What is the moral obligation of theorists in the scientific community?

[marcus]

This issue of falsifiability is interesting and should be discussed. But to discuss it calmly and peacefully we should separate it from the context of whether this or that scholarly paper succeeds in showing this or that theory to be falsifiable.

We should have a general discussion of what makes a theory empirical, how and why theories are tested----questions like that: you define the issues and contribute your ideas on it.

My take on it is this:

The reason science has worked so well for the past 400 years is that there can be a coherent COMMUNITY of scientists who can hold different opinions but can eventually settle their differences by putting them to the test of experiment.

Membership in any community involves MORAL OBLIGATIONS, the people in a community need to regulate behavior in such a way that the community can survive longterm and succeed in doing its thing.

One primary moral obligation that theorists have is to ONLY PROPOSE THEORIES THAT ARE already in principle and within a reasonable time-frame in practice TESTABLE. This helps unite the science community because different factions can settle their differences peacefully by empirical trial in good faith.

Except for math and tautologies, no statement in science can ever be proven CORRECT. There are no correct theories in empirical science. Because the next experiment could prove it wrong.

All we have theories which have not been proven wrong YET---and we can be thankful we have those!

If you disagree with any of this please post your alternative view.

There is also a "good faith" assumption that scientists of good faith can agree on things like what is a reasonable waiting time and when the observations have refuted a theory.
There may be stubborn ones who refuse to believe that their theory has been refuted or who interpret the data in totally eccentric ways, but hopefully that is a minority. the majority is assumed to be able to come to agreement on what is a valid test. The system is not utterly mechanical----it depends on human factors like trust, honesty, fairness, honor, respect for people with different opinions, willingness to change etc.----"good faith" factors.

====================
If theories proposed are not verifiable (and nothing is) then what is left for them to be be? Falsifiable. A theory can make a prediction that some definite thing will happen in an experiment that hasnt been done yet. If it is tried, and it doesn't, then the theory is dead.

If theorists go around proposing unfalsifiable theories, then the community can get FRAGMENTED. differences of opinion arise that cannot be resolved empirically.

things start to depend on RHETORIC, salesmanship, social pressure, brow-beating, economic pressure to conform, metaphysical prejudice, personality cults, semi-religious stuff, primate instinct. That's bad. It isn't science.

So theorists have a moral obligation. They have a duty to the community to only make theories which have a built-in risk of self-destruction.
A theory (to be good science) has to make some prediction about a future experiment which might, if it turns out wrong, cause the theory to be discarded.
If a theory can accommodate any possible outcome of a not-yet-performed experiment, every outcome not already excluded theories we already have, then it doesn't predict anything new. It's too flexible and indescriminate---no predictive value. theorists have a duty to build a certain stiffness in so that there are some observations a theory balks at, and cannot accept. Then it's falsifiable

===============
the tricky thing here is that for this to work the theory must be falsifiable directly by observation

falsification can't be CONDITIONAL ON THE TRUTH OF A PREMISE WHICH CANNOT BE VERIFIED.

This is OK:

A is an observation
F is that the theory is false
A => F
the observation proves the theory false

This, I think, is not OK:

A is an observation
P is an unverifiable premise, some other statement
F is that the theory is false
A => (P => F)
the observation does not prove the theory false.

the observation proves the statement (P => F) that if the premise, which cannot be proven true, WERE true, then the theory would be shown to be false.

==========
ultimately this is about responsibility to the community so that it stays united and healthy
so in practice we can fudge some and we can say that P is still OK if it is one of those comfortable Apple Pie Motherhood things that everybody accepts without even thinking about it.

But suppose P besides being unverifiable is even something that half the people don't think is true anyway! Suppose P is something that a good scientist can reasonably be SKEPTICAL about?

In that case it looks pretty obvious to me that falsifiability has not been shown.
===========

So anyone can put up a different scheme, and we can see how things sort out.

I think what I am making a case for here is DIRECT FALSIFIABILITY FROM OBSERVATION

not conditional on any doubtful and in principle unverifiable premise.

in other words

A => F

And I am arguing that this is an obligation of the theorist, to make theories so they can be testable this way, and that this obligation has a PRAGMATIC REASON which is the health and success of the community.

any comment?

 

[garrett]

Hmm, what about the value of new theory that makes existing theory conceptually and computationally simpler?

An example: QFT calculations invented by Schwinger were notoriously complicated, but essentially complete as a predictive model for QED. Feynman came along with his path integrals and made everything much easier. No new falsifiable predictions, but I'd call the theory very successful and valuable.

