR will fail. Quantum GR does not explain their interactions, but can be integrated with the standard model of particle physics that describes their interactions.

But my question is what test will falsify GR?

In addition to the experiments mentioned by @Ibix and @pervect, some ideas for tests that have not yet been done to test GR more stringently are discussed in https://arxiv.org/abs/0903.0100.

 

[martinbn]

I know that general relativity fails in singularities like the center of a black hole or the big bang.

No, GR predicts them. It doesn't fail at them. So far there are no obeservations or experiments about the singularities to see if GR fails or not.

GR also fails at fundamental particle levels like electrons, protons and neutrons etc. I.e. GR cannot explain interactions of various fundamental particles?? (Am I correct?)

GR is not a theory of everything, so it doesn't fail here either.

But these does not make GR wrong because it works every where else. Above things don't falsify GR, right?

But my question is what test will falsify GR?

Like, what experimental setup can be done (theoretically) that if successful will prove GR wrong?

I.e. What are the falsification tests people have come up for GR?

Any test whose outcome differes from the predictions of GR will show the limits og GR. So far GR passes all test.

 

[vanhees71]

I'd say, unavoidable singularities of a theory show clearly limitations of that theory.

 

[martinbn]

I'd say, unavoidable singularities of a theory show clearly limitations of that theory.

Not if the singularities are part of nature, then the theory makes good predictions.

 

[vanhees71]

Hm, well. How do you measure whether there's a true singularity? Maybe the singularities are all hidden behind a horizon (cosmic censorship). Then they wouldn't be observable at all and you could never falsify their existence, but we are drifting into philosophical questions which never have a definite answer ;-)).

 

[martinbn]

Hm, well. How do you measure whether there's a true singularity? Maybe the singularities are all hidden behind a horizon (cosmic censorship). Then they wouldn't be observable at all and you could never falsify their existence, but we are drifting into philosophical questions which never have a definite answer ;-)).

May be, but you simply reject them because of your philosophy, not scientific evidence.

 

[vanhees71]

True.

 

[Nugatory]

May be, but you simply reject them because of your philosophy, not scientific evidence.

There’s nothing wrong with rejecting an idea on the basis of philosophy as long as we know that’s what we’re doing. It’s how we keep our thought processes from being gummed up by an accretion of unfalsifiable ideas.

(This is commentary not argument)

 

[eaglechief]

If someone could show that velocity of light is a parameter in time and not a constant, than STR would be falsified and GTR, as well ... Is that assupmtion correct ?

 

[Nugatory]

If someone could show that velocity of light is a parameter in time and not a constant, than STR would be falsified and GTR, as well ... Is that assupmtion correct ?

Yes.
However it is also true that if someone could show that I had wings I would be able to fly; that true statement is of little relevance when I'm trying to get up onto the roof without a ladder. The equally true statement that relativity would be wrong if the speed of light were not constant is of equally little relevance when we live in a universe in which the speed of light is constant.

 

[Ibix]

Yes.

Do you think so? I was going to say no, because local Lorentz invariance requires that $c$ be invariant, not constant.

 

[eaglechief]

at least, wiki says so ...

"The speed of light in vacuum, commonly denoted c, is a universal physical constant important in many areas of physics. "

 

[Nugatory]

Do you think so? I was going to say no, because local Lorentz invariance requires that $c$ be invariant, not constant.

I think that we're in a conversation that isn't far enough along to need that distinction yet, and that OP used "constant" when "invariant" might have been a better word.

 

[Ibix]

I think that we're in a conversation that isn't far enough along to need that distinction yet, and that OP used "constant" when "invariant" might have been a better word.

Not sure I agree. The question seemed to me to be asking what if $c$ varied over time. I think a measurement that concluded that $c$ (or $\alpha$, although happy to duck that distinction at this point) was different in the past wouldn't bother relativity. One that showed that $c$ was different for different local frames would be a problem, I agree.

 

[atyy]

Not if the singularities are part of nature, then the theory makes good predictions.

Classically, that is true. But I don't think it's true quantum mechanically, where we can treat general relativity as an effective field theory at low energies, but it's not even clear what we get if the energy gets high enough (string theory, asymptotic safety, fundamental failure of local Lorentz invariance?).

