What makes a good physics theory

[Silviu]

Hello! What makes a theoretical model stand out compared to others. I was thinking about the Higgs mechanism and the fact that there were quite a lot of other models (proposed in the 60's-70's) to explain symmetry breaking, which (I assume based on the fact that they got published) were consistent mathematically and physically. Yet the first model they chose to test at CERN was the Higgs and it was the right one. I guess it was not a complete coincidence, but at the same time no one was 100% sure it would work (otherwise they wouldn't need to test it experimentally). So how did they chose it? What made it stand out compared to the others models? And in general, what makes experimentalists pick a model to test out of the hundreds that get published? Thank you!

 

[russ_watters]

Generally, you weigh:
1. Ability to make testable predictions.
2. Accuracy of those predictions.
3. Domain of applicability.
4. Simplicity.

And it should go without saying:
5. internal consistency.

 

[Orodruin]

4. Simplicity.

And it should go without saying:
4. internal consistency.

You mean unlike your numbering?

 

[russ_watters]

You mean unlike your numbering?

lol. Fixed. I love irony so much I'd like to say I did that on purpose, but sadly no.

 

[Silviu]

Generally, you weigh:
1. Ability to make testable predictions.
2. Accuracy of those predictions.
3. Domain of applicability.
4. Simplicity.

And it should go without saying:
5. internal consistency.

Thank you for your reply. For 3. and 4. why would that be a necessity? Is there a reason why nature would choose something simple over something complex (by our definition of simple) and at the same time, how do you quantify the applicability? Isn't this usually coming long after the discovery is done?

 

[russ_watters]

Is there a reason why nature would choose something simple over something complex (by our definition of simple) and at the same time...

Not nature, scientists. You asked how scientists pick a good theory. Simpler tends to be better because more complex theories have more components in them that could be wrong. In addition, these additional components are often built on assumptions that may not even be necessary, such as the old aether theories of light. The aether was a supposed medium which facilitated the propagation of light. The problem is, current theories that include it don't provide a way to detect it (not that they aren't good enough, but rather that being unable to detect it seems to be a feature). So there's no way to tell if it is a real thing or a figment of the scientist's imaginations.

This is Occam's Razor: https://en.wikipedia.org/wiki/Occam's_razor

how do you quantify the applicability?

Some will be built-in to the theory (magnetic forces are vastly stronger than gravity, so gravity doesn't help understand how atoms are put together), but other parts have to be tested.

Isn't this usually coming long after the discovery is done?

Often, yes, when testing is required; Newton's theory of gravity was superior to previous ideas and it works very well for a wide variety of situations. But it was found to be inaccurate, for example, in stronger gravitational fields. Einstein's General Relativity has a greatly expanded domain of applicability but again, it doesn't help describe how atoms work and the theory that does, Quantum Mechanics, is built on a fundamentally different vision of how the universe works, and they seem to be incompatible with each other...but fortunately both work well in their well separated domains.
http://nautil.us/issue/29/scaling/will-quantum-mechanics-swallow-relativity

Now, you seem to be wondering more about new theories. Today, there is very little left of the behavior of the universe that hasn't been tested yet. So new theories first and foremost need to be able to correctly predict all the behavior seen, as well as existing theories. And a broader theory is thought to be better because it covers more ground.

 

[bhobba]

Feynman nailed it:


But there are a few loose ends. Some say Feynman was a falsificationist like Popper. Popper is not too bad, better than Kuhn IMHO. But unfortunately these guys don't capture the subtleties involved eg we can't isolate quarks, yet we believe in them. Exactly how do you falsify that.

Heisenberg and Dirac had a nice discussion about other subtleties - if you want to be critical of Kuhn, as I am, bring up Dirac whose views I agree with:
http://philsci-archive.pitt.edu/1614/1/Open_or_Closed-preprint.pdf

Weinberg has some interesting things to say as well:
http://www.physics.utah.edu/~detar/phys4910/readings/fundamentals/weinberg.html

Why Truth, Beauty and Simplicity - see Gell-Mann:
https://www.ted.com/talks/murray_gell_mann_on_beauty_and_truth_in_physics

It's interesting stuff, but possibly more philosophy than physics.

Thanks
Bil

 

[sophiecentaur]

Today, there is very little left of the behavior of the universe that hasn't been tested yet.

OMG! That could be a bit of an overstatement. At any minute someone could find a massive anomaly and it would be Quantum Mechanics all over again. Maybe most present theories have been tested but.