 

[josh1]

Forget falsifiability. What about the capacity for a theory to evoke feelings of corny sentimentality - what one might call it's "schmaltzifiability"? Also, why do bad things happen to good people? Perhaps a bunny with a pancake on it's head might help.

 

[marcus]

Hmm, what about the value of new theory that makes existing theory conceptually and computationally simpler?

An example: QFT calculations invented by Schwinger were notoriously complicated, but essentially complete as a predictive model for QED. Feynman came along with his path integrals and made everything much easier. No new falsifiable predictions, but I'd call the theory very successful and valuable.

falsifiability is far from the only criterion for judging theories.
it is just a bad thing not to have.

 

[garrett]

falsifiability is far from the only criterion for judging theories.
it is just a bad thing not to have.

Ahh, I thought when you said

One primary moral obligation that theorists have is to ONLY PROPOSE THEORIES THAT ARE already in principle and within a reasonable time-frame in practice TESTABLE.

you were arguing that falsifiability should be essential.

Maybe here's a way of keeping this criteria: Does a theory qualify as falsifiable if its claim is that it nicely reproduces another falsifiable theory?

 

[Chronos]

Building a house on untitled ground is too risky for my taste.

 

[marcus]

Ahh, I thought when you said
you were arguing that falsifiability should be essential.
...

[EDIT]
It finally sinks in what you were saying, Garrett
How about we say that if two theories predict the same numbers (like Schwing and Feyn) then they are just two versions of the SAME theory. One can have several formulations of the same thing----with the same predictive content.

I DO think falsifiability is essential. But there are other good things to look for too, like elegance.
But perhaps falsifiability is THE essential feature of a theory in empirical science.

It is something you can't do without---testability.

But what I meant to say, garrett, is that there is much more that theories give us besides that. I mean the creative unifications like connecting electricity to magnetism---the unification that revealed the possibility of radio waves

Theories can make us see old phenomena in new ways and open our eyes to completely new phenomena that no one had thought to look for. A theory can EXPAND OUR IMAGINATION OF NATURE.

Yes falsifiable is essential, if something isn't made to be tested then then it's trash. but once it can be tested we should still expect more than mere testability from a great theory.

you can probably think of more and better illustrations of this, so I won't elaborate

=============
you mentioned elegance (illustrated by Feynman path integral as opposed to Schwinger less illuminating version). agreed

 

[George Jones]

This is OK:

A is an observation
F is that the theory is false
A => F
the observation proves the theory false

I tend to look also at this from an equivalent, contrapositive point of view.

~F => ~A

Equivalently,

T => B

where T = ~F = theory is true. and B = ~A. If the theory being true impliles B, and ew observe the logiical negation of B, then the theory is false, as you have written,

John Baez and Martin Hardcastle had an http://groups.google.ca/group/sci.physics.research/msg/3a09db4f289e679a?hl=en&" on sci.physics.research a few years ago about the difference between universal "for all" quantifiers and existential "there exits" logical quantifiers in statements of theories.

Regards,
George

 

[selfAdjoint]

Feyerabend, who started out as a Popperian, wound up sharply criticising Popper's falsification criterion as an improper idealization. In the real world, Feyerabend emphasized, things are alwys more ambiguous, one sparrow doesn't make a Summer, and one false prediction can generally be "handled". On the other side, modern theories and their interactions with reality are so complex that it is impossible for one mind to comprehend them and derive simple boolean propositions like this.

Jaques Distler certainly comes close to mastering string theory. I was able to reduce Distler et al's claims to a simple string of logical propositions, but look what happened; side issues immediately vitiated them. Feyerabend would be chuckling.

 

[marcus]

... In the real world, Feyerabend emphasized, things are alwys more ambiguous, one sparrow doesn't make a Summer, and one false prediction can generally be "handled". ...

In this thread we don't have just one person's view of falsifiability or the criteria (essential and otherwise) for judging a scientific theory. You me and Popper may all have different notions, and anybody who isn't Popper (who passed away in 1994) is welcome to give their own interpretation of what it means and why we want theories to be falsifiable.

My personal view is that life cannot be reduced to a mechanistic problem-solving model, the progress of science is not mechanical or reducible to simple rules. So I welcome your observation that it may take several Wrongies before you can drive a stake thru the heart of a tough theory.

I also think that (as you know better than I) the ground of 20th Century physics is littered with the corpses of theories which were at one time somebody's pet brainchild and which were passionately believed in and promoted by leaders and followers alike.

