Today’s second most viewed article in Newsweek is an interview with Steven Weinberg about what we’ll learn at the LHC. Unfortunately it almost immediately turns into a discussion about religion and is linked to on the Newsweek site as Will Physicists Find God? The interviewer wants to know whether the Nobelist will be willing to reconsider his well-known atheism based on what is found at the LHC. Weinberg does a good job of answering these questions politely and sensibly. He gets a bit into philosophy of science, noting that the hypothesis of the existence of God is testable (in the same sense that string theory is testable), since thunderbolts coming out of the sky and striking atheists dead would give strong evidence that He (or She) exists.
Sean Carroll, quoting from a book by David Deutsch on parallel universes, attacks Weinberg as not understanding how science works in a blog posting about Science and Unobservable Things, and in a discussion with John Horgan at Bloggingheads entitled Cosmic Bull Session. He specifically is critical of a claim by Weinberg that “the important thing is to be able to make predictions”, arguing that such a statement is “going a bit too far.”
This month’s Discover magazine has a cover story on theories of what happened before the Big Bang. The article begins with St. Augustine speculating on what God was doing before the first day of creation, then moves to discuss the work of several modern “cosmology heretics”. The discussion doesn’t include the work of the Bogdanovs, but does cover three such theories, from Steinhardt and Turok, Carroll and Chen, and Barbour, ending up with a discussion of the crucial problem of testability of such theories. Steinhardt and Turok are rather concerned about this, and point to one negative prediction (shared by many cosmological models) that they can make: effects of gravitational waves will not be seen in the CMB polarization. Carroll and Barbour on the other hand don’t seem to have a problem with not being able to predict anything, with Barbour described explicitly as having “no way to test his concept of Platonia.”
For more recent research on the multiverse, see philosopher Klaas Kraay’s Theism and the Multiverse, where he argues that:
theists should maintain that the world God selects is a multiverse. In particular, I claim that this multiverse includes all and only those universes which are worth creating and sustaining. I further argue that this multiverse is the unique best of all divinely-actualizable worlds.
You said that Carroll “. . . specifically is critical of a claim by Weinberg that ‘the important thing is to be able to make predictions’, arguing that such a statement is ‘going a bit too far.'”
Not exactly, the quote says just a bit more. Deutsch is apparently critical of the following quote from Weinberg:
“The important thing is to be able to make predictions about images on the astronomers’ photographic plates, frequencies of spectral lines, and so on, and it simply doesn’t matter whether we ascribe these predictions to the physical effects of gravitational fields on the motion of planets and photons or to a curvature of space and time.”
Of course it “does matter” whether or not an eclipse is predicted by Ptolemy’s geocentric theory of the solar system — wasn’t this really the point that Deutsch and Carroll were trying to make?
Of course, you’re trying to make a broader point, but . . . . .
After quoting the longer passage, Sean later pulls the part that I quoted out of it and refers to this specific part, in the way that I indicated.
The full Weinberg quote makes perfect sense. Weinberg understands the scientific method well, and is expressing himself accurately and carefully. He’s not discussing eclipses and Ptolemaic theory, and what he writes is not “crazy”, as Sean describes it. Both Sean and Deutsch are wilfully misinterpreting Weinberg’s words and setting up a silly straw man to knock down. I get a lot of this, but it being done to Weinberg.
Peter have you seen D. Gross’s talk on status of string theory at IPMU
opening symposium. It can be found here
The video is also apparently online, though I cannot find the link.
You said ” [a]fter quoting the longer passage, Sean later pulls the part that I quoted out of it and refers to this specific part, in the way that I indicated.”
Actually, immediately following the quote from Weinberg, Sean says “[t]hat’s crazy, of course — the dynamics through which we derive those predictions matters enormously. (I suspect that Weinberg was trying to emphasize that there may be formulations of the same underlying theory that look different but are actually equivalent; then the distinction truly wouldn’t matter, but saying “the important thing is to make predictions” is going a bit too far.)”