 

[atyy]

Hm, well. How do you measure whether there's a true singularity? Maybe the singularities are all hidden behind a horizon (cosmic censorship). Then they wouldn't be observable at all and you could never falsify their existence, but we are drifting into philosophical questions which never have a definite answer ;-)).

Maybe one of these theories with extra dimensions will be correct, then there could be a failure of cosmic censorship
https://www.livescience.com/53857-5d-black-holes-break-general-relativity.html
https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.116.071102

 

[martinbn]

Classically, that is true. But I don't think it's true quantum mechanically, where we can treat general relativity as an effective field theory at low energies, but it's not even clear what we get if the energy gets high enough (string theory, asymptotic safety, fundamental failure of local Lorentz invariance?).

What makes you think that? Any theorems in that direction?

 

[martinbn]

Maybe one of these theories with extra dimensions will be correct, then there could be a failure of cosmic censorship
https://www.livescience.com/53857-5d-black-holes-break-general-relativity.html
https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.116.071102

This is gramatically way too complicated for me. What is the proper term for conditional sentences of the form "If extra dimensions then possibly there might be numerical evindence that could perhaps be interpreted as suggesting that WCC may not be true."

 

[jbriggs444]

at least, wiki says so ...

"The speed of light in vacuum, commonly denoted c, is a universal physical constant important in many areas of physics. "

Since it is not a unitless constant, it is as much conventional as physical. We can change its value over time with a flick of a pencil and eraser. And we have done so.

 

[pervect]

What definitions of the distance (say meters) and time (say seconds) are being used by the Original Poster (OP)? We can conclude that they are not using the most modern definitions, because they are interested in experimentally measuring the speed of light, and that's not compatible with the modern SI definition of the meter and second. So presumably the OP is using some other definition. I've found it unwise to make assumptions about what what defintitions people might be using in such threads.

Unfortunately there's no meaningful answer to experimentally measuring the speed of light without knowing the experimental basis of the distances and time standards being used. And the only conclusion we can be certain of is that they're not the modern one , which goes as follows;

The metre, symbol m, is the SI unit of length. It is defined by taking the fixed numerical value of the speed of light in vacuum c to be 299 792 458 when expressed in the unit m s–1, where the second is defined in terms of the caesium frequency Cs.

So if the OP was using the modern standard, they wouldn't be talking about measuring the speed of light, because it is defined as a constant. So they are implicitly using some other definiton. We need to know what that is to attempt to answer the question.

The dimensionless approach would be to talk about the fine structure constant instead of the speed of light but we've had those discussions in the past and gotten "blank stares". I'm still willing to have such a discussion if the OP is interested and feels it's relevant, though.

Otherwise, we need to get into the nitty-gritty of the time and distance standards to attempt to provide an answer.

 

[Vanadium 50]

I am disappointed in the direction this thread has turned down.

Suppose I built a box with an LED, a mirror and a light sensor and set it up to look at the delay between emission and detection, and from that calculate the speed of light. Every second it blinks 300000. Now suppose one morning it, and every box in the universe like it, started blinking 600000. It doesn't matter if what "really changed" was the speed of light, our time standard, our length standard, or some combination. SR would be in trouble.

But the whole idea is immensely silly, and in my view should not have been brought up instead. SR is an amazingly well-tested theory - much better tested than, for example, the theory that Covid is caused by the Coronavirus - and bringing it up to the exclusion of real tests of GR is an ill-advised and ill-founded distraction. One might as well ask if the sudden appearance of Zeus with his energy-non-conserving thunderbolts would falsify GR.

There are at least ten parameters that can be measured and have values predicted by GR and with different values predicted by alternatives. As an example, the amount of space curvature (not spacetime curvature) is smaller in scalar-tensor theories than in GR.

 

[vanhees71]

Well, that example would lead us again to the unfruitful discussions of one-path light speed. As discussed at length recently, for that you need clock synchronization conventions, as was realized by Einstein already in 1905 when writing his famous paper on special relativity.