My personal view is not to adhere to a systematic philosophy of anybody including Popper, but to try to be pragmatic about it and see certain kinds of ethical behavior as important for the unity and survival of the community.

there are a lot of different points to be made including about the economical use of time, and longterm credibility and availability of government funding---there is probably no one mechanism or recipe that by itself ensures scientific progress (if we could even define that exactly)

but prediction of new results (not to indiscriminately "predict" everything that happens) is so valuable you can say it is essential, and valuable if not essential for some secondary reasons as well
1. it let's the guys kill off theories so they don't waste so much time barking in wrong alleys
2. it keeps the community more cohesive instead of breaking up into schools and factions that go their own way (which they still do to some extent but ultimately there is a way to resolve differences)
3. predictiveness of theories makes them more INTERESTING
4. predictiveness makes the discoveries of science have more practical value to mankind because it let's you sometimes design a bridge or a transistor that works (but be careful how you use it)
5. predictiveness builds up prestige and credibility so that scientists can sometimes say things that politicians don't want to hear and still be listened to, like about ozone and greenhouse.

I am not saying that practical value to mankind is essential to a good theory! A lot of these things are not essential to science. they are just BYPRODUCTS of the traditional predictivness of scientific theories.

the essential one is #1 on the list----maybe we would all agree about that. It let's you kill off the wrong ones.

And the other thing to mention, just for completness, is that you never can empirically demonstrate that a theory is a correct description of nature (correct descriptions do not exist as far as we know)

empirically all you can do is prove a theory wrong. So predictiveness does not mean verifiable, it means falsifiable.

And even then there is no absolute clear meaning of the terms, that is how it always is, discussion depends on good will and tolerance, words are not perfect. So one can get caught in quibbles. But that is the general idea as I see it of why one needs to hold theorists to the standard of making theories predictive of definite results---so they can be falsified directly by observation.

 

[Mike2]

This brings up the issue of what we all accept to start with before discussion can even begin. It seems we must start with the precepts of reasoning, the general principles of logical reasoning, before we can even begin to discuss the specifics of how this is applied to physics. We accept propositional calculus, predicate logic, and the rules of mathematics. And we accept that these logical principles are applicable to the physical world. Then we search for some fundamental precepts on which to base our physics.

But what if someone comes up with a physical theory solely derived from these logical and mathematical rules of consistency. I wonder if category/topos theory might not be the underlying basis of background independent QFT, for example. Certainly in an effort to explain everything, we should expect that one day we will discover that physics can be derives solely from the first principles of logic and math.

The question is: if no error in logic or math can be discovered in such a theory should we reject it because we are not able yet to test it? Or would the sheer logical consistency of it require that we accept it?

 

[marcus]

Or would the sheer logical consistency of it require that we accept it?

here's one possible answer (BTW you raise interesting issues)

ultimately the content of a theory is in its predictions
(not general vague predictions but specific definite ones)

if two theories predict the same numbers (and would be falsified by the same set of experimental results) then they are in a certain sense the SAME theory-------like Schwing and Feyn's two versions of QED that turned out to be the same.

if a theory does not make definite predictions then it has no content.

in that case mere logical consistency is useless to us

you ask "should we accept it?" One answer would be that we never accept theories. they are not to be believed in. what you do with theories is USE their predictions, and test them, and eventually replace them.

there is no physical evidence that a single fundamental "right" theory can be formulated using human mathematics----any thinking that assumes such a thing exists is semi-religious---has made a metaphysical leap of faith somewhere along the line.

what we can see in the history of science is a series of gradually improved models---with no implication of a limit or endpoint

logical consistency should be a requirement but it's hardly a recommendation-----silly or superfluous stuff can also be consistent.

other people may think differently about your suggestion
=============

you raise an interesting question about what are the UNSPOKEN UNQUESTIONED PREMISES of some given community of people.
that one of the hardest things to determine, it is like what is inside the earth----the layers of rock and mantle and molten stuff and iron.

there is no automatic recipe or routine for science. if we could IDENTIFY AND CODIFY what are people's unquestioned premises and reduce everything to explicit principles (hah hah good luck!) then we could make a machine that would construct theories and test them. the whole process could be automated. just turn the crank

then it wouldn't need to involve intangibles like the Sense of Beauty and the Limbic System.

but so far we are all made of meat, the last time i checked. it is amazing that things have come along this well

 

[wolram]

There is no way of kitty footing around this, make a prediction, propose an
experiment to falsify it, if the establishment does not have the means to test your theory then it is the fault of the establishment, if goverments will not
fund scientific theory what is the use of scientists.