Perhaps not quite as clear as it could be, but not really ” . . . wilfully misinterpreting Weinberg’s words and setting up a silly straw man to knock down . . ” either, is it?
As for your statement that Deutsch is willfully misinterpreting Weinberg’s words, what exactly do you base that on? I have an old copy of the book around somewhere and I guess I could pull it out, but it seems like a lot of effort for the issue involved. I suspect it was part of a much longer and detailed argument by Deutsch that the purpose of science is more than making predictions — it’s explaining how the world works. You may not agree with that emphasis, but I think Deutsch is usually careful and sincere in this regard.
Anyway, I do enjoy your blogging even if I don’t always agree — keep it up.
Thanks for the link. It looks like Gross’s talk was much the same as other ones that I’ve discussed here in recent years, urging string theorists to not give up the search for a new formulation of string theory that would have a unique vacuum and make predictions. Also, the talk was at the opening of the new IPMU in Tokyo, which I should have written about.
I still don’t get the “that’s crazy, of course” bit (and you’re right, that’s Sean, not Deutsch). There’s nothing crazy at all about what Weinberg writes, unless you misinterpret it.
I took a more careful look at the original source from Deutsch. He’s trying to use the Weinberg quote to set up a contrast between “predictive” and “explanatory” power of a theory, but both sides of this are slippery. The problems with defining a “prediction” are well known, and an “explanation” is even trickier to define. We want not just an “explanation”, but a “scientific explanation”, and to me such an explanation inherently comes with some way that you can test it by observation of the real world, i.e. it makes predictions. Sure, you can come up with constructions of a theory with predictive, but not explanatory power, and that would be an unsatisfactory theory. But there really is no such thing as a scientific theory with explanatory power, but no predictive power, and unfortunately that’s what some string theorists and multiverse proponents are trying to sell to the public these days.
The original Weinberg quote is one I’m fond of, from page 147 of his text Gravitation and Cosmology. He was making the point that we are not necessarily bound by experimental facts to Einstein’s geometric interpretation of gravity:
…Einstein and his successors have regarded the effects of a gravitational field as producing a change in the geometry of space and time. At one time it was even hoped that the rest of physics could be brought into a geometric formulation, but this hope has met with disappointment, and the geometric interpretation of the theory of gravitation has dwindled to a mere analogy, which lingers in our language in terms like “metric,” “affine connection,” and “curvature,” but is not otherwise very useful. The important thing is to be able to make predictions about images on the astronomers’ photographic plates, frequencies of spectral lines, and so on, and it simply doesn’t matter whether we ascribe these predictions to the physical effect of gravitational fields on the motion of planets and photons or to a curvature of space and time. (The reader should be warned that these views are heterodox and would meet with objections from many general relativists.)
I.e., there is more than one way to skin a cat.
actually, ptolemy’s theory was discarded not by its metaphysical implications but by its inability to accurately describe planetary orbits, in particular that of mars. it is simply a false statement regarding a crucial point in science history that the ptolemaic system was obsoleted by its dynamical content or anything the like.
actually (in one of these truly unscientific what if statements) i would argue that had the ptolemaic system described planetary and lunar orbits just a tad better (so that naked eye observations would not reveal inconsistencies), people at the time would have probably tried harder and harder to mate it to aristotelian dynamics.
but fortunately, at this juncture, observational data were good enough to guide the way. if you just recapitulate, keplers milestone achievements were based on tychos and his own data, that were orders of magnitude better than previous ones. in fact, the very new ingredient of observing planetary positions out of opposition could be viewed as a major experimental breakthrough that allowed kepler to derive the elliptical form of the orbit.
this particular episode is very general to scientific advancements i think. and i challenge you to give me one single instance of advancements in fundamental physics that were not based on solid experimental (including of course observational) data.