 

[Vanadium 50]

that example would lead us again to the unfruitful discussions of one-path light speed.

If I measure 300000 today and 600000 tomorrow with the same device, that's a problem. Synchronization conventions do not affect the number being read out.

 

[jbriggs444]

If I measure 300000 today and 600000 tomorrow with the same device, that's a problem. Synchronization conventions do not affect the number being read out.

You measure 300000 for what, exactly? A count per unit time? With time measured based on the period of the radiation associated with...? Perhaps the dimensions of the box changed. That does not affect special relativity.

If we want to avoid a debate on whether the speed of light changed or the size of the box changed, we get right back to worrying about unit definitions, the fine structure constant and rewording the question.

[Though I do understand a sentiment that these quibbles feel like cheats designed to avoid the thrust of the OP's question]

 

[Vanadium 50]

I am not going to answer. It's pointless quibbling, and a distraction from the question at hand. We could equally well argue about the color of Zeus' thunderbolts.

If you think less of me because I couldn't come up with an unquibbleable example in a paragraph, well, so be it.

 

[jbriggs444]

If you think less of me because I couldn't come up with an unquibbleable example in a paragraph, well, so be it.

No, no. I get where you are coming from and do not disagree. I'll be quiet now.

 

[PeterDonis]

You measure 300000 for what, exactly?

He means the device is measuring the speed of light by some well-defined process--for definiteness, say it's a light source, a mirror, and an atomic clock--and outputting the result in meters per second.

If nothing changes about how the device is constructed from today to tomorrow, then our understanding of SR says the result it outputs should not change from today to tomorrow. So if it did, that would mean something was wrong with our understanding of SR. That's what I take @Vanadium 50 to be saying.

 

[pervect]

I am disappointed in the direction this thread has turned down.

Suppose I built a box with an LED, a mirror and a light sensor and set it up to look at the delay between emission and detection, and from that calculate the speed of light. Every second it blinks 300000. Now suppose one morning it, and every box in the universe like it, started blinking 600000. It doesn't matter if what "really changed" was the speed of light, our time standard, our length standard, or some combination. SR would be in trouble.

Your box (es) is (are) made out of atoms, I presume. Following this line of reasoning leads one to the realization that one is actually experimentally concerned with the fine structure constant of said atoms. As you say, it doesn't matter whether the atoms are changing, or the speed of light is changing, something is happening and we can determine it experimentally. The dimensionlessness of the fine structure constant is what allows us to unequivocally say that something is changing, that we're not playing any such games.

Without knowing that your box is made up of atoms, though, one could be using any number of different notions of distance. For instance, one might have a long physical meter stick as they did in ages past, and perhaps this "standard" is loosing atoms over time via various mechanisms included sublimation. Then it would be a prediction of SR based on this standard, that the speed of light is slowly changing over time. What happens to the speed of light when your standard meter stick totally evaporates is an interesting question, which I won't attempt to answer, as I believe that using a physical meter stick is an inferior standard, though at one time it was the best we had.

Without knowing that your box is made out of atoms, and that you are implicitly relying on the properties of atoms to determine both your time scale (via the cesium atoms), and the length of your box, one really can't say what is changing, and the focus gets lost.

One reason to talk about the fine structure constant is the notion that boxes and other "physical" objects are made out of atoms, and that if we are assuming that those atoms are the basis of our time and distance standards, then the appropriate language to use for the speed of light varying is the fine structure constant is varying. There is no ambiguity in that, because the fine structure constant is dimensoinless.

And it saves a lot of wrangling.

However, we do not know if the Original Poster has the notion that distance and time standards are based on the properties of atoms. Most likely, they haven't actually thought about the issue at all. If they're willing to accept the idea that atoms are the basis of distance and time standards, though, there isn't any real problem that can't be gotten around. And this route leads to the fine structure constant being the interesting physical quantity, because it's dimensionless, and no amount of fiddling with units can change the value of a dimensionless constant. Problem solved.

 

[vanhees71]

If I measure 300000 today and 600000 tomorrow with the same device, that's a problem. Synchronization conventions do not affect the number being read out.