 

[Mike2]

here's one possible answer (BTW you raise interesting issues)

ultimately the content of a theory is in its predictions
(not general vague predictions but specific definite ones)

agreed. The only reason we study is to make better decisions, which can only affect the future.

there is no physical evidence that a single fundamental "right" theory can be formulated using human mathematics----any thinking that assumes such a thing exists is semi-religious---has made a metaphysical leap of faith somewhere along the line.

the evidence that we do indeed think that there is a TOE derivable from sheer logic is... we keep trying. If we get stuck on some observation, we work at it until we have an explanation based on something more fundamental. This process of trying to understand and explaining things by more fundamental principles can stop ONLY when everything is explainable from sheer logic itself. Since we will always continually try to understand everything, there must be some innate expectation in us that everything is ultimately logical. So we pursue it in the hope it will explain some reason for existence as well.

what we can see in the history of science is a series of gradually improved models---with no implication of a limit or endpoint

The end point is where we have derived a TOE in terms of sheer logic itself and know for a certainty what will happen tommorrow without the need of any observations at all.

 

[arivero]

what we can see in the history of science is a series of gradually improved models---with no implication of a limit or endpoint

Actually we do not see a series, we only see four or five points (different theories) for a given branch of science, and we haven't any proof of infinite iteration, as a series has. Take astronomy. We have celestial spheres, then epicicles, then kepler theory, then Newton theory, then poincare, then einstein. Six points do not constitute "a series".

 

[josh1]

Look, if someone says to me they've got it all figured out except they can't predict anything, then I tell them they haven't gotten it all figured out. If they say they've got a theory that addresses various issues but which can't yet generate a testable prediction - so they know they haven't got it all figured out - but know of no reasons why the theory won't eventually be predictive, then I say fine. There really isn't anything else to say. So can we please put this issue to sleep. It's enough already.

 

[LeonhardEuler]

I remember reading in Roger Penrose's book, The Road to Reality, where Penrose was trying to explain that falsifiability was not a realistic criterion. He brought up the example of a theory that predicts the existence of at least one magnetic monopole somewhere in the universe. Clearly this theory could never be falsified because you would never know for certain that you had searched the entire universe and that there was not more stuff elsewhere. From what I remember Penrose claims that this theory would be usefull in explaining why charge is quantized (or maybe it was spin), so it has scientific value.

I think he raises a valid point in that falsifiability is not the ultimate criterion for judging the scientificness of a theory. However, in the example he uses, the theory is still testable because it is proovable to the extent that most other theories are falsifiable. It is true that you can't prove the whole universe is not a dream, so that you can't really prove the theory in this example, but it is also true that you can't disprove any theory because you could never prove that the experiment wasn't a dream. But, in a practicle sense, the statement that there is a magnetic monopole in the universe is proovable if true. Penrose fails to convince me that testability is not a necessary criterion for a scientific theory, but I do think "testable" would be a better criterion than "falsifiable".

 

[marcus]

I remember reading in Roger Penrose's book, The Road to Reality, where Penrose was trying to explain that falsifiability was not a realistic criterion. He brought up the example of a theory that predicts the existence of at least one magnetic monopole somewhere in the universe...

Leonhard thanks for the comment! This is a smart comment and points to an interesting passage of Penrose's book starting on page 1020
the section 34.4

he argues skillfully that "observational falsifiability" may be too restrictive but does not say exactly how to open it up slightly so that it would be a better standard but not TOO open so as to let untestable stuff in.

actually I think his argument is wrong----but it sounds good and it is challenging and I enjoyed re-reading it at your suggestion---and he could be right.

I could be wrong but I think he confuses a STATEMENT made by a scientific theory with the THEORY ITSELF.

It is not a theory that K>0, that the largescale curvature of the universe is positive. He gives this as an example of a theory, as he is making his argument. I think it is wrong because it is not a theory that K>0. He gives many many examples just like this. He also calls it a theory to say that K = 0.

What is a theory is not that K > 0, the theory is some model of cosmology in which one of the conclusions could be K > 0. or could be K = 0. It is the whole theory which should be falsfiable (or it is not kosher).

So if you have a theory which says, among other things, something about K and maybe that statement is not by itself falsifiable, the theory should still (if it is scientific) make SOME falsifiable statement about SOMETHING.

there has to be some future experiment that could in some imaginable circumstance go against the theory.

I think he failed to show that falsifiability is too stringent a requirement.
there may be other things wrong with what he says around pages 1020-1021 and I will try to get back to this. thanks again for bringing it up!

 

[hossi]

Some of my thoughts: how high you rate falsifiability depends on what you consider to be science, or what you consider to be important. That's a sociological issue that changes with time. For some the quest is to understand the mathematical structure of the universe, for some it's figuring out how many pions a black hole emits. Unfortunately, for most governments, science is when you can make a patent out of it. The questions humans wonder about change from decade to decade, they are influenced by politics and religion, and that's not something science is detached from. And it should not be. Though nowadays most people would do so, progress is not something you define in technological terms.