 

[russ_watters]

OMG! That could be a bit of an overstatement. At any minute someone could find a massive anomaly and it would be Quantum Mechanics all over again. Maybe most present theories have been tested but.

I don't think that's likely, and it has to do with "the relativity of wrong" we discuss a lot. There are four fundamental interactions. Is there another we don't know about? Maybe, but it isn't likely and even if there is, for us to have not noticed it, it can't have much impact on the accuracy of our current theories.

I'm using the word "impact" here to describe how wrong a current theory is and thus how big of a change a new theory can make to our understanding of a phenomena: if the error is small, the change a new theory could make is small.

Looking at a specific one, gravity, over the years (centuries), our understanding of gravity has grown in steps and leaps. The general theories of gravity (Newton's, GR) represented big leaps, with Newton's theory being a bigger leap because previous ideas were much more wrong than Newton's gravity is. And GR has the potential to be wrong by an even much smaller amount. So since GR most of the work is on special/extreme cases and their implications. These are small steps. Very little (or nothing?) is changing to the theory on a basic level. And these small steps can't add-up to a giant leap because there simply isn't room for it. The errors or unknowns in GR simply aren't big enough to allow for a new theory to make a big change in the accuracy of current predictions.

Another use of "impact" might be how directly a theory or phenomena has influence on our daily lives. An example would be looking at particles that make up matter. Molecules to atoms to electrons and neutrons, these were big discoveries with wide applicability...most of our lives are in the domain of chemistry (molecules and atoms), knowing about the nucleus but not really doing anything with it. Then, nuclear power deals with splitting the nucleus to access the protons and neutrons. But splitting neutrons and protons to get quarks, while interesting, have less "impact" because they don't exist (very often? at all?) free in nature at the energy levels commonly seen.

This use of "impact" has more of a chance to be wrong than the first because one never knows when a new technology or application will change the "impact" of a layer of theory. Currently, most of our electricity is fueled by chemistry, but if we change to nuclear or solar, it will be fueled by nuclear energy. But even then, nuclear energy can never be as "impactcful" as chemical energy, sinc chemical energy runs our biology. Perhaps an application for quarks will be discovered that will be "impactful" to our lives too, but I doubt it.

 

[haushofer]

It should contain strings. Lots of them.

 

[sophiecentaur]

I don't think that's likely,

Well - we are only scratching the surface of dark matter and dark energy. If we ever manage to sort out those two we could have to re-think a lot about the Universe.

 

[russ_watters]

Well - we are only scratching the surface of dark matter and dark energy. If we ever manage to sort out those two we could have to re-think a lot about the Universe.

Like what?

 

[sophiecentaur]

Like what?

Be fair. How would I know?
But dark energy and dark matter are not fully explained, are they? And dark matter accounts (could account) for a vast proportion of the stuff of the Universe. I'd say that represents a significant gap in our knowledge. I don't think anyone could say we have 'characterised' it.

 

[Rap]

Yes, Feinman nearly nailed it. He did not, however deal with "simplicity". If you have two competing theories, both of which give the same correct answers, which do you choose? The "simplest". I think an accurate mathematical account of the orbits of the sun and planets could be made using an earth-centric model with the planet's paths being described as circles with an enormous number of superimposed "epicycles". Anyone who asked what "caused" the epicycles could be told to "shut up and calculate". But Newton's theory of gravitation was by far simpler and more intuitive.

The problem with quantum mechanics is that clinging to classical "intuition" is clearly counterproductive. So we are down to simplicity and the need to develop a very much deeper "intuition", and the "shut up and calculate" impulse is stronger.

 

[Ian J Miller]

I would add:
(6) Gives understanding that takes you somewhere where otherwise you would not have gone.

The problem with this is we have examined an awful lot, which means that "where you would not have gone" is somewhere that is extremely difficult to go. As an example (and for illustrative purposes for this answer only) suppose I say that gravity is a caused by a low probability quantum effect. (The problem here is fixing the probability requires a constant, which makes it difficult to evaluate.) Now, the consequence is that the probability of a gravitational force between two hydrogen molecules, say, is extraordinarily low, and they only exert such force from a large assembly of them. So here we have a proposition that will change your concept of gravity, which nobody understands, it explains why gravity is so weak, it makes testable predictions but it is essentially impossible to test, at least as I understand modern experiments. So the question now is, is this a good theory or an awful one (leaving aside the lack of detail)?