Of course they do. You need two clocks at different points to measure a one-way speed of whatever, including the speed of a light signal. The standard Einstein synchronization is for a (thought) set of clocks relative at rest in an inertial reference frame (SR). In this sense the one-way speed of light is set to $c$ by defining the corresponding Galilean (pseudo-Euclidean) coordinates. Already synchronized clocks of one such global inertial frame are not synchronized with the synchronized clocks of another such frame moving with constant speed relative to the former.

In GR you can establish such a clock-synchronization only locally, and the issue becomes more complicated.

 

[eaglechief]

So if the OP was using the modern standard, they wouldn't be talking about measuring the speed of light, because it is defined as a constant. So they are implicitly using some other definiton. We need to know what that is to attempt to answer the question.

Thanks for the answers but i think my point concerning velocity of light was misunderstood.
It is currently described and agreed as a constant in space, but it is not generally accepted that c is a constant in time, as well, as several VSL-approaches of the last years show.

Wikipedia VSL-Theories

Even Albert Einstein was discussing that in 1911 ...

 

[Olorin]

Wouldn’t an observational evidence of Strong Equivalence Principle violations be a proper falsification of General Relativity? It seems that GR is the only metric theory of gravity which relies on the SEP. And we have a recent study that casts a lot of trouble for the SEP: https://arxiv.org/abs/2009.11525. If confirmed, do we have falsification?

 

[PeterDonis]

it is not generally accepted that c is a constant in time

Yes, it is. Scientists have looked very carefully for evidence of variation in time (not of $c$ but of the fine structure constant, which is the meaningful variation to look for) and have not found any:

https://en.wikipedia.org/wiki/Fine-structure_constant#Past_rate_of_change

Wikipedia VSL-Theories

None of the proposals described in that article have gained any acceptance. Note also that some of them really mean violation of Lorentz invariance (instead of variation of the fine structure constant) when they say "variation of c with time"; scientists have also looked very carefully for evidence of violations of Lorentz invariance and have not found any:

https://en.wikipedia.org/wiki/Modern_searches_for_Lorentz_violation

 

[pervect]

Thanks for the answers but i think my point concerning velocity of light was misunderstood.
It is currently described and agreed as a constant in space, but it is not generally accepted that c is a constant in time, as well, as several VSL-approaches of the last years show.

Wikipedia VSL-Theories

Even Albert Einstein was discussing that in 1911 ...

I think it is my point that's not been understood here. At a minimum, it's not been acknowledged. Perhaps you could quote some references that suggest that the speed of light is "not generally accepted" as "constant with time", and, to get back to the point I made, describe in principle how you are measuring the speed of light. To measure the speed of light, you need two standards, one for time, one for space. And you haven't told us what these standards are even in principle. That was basically the point, and I have no idea how you think that the speed of light should be measured, and why you think the standards that you are using to measure the speeds themselves are "constant with time".

I can say that if you look at the actual language of the current version of the standards, there's no mention of the speed of light changing with time, and is defined as a constant.

See for instance (again) https://www.bipm.org/metrology/length/units.html

The metre, symbol m, is the SI unit of length. It is defined by taking the fixed numerical value of the speed of light in vacuum c to be 299 792 458 when expressed in the unit m s–1, where the second is defined in terms of the caesium frequency

Cs.

It's not my intent to say that you absolutely must adopt the current generally accepted and well documented standard, set by the body (the bipm) that is charged with the task of setting standards, but given that you apparently do not, in order to have a meaningful conversation you would at least have to tell us what standard(s) you are using.

For instance, you might be using the old meter prototype standard, or some variant therof that has been idealized.

 

[PeterDonis]

we have a recent study that casts a lot of trouble for the SEP

That study is not investigating the SEP directly. It's investigating a claimed "External Field Effect" in the context of MOND. So you have to be careful about taking its claims of SEP "violation" at face value. There is quite a bit of model-dependent interpretation involved.

All the paper is really claiming is "if we use our model of galaxy rotation curves, it matches the data better for these particular galaxies than the $\Lambda C D M$ model". But a model has to account for all of the data, not just data for one particular set of galaxies or data that are claimed to illustrate one particular group of theorists' ideas about one particular claimed effect.