The question is then where to draw the lines between engineering, physics, mathematics and philosophy. I personally tend to put a lot of philosophy into the physics, and I always found mathematicians to be the better philosophers, but that is a side issue. The point is that I find it very short sighted to cook up some static rules for what science is, and on that, build principles for what we have to do.

However, a discussion what guidelines physicist should pursue in a given situation -- and, naturally, you are talking about the present situation -- is definitely necessary. It is necessary to stay in contact with the society we live in (and from) and the life that we lead. It is necessary to readjust the field, esp. if things seem to go wrong. In this regard, a demand for falsifiable theories is certainly good.

But I should also mention that I already see the collective scream for PREDICTIONS and testable theories to go into extreme on the other side. It's just not my understanding of physics to come up with some more or less motivated model with some parameters that fit some data or not. A brilliant example for such a thing is MOND. Certainly, it's intriguing that it works, but what to we LEARN from it? Most people I have talked to say that even if it's right, then there has to be a REASON for it. Something to understand WHY it looks (as ugly) as it does. Or take the biggest part of the DSR-predictions. Where is the theory that they work with? Is is self-consistent? Does it reproduce the SM? (soccer ball?) Is it sufficient to have some more or less convincing equations that might (in some papers) shift the GZK bound towards the right direction (even though the data situation is far from clear). How come there are suddenly so many people basically jumping on these issues completely disregarding the basic problems?

This ALSO is something that shouldn't happen!



B.

PS: marcus btw your intro about moral obligations, good faith, and the like in the scientific community sound kind of familiar. do you have a reference on that?

 

[marcus]

I agree MOND is a brilliant example.
===B===
...A brilliant example for such a thing is MOND. Certainly, it's intriguing that it works, but what do we LEARN from it? Most people I have talked to say that even if it's right, then there has to be a REASON for it. Something to understand WHY...
===endquote===

I think it is an example of how science can proceed. first get an ad hoc theory like MOND that makes predictions different from GR, then test it to see if it is right, it it is right at a phenomena level then look for the deep reasons WHY.

I have never been interested in MOND per se as some kind of final theory but only as a effective theory that would force people to look for possibly beautiful reasons, something in quantum gravity perhaps since it is a theory of space time and its interaction with matter.

so I have always been interested to know if MOND checks out. if it is right there must be something fundamental hiding behind the phenomena. (just my guess)

So I think it is a good way for science to proceed. Because if they never formulated and checked MOND they would have no clue of that divergence from straight GR.

half the time the early theory is ugly----like you say---but it is OK

I hope that the LISA pathfinder is used to check MOND in the Solar System the way Bekenstein and Magueijo have proposed, or a probe like LISA does this. And I hope that either MOND is KILLED or not killed. Either way it would be very helpful information for people developing quantum gravity.

you ask. "what do we learn from it?" Obviously we learn a huge huge thing, if we can test and it passes the test. We learn that the theories of QG must be rebuilt in a fundamental way to give mond effects, it is a huge challenge.

I try not to hope either way, but I almost can say I hope that MOND is falsified, and quickly. but frankly it would be very interesting to see how the QG people would cope if it would pass a test and were not falsified.

 

[Danny]

To me MOND is a toy model not a theory.

 

[marcus]

in case anyone is interested in Solar System tests of MOND there is this paper

http://arxiv.org/abs/astro-ph/0602266
MOND habitats within the solar system
Jacob Bekenstein, Joao Magueijo

"MOdified Newtonian Dynamics (MOND) is an interesting alternative to dark matter in extragalactic systems. We here examine the possibility that mild or even strong MOND behavior may become evident well inside the solar system, in particular near saddle points of the total gravitational potential. Whereas in Newtonian theory tidal stresses are finite at saddle points, they are expected to diverge in MOND, and to remain distinctly large inside a sizeable oblate ellipsoid around the saddle point. We work out the MOND effects using the nonrelativistic limit of the TeVeS theory, both in the perturbative nearly Newtonian regime and in the deep MOND regime. While strong MOND behavior would be a spectacular 'backyard' vindication of the theory, pinpointing the MOND-bubbles in the setting of the realistic solar system may be difficult. Space missions, such as the LISA Pathfinder, equipped with sensitive accelerometers, may be able to explore the larger perturbative region."

the version of MOND here is the TeVeS theory that Bekenstein worked out. it is a version of MOND which is covariant, in that sense like General Relativity----but has the same sort of effects as earlier versions of MOND as far as rotation curves of galaxies.

For a primer on TeVes MOND see
http://arxiv.org/abs/astro-ph/0509519
A Primer to Relativistic MOND Theory
Jacob D. Bekenstein (Jerusalem), Robert H. Sanders (Groningen)
6 pages, 1 figure, to appear in proceedings of IAP05 in Paris(EDP-Sciences 2005)

"We first review the nonrelativistic lagrangian theory as a framework for the MOND equation. Obstructions to a relativistic version of it are discussed leading up to TeVeS, a relativistic tensor-vector-scalar field theory which displays both MOND and Newtonian limits. The whys for its particular structure are discussed and its achievements so far are summarized."

 

[marcus]

B you asked about sources on my notions of why science needs theories to be testable. don't have any authoritative philosophy of science essays to offer. what I wrote is my own words and has to stand or fall on its own.

I really believe that ethics and morality is rooted in the needs of a community----you need a moral code as soon as the group is larger than face-to-face where everybody knows everybody else and you learn what to expect. (this may be wrong! I have not studied ethics!)

And I think that this extends to the scientific community and its need to resolve differences so there are no permanent divisions into like religious sects. Making theories always be clearly testable by observation at least among reasonable people of good will (which is real vague but what can you do) is only ONE of several moral obligations that go with doing science.

Anyway that is my own take and I cannot quote anybody as an authority for it. I have not read Karl Popper and probably would not like him if I did read him. A lot of philosophy is tiresome.

I think the whole big discussion of falsifiability was caused by Lenny Susskind pushing the Anthropic Principle and the discussion will probably end as soon as Susskind and the like stop pushing it and the Anthropic Principle does a fadeout.

the basic thing wrong with the AP is that you can never have a result of an experiment which is incompatible with the existence of conscious life, so whatever result occurs of whatever future experiment that always was "predicted" by the AP. So the AP has no predictive value. It is mush that can accommodate anything that happens.

A real scientific theory will tell you something that cannot happen, which if it happens then that theory is wrong. My opinion, you are welcome to disagree.

The other thing I think is wrong with the AP (besides that it is not falsifiable---ie. has no predictive value) is that it HAS NOT BEEN SHOWN TO BE NECESSARY. I don't think it's been proven that anyone needs it.
Some senior scientists may think their life's work depends on it. But I don't think physics or science as a whole needs it. Too many other possible ways that fundamental constants could be determined, ways that should be looked into. (Again personal opinion, can't offer authority for this either.)

this kind of talk is largely just in the air---Zeitgeist----you no doubt have heard it or read it from other people. If you think of a good authoritative source, especially if it is online, please let me know! Also if you disagree!

I noticed on some post of yours you said "dont use the AP if you don't have to!" I agree very much and also think that you never have to, and that it would never do any good if you did. But yeah, don't use it if you don't have to (that's good enough as a first approximation)

 

[hossi]

===B===
...A brilliant example for such a thing is MOND. Certainly, it's intriguing that it works, but what do we LEARN from it? Most people I have talked to say that even if it's right, then there has to be a REASON for it. Something to understand WHY...
===endquote===

I think it is an example of how science can proceed. first get an ad hoc theory like MOND that makes predictions different from GR, then test it to see if it is right, it it is right at a phenomena level then look for the deep reasons WHY.

I am not sure I made the point I wanted to make. I am not against MOND. I personally don't like it but I agree with you that it is a way how science CAN proceed. But that is not the only way and it shouldn't. And it's not a level that we ever should be satisfied with - certainly not a level I will ever be satisfied with.

The point I wanted to make above was that one should not just cry for falsifiability, because that is not sufficient. As you say, if we have some falsifiable model that might be good, but only because it could enable us to look "for the deep reasons". Just "looking for the deep reasons" without aiming at "falsifiability" in the first place is as short sighted as demanding "falsifyiabiliy" without aiming to find "the deep reasons".



B.

 

[marcus]

our positions are quite close although different and I would be happy to leave it thus. I probably understood your point

I personally don't like it but I agree with you that it is a way how science CAN proceed. But that is not the only way and it shouldn't. And it's not a level that we ever should be satisfied with - certainly not a level I will ever be satisfied with.
.

I, too. I do not like it personally. I'm inclined to be dissatisfied with something at the level of a bizarre overcomplicated effective theory with no hint of a fundamental spacetime matter mechanism.

(It is clunky and uses a lot of unexplained machinery. It is the kind of thing not to be satisfied with.)

such things probably give Lady Nature a headache and she is disgusted to hear about them

but I am glad that you also think that science can proceed THROUGH such things----from the effective ad hoc through to the fundamental.