 

[Olorin]

That study is not investigating the SEP directly. It's investigating a claimed "External Field Effect" in the context of MOND.

But if the SEP holds true, neighbouring objects should not exert any gravitational influence on the studied galaxies and thus should not have any dynamical effects on them. But the contrary is indeed claimed to be observed. The fact that this is done in the context of MONDIAN dynamics only further emphasizes the fact that such an effect is manifest. The study is definitely a test of the SEP. Furthermore, we need to bear in mind that the lead author was a proponent of LCDM and was convinced that his data would falsify MOND once and for all and ended up finding that the opposite was truth. That’s something worth noticing.

But a model has to account for all of the data, not just data for one particular set of galaxies or data that are claimed to illustrate one particular group of theorists' ideas about one particular claimed effect.

Exactly’ and that’s exactly what MOND has been doing recently. LCDM not so much. 2020 has been a pretty amazing year for MONDIAN astrophysics. See this for example: https://darkmattercrisis.wordpress.com/2020/11/10/the-crisis-in-cosmology-is-now-catastrophic/.

This review about MOND consecutive recent successes and LCDM lack off is pretty telling, wouldn't you agree? I find all those studies extremely telling and I defy agree with Pavel Kroupa statements. Has anyone specific objections on any of those arguments? They directly concern the survival of the concordance cosmology model of which GR is the theoretical pilar, its very foundation. can we agree on the idea that a falsification of GR is a falsification of the SMoC and vice versa somehow?

 

[Sagittarius A-Star]

and I have no idea how you think that the speed of light should be measured, and why you think the standards that you are using to measure the speeds themselves are "constant with time".
...
For instance, you might be using the old meter prototype standard, or some variant therof that has been idealized.

The length of the old meter prototype $L_1$ is a multiple (factor := $k$) of the Bohr radius $a_0$:
$L_1 = k * a_0 = k * \frac {\hbar}{m_ec\alpha}$.

Source:
https://en.wikipedia.org/wiki/Bohr_radius

=> $\ \ \ c = k * \frac {\hbar}{m_e L_1 \alpha}$

So, if $L_1$ is assumed to be "constant with time", then you will measure a constant $c$.

 

[PeterDonis]

if the SEP holds true, neighbouring objects should not exert any gravitational influence on the studied galaxies and thus should not have any dynamical effects on them. But the contrary is indeed claimed to be observed.

Yes, I know that's what the paper claims, but, as I said, that claim is based on quite a bit of model-dependent theoretical interpretation. It's not the same as if we could run a controlled experiment in which we could directly test the SEP.

that’s exactly what MOND has been doing recently

Yes, but the whole question is still open; there is not general agreement among cosmologists about how these issues (and many others) are going to be resolved. If you are asking if it is possible that "we have to modify GR" will be part of the final resolution, yes, of course that is possible. But "modification" is not the same as "falsification". See below.

can we agree on the idea that a falsification of GR is a falsification of the SMoC

No. The implication doesn't work in either direction.

First, as far as GR is concerned, "falsification" is not the right term, any more than it was for Newtonian physics when GR was discovered. Newtonian physics still works just fine for many applications. So will GR even if we someday find that we have to modify it in order to account for certain observations in cosmology. (These remarks would apply to the OP of the thread as well.)

That said, we could end up finding issues with GR in areas that do not have any implications for the SMoC--for example, issues to do with black holes. We could also end up finding issues with the SMoC that require it to be modified in ways that don't involve GR. ("Modify our assumptions for the equation of state and density of dark matter" is one obvious way that could happen.)

 

[PeterDonis]

This review about MOND consecutive recent successes and LCDM lack off is pretty telling, wouldn't you agree?

If you only read what the prosecution says, of course you're going to think the case is open and shut.

As I said in my previous post, this whole area is still open. I would be careful putting too much credence in what partisans on any side say, particularly in informal contexts like blog posts.