The point I wanted to make above was that one should not just cry for falsifiability, because that is not sufficient...,

I see the word "just". I think one should cry and cry for falsifiability all the time. But not just for that. It is certainly NOT sufficient---not the only criterion by far. but one should insist on it. Theories must always be making a reasonable attempt to achieve it and they should exhibit testability ASAP or sooner.

just because it is not sufficient is no reason not to insist on it zealously (but not just that).

And the more fundamental and beautiful a theory is THE MORE ONE MUST INSIST on it predicting new phenomena that make it falsifiable and testable.

it is what distinguishes the real gold of new science from the fairyland daydream stuff.

that is merely my opinion. You also say this:

... if we have some falsifiable model that might be good, but only because it could enable us to look "for the deep reasons". Just "looking for the deep reasons" without aiming at "falsifiability" in the first place is as short sighted as demanding "falsifyiabiliy" without aiming to find "the deep reasons".

that is right. Except that there is a division of labor. It is everybody's job (including mine) to demand falsifiability----this to uphold and defend the empirical tradition. But it is the theorist's job to aim to find the deep reasons. All I can do is wait and be ready to applaud if (s)he finds them. So without necessarily being "short-sighted" there is a legitimate division of responsibility.

BTW I LIKE that we differ slightly in a nuance or perhaps in some logical point. I like to be here where people have a lot of different views. Rather than argue any more about this one point, I will say that I totally agree with you

 

[selfAdjoint]

The point I wanted to make above was that one should not just cry for falsifiability, because that is not sufficient. As you say, if we have some falsifiable model that might be good, but only because it could enable us to look "for the deep reasons". Just "looking for the deep reasons" without aiming at "falsifiability" in the first place is as short sighted as demanding "falsifyiabiliy" without aiming to find "the deep reasons".

IMHO this goes beyond strategy to wisdom. An insight one wishes others would follow.

 

[turbo]

There is no way of kitty footing around this, make a prediction, propose an experiment to falsify it, if the establishment does not have the means to test your theory then it is the fault of the establishment, if goverments will not fund scientific theory what is the use of scientists.

If someone with a new idea (or a new way of looking at old ideas with modern data) can come up with a model that is logically consistent, and offer a half-dozen methods by which the model might be falsified, he will be immediately shouted down by the "faithful". (VOE, here, with people like Nereid demanding quantified predictions, while conveniently ignoring qualitative predictions that do not suit them) Meanwhile whacky concepts like string theory, now grown into M theory, that can make NO testable predictions are treated like celebrity guests - put out the nice hand towels and keep the shrimp coctails coming! Modern cosmology is in crisis, and the crew members keep re-arranging the deck chairs even after the iceberg has been hit.

 

[turbo]

The point I wanted to make above was that one should not just cry for falsifiability, because that is not sufficient. As you say, if we have some falsifiable model that might be good, but only because it could enable us to look "for the deep reasons". Just "looking for the deep reasons" without aiming at "falsifiability" in the first place is as short sighted as demanding "falsifyiabiliy" without aiming to find "the deep reasons".

One should search for truth and try to define the parameters by which the truth might be revealed. That done, it is incumbent on the model-builder not just to stand back and wait to see if his model holds up, but to explain himself or herself to the community at large the implications of the model, and the means by which it might be falsified or supported. Science is not like a crap-shoot, in which you can cast your lot and hope you get lucky. If you understand what you are doing, you should be able to give reasons backing up your choices, and you should be able to to set some limits at which your interpretation fails. If you cannot do this, you are likely trading in belief and speculation, and are not dealing with logic and science.

 

[hossi]

One should search for truth and try to define the parameters by which the truth might be revealed. That done, it is incumbent on the model-builder not just to stand back and wait to see if his model holds up, but to explain himself or herself to the community at large the implications of the model, and the means by which it might be falsified or supported. Science is not like a crap-shoot, in which you can cast your lot and hope you get lucky. If you understand what you are doing, you should be able to give reasons backing up your choices, and you should be able to to set some limits at which your interpretation fails. If you cannot do this, you are likely trading in belief and speculation, and are not dealing with logic and science.

Hi turbo,

I totally agree. Unfortunately it seems, today half of the science is "cast your lot and hope you get lucky" wereas the other half is "don't cast anything and hope a lot".

The only statement of yours I don't like is the first sentence. If you start talking about truth, sooner or later you will have to define what you mean by that -- do we really want to start a discussion about what truth and reality is? I suspect you mean it as a synonym for describing nature?



B.

 

[hossi]

IMHO this goes beyond strategy to wisdom. An insight one wishes others would follow.

Not sure I understand what you say I surely wish others lots of insight The 'strategy to wisdom' is exactly what needs to be readjusted. But can you do so without discussing what 'wisdom' means? And can that ever be independed of the influence of the society we life in?