 

[vanhees71]

I think the usual strategy to test GR is to use some model or fit functions for measured quantities with some parameters to fit. The fit functions are chosen such that GR predicts certain values for parameters included in these fit functions and measure with which significance/confidence the values of these constants agree with or deviate from GR. An example is to use pulsar-timing measurements and post-Newtonian approximations with the various coefficients of the corresponding expansion, which have well-determined values within GR. These high-precision observations (via radio astronomy) so far haven't shown any deviations.

I think the case is not so clear concerning the question, whether there's dark matter or deviations from GR (alternative models like MOND) to explain the findings from cosmology (CMBR fluctuations, structure formation) and astronomy (velocity curves for stars in galaxies, dynamics "galaxy collisions").

For me among the most convincing observations in favor of GR and against alternative theories of gravity is the finding of galaxies which seem to have only a small amount of dark matter, i.e., where the velocity curves of stars follow the mass distribution of the visible matter, which thus makes up the galaxy with no or only a negligible amount of dark-matter content. If MOND were right, there should be the same deviations of the velocity curves from the predictions of GR (or Newton for that matter) as in the galaxies, where the "Standard Model" assumes the presence of dark matter.

 

[Olorin]

Yes, I know that's what the paper claims, but, as I said, that claim is based on quite a bit of model-dependent theoretical interpretation. It's not the same as if we could run a controlled experiment in which we could directly test the SEP.

Of course, every test of every theory is based on model-dependent interpretation: You build a theory and compute a numerical prediction that is derived from your model and test it against the experimentally obtained data. That’s just how science works. In this case, you take MOND equation for rotation curves and you apply it to galaxies. Furthermore your model predicts that in some cases the neighboring galaxies should be able to change the acceleration scale under which the MOND regime manifests itself, the so called External Field Effect (EFE). You can compute it very precisely for each galaxy. Now the idea is, where the EFE is computed to be negligible, the rotation curves are readily fitted by the MOND equation and where EFE is computed to be significant, the fit is not that good without taking it into account, and when you take into account, the fit is spot on. Hence, this demonstrates that there is a correlation between the neighbooring astronomical objects and the stellar dynamics within the studied galaxies with a 5 sigma confidence interval. This correlation between neighboring object and stellar dynamics should not exist if the Strong Equivalence Principle holds true, hence the conclusion that this experimentally evidenced correlation falsifies it and puts General Relativity on the edge of falsification as it is the only metric theory that needs the SEP to be valid.

 

[Olorin]

First, as far as GR is concerned, "falsification" is not the right term, any more than it was for Newtonian physics when GR was discovered. Newtonian physics still works just fine for many applications. So will GR even if we someday find that we have to modify it in order to account for certain observations in cosmology. (These remarks would apply to the OP of the thread as well.)

I disagree with your take on falsification. You are being confused between the ontological validity of a theory, that is if in its essence it makes an acceptable and valid explanation of natural phenomena, and its effectiveness at describing the results of experimental observations and hence its utility for technology. Ontology and effectiveness are 2 very different aspects of our theories. Take Newtonian gravity: we know and Newton knew that this approach was ontolligically wrong, he knew that an instataneous action at a distance as a propriety of a force law was wrong, but it didn’t prey on the effectiveness of the model and we still use it to compute orbits etc...same can be said for Newtonian mechanics, now superseeded by quantum mechanics etc...So in this sense Newtonian mechanics is wrong, it doesn’t makes a valid and acceptable explanation of natural phenomena in the general and ontological sense, but it is nevertheless still effective for modeling a large array of physical situations.

In the same fashion, GR can one day be falsified, that is proven wrong in the ontological sense, that is if we demonstrate that its very essence as a metric theory of gravity doesn’t make sense anymore as a correct and valid explanation of the nature of gravity. One of the straightforward ways to do it is to test for the validity of its postulates that is its foundational pillars. The 3 equivalence principles are those pillars: the weak, the strong and the Einstein. For example and to add to the initial topic, there are 3 experiments being set up at CERN ( Aegis, Alpha g and g bar) that will test for the weak and indirectly the Einstein equivalence principles validity for antimatter, by testing anti hydrogen atoms interaction with the Earth gravitational field. If for exemple the outcome of these experiments is anti hydrogen falls up with an acceleration of -g then that’s a 200% violation of both the weak and the Einstein equivalence principles and GR is falsified in its essence because it wouldn’t make any sense to describe gravitational phenomena as resulting from the space time curvature of the Earth anymore as different test objects would now behave differently in the presence of one specific curvature. This still wouldn’t remove the effectiveness of GR formalism to agree with some observables but it will render its overall physical framework nonsensical. So this was my take on ontoligical validity, falsification and effectiveness of a scientific theory. Hope things are clearer for all.

 

[Olorin]

If you only read what the prosecution says, of course you're going to think the case is open and shut.

As I said in my previous post, this whole area is still open. I would be careful putting too much credence in what partisans on any side say, particularly in informal contexts like blog posts.

The evidence pieces have been mounting in an unprecedented way during recent times in favor of the prosecution though. I‘m not saying that the case is open and shut as I think it never really is in science (regarding the discovery of processes at least because for the ingredients is a bit different imo). But I'll be all ears about everything the defense has to say in opposition to each and every prosecution piece of evidence that is presented in the blog article ( maybe not here as it might be off topic).
As for the partisanship, I think it can’t be avoided in science as there will always be better roads at solving persistent problems than others. In other words, all the paths are not equivalent as a physical problem doesn’t realistically have multiple final solutions. Thus commitment to what seems the most plausible solution or enlightening paths will always be preferable.
As for the blog form, this is not your usual amateur blog, it is Pavel Kroupa’s blog, one of the leading astronomers and cosmologist of our time and listening to his insights in a different communicating format is always relevant and very instructive, especially since it is so well sourced and documented with all the references to peer to peer reviewed articles.

 

[Olorin]

I think the usual strategy to test GR is to use some model or fit functions for measured quantities with some parameters to fit. The fit functions are chosen such that GR predicts certain values for parameters included in these fit functions and measure with which significance/confidence the values of these constants agree with or deviate from GR. An example is to use pulsar-timing measurements and post-Newtonian approximations with the various coefficients of the corresponding expansion, which have well-determined values within GR. These high-precision observations (via radio astronomy) so far haven't shown any deviations.

Well that’s exactly what has been done in the SEP test studies I quoted but using the prediction of the MOND conjecture and in this case it has shown correlations between the dynamics of studied galaxies and neighboring objects just as predicted by the conjecture and excluded by GR. It is a perfectly well designed test and SEP/GR didn’t pass this one. In other words, if SEP/GR were correct, it shouldn’t be possible to correlate 2 independent observables ( rotation curves and neighboring galaxies physical properties) under an a priori theoretical framework with a 5 sigma confidence interval on the correlations.

I think the case is not so clear concerning the question, whether there's dark matter or deviations from GR (alternative models like MOND) to explain the findings from cosmology (CMBR fluctuations, structure formation) and astronomy (velocity curves for stars in galaxies, dynamics "galaxy collisions").

The mounting pieces of evidence are trending towards the MOND conjecture and a profound overall crisis in current cosmological paradigm that will most likely need a great amount of revision ( see Pavel Kroupa blog post « the crisis in cosmology is now catastrophic » linked above). Still unclear what will bring and push the consensus in that direction. My bet is on gravitational tests of antimatter at CERN and an ontological falsification of GR ( as explained in a post above).

For me among the most convincing observations in favor of GR and against alternative theories of gravity is the finding of galaxies which seem to have only a small amount of dark matter, i.e., where the velocity curves of stars follow the mass distribution of the visible matter, which thus makes up the galaxy with no or only a negligible amount of dark-matter content. If MOND were right, there should be the same deviations of the velocity curves from the predictions of GR (or Newton for that matter) as in the galaxies, where the "Standard Model" assumes the presence of dark matter.

Well this is defy not convincing at all since the authors of the studies ( van dokkum and al. If my memory serves ) have forgotten about the External Field Effect that exist in MOND ( see: http://astroweb.case.edu/ssm/papers/MNRASv480p473.pdf ). It is now clear that they got their conclusions wrong regarding MOND not only for ignoring the external field effect but also for a lot of other reasons.

 

[PeterDonis]

I disagree with your take on falsification.