B.

 

[hossi]

BTW I LIKE that we differ slightly in a nuance or perhaps in some logical point. I like to be here where people have a lot of different views. Rather than argue any more about this one point, I will say that I totally agree with you

Well, I too think that we mostly agree. However, just one remark:

And the more fundamental and beautiful a theory is THE MORE ONE MUST INSIST on it predicting new phenomena that make it falsifiable and testable.

it is what distinguishes the real gold of new science from the fairyland daydream stuff.

Yes, but it's also a question of timescale and of alternatives. Say, a theory that is fundamental and said-to-be-beautiful should at some point predict something. I think we agree on that.

If you find the approach from a fundamental level viable then you have to take into account that not all such approaches will immediately be testable, or it will take time to figure out a way to make predictions.

But how much time do you give the researchers. 10 years? 20 years? 30?? How much do you invest in their efforts? How much do you invest in alternatives? How do you distribute your support? At which point do you say it's enough?

I am not asking that because I want you to answer. These are questions that ought to be discussed - Repeatedly - in the scientific community. To make sure we do not get lost in fairyland. The answer won't always be nice, it won't always be easy, but it would help us save a lot of brain capacity.



B.

 

[turbo]

The only statement of yours I don't like is the first sentence. If you start talking about truth, sooner or later you will have to define what you mean by that -- do we really want to start a discussion about what truth and reality is? I suspect you mean it as a synonym for describing nature?

The best I can do is point to Einstein, who used thought experiments to point out truths, and who refined his ideas to demonstrate them. If you observe that A follows from B, and reliably so, you should search for a way to explain the relationship between A and B. You can posit an explanation for the relationship, but a real theorist would not only posit a relationship, but would provide meaningful criteria by which the theory could be proven or falsified. Modern cosmology has run off the track in this regard. "Darn, if I could only have 27 dimensions, all but 3 or which are hidden, I can explain the whole Universe, as long as I don't have to predict anything and don't have to subsequently explain any competing observations" - it seems like the ultimate cop-out to me, but the parade-goers admire the emporer's new clothes regardless.

 

[hossi]

The best I can do is point to Einstein, who used thought experiments to point out truths, and who refined his ideas to demonstrate them. If you observe that A follows from B, and reliably so, you should search for a way to explain the relationship between A and B. You can posit an explanation for the relationship, but a real theorist would not only posit a relationship, but would provide meaningful criteria by which the theory could be proven or falsified. Modern cosmology has run off the track in this regard. "Darn, if I could only have 27 dimensions, all but 3 or which are hidden, I can explain the whole Universe, as long as I don't have to predict anything and don't have to subsequently explain any competing observations" - it seems like the ultimate cop-out to me, but the parade-goers admire the emporer's new clothes regardless.

Turbo, I just don't get your answer. I think we are on the same track, but despite you pointing to Einstein, talking about truth is still not helpful. I.e. it is problably true that supersymmetric string theory is anomaly free only in 10 dimensions. But you are referring to truth as something observable in nature - that was what I meant with "I suspect you mean it as a synonym for describing nature". B.

 

[turbo]

At some point, we have to deal with fundamentals. We have to make distinctions between what is arguable and what is demonstrably true. If we cannot come to agreement on what is demonstably true, there will be conflict forever. If we allow the dialogue to be dominated by those who argue between 27 dimensions and 24 dimensions, we will all lose, since neither proponent can give a single testable prediction by which their model might be falsified. "String", "M" or whatever you want to call it is a loser out of the gate, unless they are willing to engage with observational astronomy, and they are not.

 

[Chronos]

One should search for truth and try to define the parameters by which the truth might be revealed. That done, it is incumbent on the model-builder not just to stand back and wait to see if his model holds up, but to explain himself or herself to the community at large the implications of the model, and the means by which it might be falsified or supported. Science is not like a crap-shoot, in which you can cast your lot and hope you get lucky. If you understand what you are doing, you should be able to give reasons backing up your choices, and you should be able to to set some limits at which your interpretation fails. If you cannot do this, you are likely trading in belief and speculation, and are not dealing with logic and science.

There is no "truth' in science, that is a philosophical concept. In science there is only the "preponderance of observational evidence". In that context, defining "the parameters by which the truth might be revealed" means all models must be rigorous, mathematically consistent, and predictive. Thus it is incumbent upon the model-builder to predict the results of crap-shoots before the dice are rolled, not interpret them ad-hoc. My objection is to the suggestion mainstream scientists tend to leap to conclusions - hoping they get lucky when the observational evidence rolls in. Albeit, I agree that is a fair characterization of most non-mainstream scientists.