My objection was to the word "falsification". If you're willing to say that Newtonian mechanics has been "falsified" because GR says gravity is not a force any more, for example, even though people all over the world use Newtonian mechanics every day to make predictions that are accurate enough for their purposes, then that's your preference for choice of words; it doesn't mean we disagree about what Newtonian physics can and can't do in terms of predictions.

A drawback of using the word "falsification" this way is that, by this criterion, every theory we have is potentially false and we can never say a theory is "true" because we could discover something in the future that makes us have to modify or replace it. But again, that's a matter of choice of words; it doesn't change what any given theory is capable of in terms of predictions. And predictions are what we actually use theories for.

The evidence pieces have been mounting in an unprecedented way during recent times in favor of the prosecution though.

Sure, if you only read what the prosecution says.

As for the blog form, this is not your usual amateur blog, it is Pavel Kroupa’s blog

That doesn't make it a valid reference for PF discussion. Physicists can get away with saying things in informal contexts--not just blogs but even published books, if they aren't textbooks (check out all the PF threads we've had triggered by some overstatement in one of Brian Greene's books)--that they would never get away with in a textbook or peer-reviewed paper. That's why we don't treat physicists' blogs and other informal sources as valid references, no matter who the author is or how knowledgeable they are about the field in question.

What you personally want to put credence in is, of course, up to you. I'm just explaining why you shouldn't expect others here to agree with you if all you have is an informal source, especially if it's only presenting one side of the argument.

 

[Olorin]

A drawback of using the word "falsification" this way is that, by this criterion, every theory we have is potentially false and we can never say a theory is "true" because we could discover something in the future that makes us have to modify or replace it. But again, that's a matter of choice of words; it doesn't change what any given theory is capable of in terms of predictions.

That’s the exact definition of «falsification» you can find in Popper’s books and whom coined the very term we are discussing here.

And predictions are what we actually use theories for.

Science is not only about making predictions and being usable, it is all about finding and understanding the inner workings of nature and of the physical, material world. A lot of wrong theories in the ontological sense can be used to make valid predictions. It doesn’t make them valid or true at explaining our universe, which is the very essence of fundamental science.

That doesn't make it a valid reference for PF discussion. Physicists can get away with saying things in informal contexts--not just blogs but even published books, if they aren't textbooks (check out all the PF threads we've had triggered by some overstatement in one of Brian Greene's books)--that they would never get away with in a textbook or peer-reviewed paper. That's why we don't treat physicists' blogs and other informal sources as valid references, no matter who the author is or how knowledgeable they are about the field in question.

Except Kroupa’s blog post is a formal-informal context as each of its arguments is sourced with links to several peer-reviewed article. It is not like if he was making up the arguments out of thin air, he is literally quoting peer reviewed scientific literature. So I guess you could make an exception here due to very well documented and sourced content of Pavel Kroupa’s blog as it is easier for everyone to go and look for themselves instead of me having to copy paste each article link with each corresponding argument.

What you personally want to put credence in is, of course, up to you. I'm just explaining why you shouldn't expect others here to agree with you if all you have is an informal source, especially if it's only presenting one side of the argument.

Again, Kroupa’s blog is not your usual informal source of information due to that many sourced peer reviewed articles linked to his insights. As for the other side of the argument, I haven’t heard any convincing rebuttal as of yet. So still waiting to learn about serious counter claims. But this is proby not the place to discuss those.

 

[PeterDonis]

Science is not only about making predictions and being usable, it is all about finding and understanding the inner workings of nature and of the physical, material world.

Not all scientists agree with this. Particularly those who work on quantum mechanics.

I guess you could make an exception here

Nope. If he provides links to peer-reviewed papers, great--then give references to one of those papers to back up a particular claim, not what he says about it in the blog.

it is easier for everyone to go and look for themselves

No, it's easier for everyone to go to the particular place in the particular peer-reviewed paper that is relevant, as pointed to by you (or whoever is making the claim). If you're not willing to do that work to back up a claim that you are making, why should anyone else do it for you?

As for the other side of the argument, I haven’t heard any convincing rebuttal as of yet.

Where have you looked?