‘actually, ptolemy’s theory was discarded not by its metaphysical implications but by its inability to accurately describe planetary orbits, in particular that of mars. it is simply a false statement regarding a crucial point in science history that the ptolemaic system was obsoleted by its dynamical content or anything the like.’ – chris
Chris, you are completely wrong, please see the Java animation which compares the motion of Mars predicted by both theories:
‘… Mars’ retrograde motion, the way it appears to go backward in the sky. This happens because the faster Earth passes the slower outer planet (in the left part of the diagram). And on the right, we have Ptolemy’s attempt to explain retrograde motion with a fixed Earth, using epicycles. Both ideas (Copernicus’ Sun-centered system, and Ptolemy’s Earth-centered system) seem to work well. But the Sun-centered system is the true situation.’
The point is, you could endlessly add epicycles to Ptolemy’s model to ‘correct’ imperfections. Ptolemy’s model made anthropic predictions because the exact version of the infinite landscape of possible epicycles was selected specifically to match known features of the observer’s universe, including the motion of Mars.
If you then make a prediction of something new but the prediction fails, the temptation is to announce the discovery of a new epicycle to ‘correct’ the model, to claim you are doing science, not to announce that you’ve discovered you’re wrong. So every problem with the theory is hyped as being ‘scientific evidence of the need for more complexity in nature’: see e.g. The Crime of Claudius Ptolemy by Robert R. Newton (John Hopkins University Press, London, 1977). The ptolemaic system was obsoleted by its vast dynamical content of epicycles, compared to the simplicity of Newton’s three laws of motion.
You said “. . . there really is no such thing as a scientific theory with explanatory power, but no predictive power. . .”
Yes, and as anon’s post implies above, if you have a theory that seems to have predictive power it may in fact not have explanatory power and thus not be a true scientific theory. I agree that any true “scientific theory” should both make predictions and explain how the world really works.
I still have a bit of a problem with ‘. . . it simply doesn’t matter whether we ascribe these predictions to the physical effect of gravitational fields on the motion of planets and photons or to a curvature of space and time.” I think in the end it does “matter.” However, the full quote from Weinberg provided by Khris shows that the ideas were part of a broader and more complicated argument — something that’s not always conveyed in the 25 word or less quote — whether the quote is taken from Weinberg, Carroll, Deutsch or you.
Thanks again for such an interesting blog.
Going the other way, the Economist has a long piece on various projects to find scientific explanations for religion: http://www.economist.co.uk/science/displaystory.cfm?story_id=10875666
Kris and Michael,
All Weinberg is doing is making the point that what is experimentally known about the gravitational force is consistent with either a theory based on Riemannian geometry, or with one that has a massless spin-two particle to provide the gravitational force. In the latter case, geometry may be “emergent”, only an aspect of a low-energy approximation. The idea that some (unknown) more fundamental non-perturbative version of string/M-theory might have this feature is what Gross is talking about in the lecture linked above. I’m still not getting why Sean thinks this is “crazy”.
i am very sorry, but i have to accuse you of ignorance in this topic. of course retrograde motions are possible in ptolemys astronomy. and of course you could add an arbitrary number of epicycles to match the observed position of each planet but the historic fact is, that the breakthrough of copernican astronomy happened precisely with the establishment of the elliptical orbits around the true (and not the average) position of the sun. and even before that, copernicus-based planetary tables were more precise than the preceeding alphonsine tables which was percieved at their times as the major advantage.
it’s easy in retrospect to fall into the trap of believing that observational methods back then were so poor that everyone would be content with vaguely qualitative statements like the one you quoted. they were not. similarly, it is only in retrospect that the infinite addition of epicycles is a good argument. for people of the day (without a firm grasp of the concept of series limit e.g.) the complication of one additional epicycle in terms of real work was quite noticeable and models that simplified were quite welcome.
let me finally add, that perhaps the final irony in this discussion is the arguments brought against the copernican system. if one discounts all the preoccupied ‘it must not be so it can not be’ type, the one main scientific argument against copernicus was metaphysical in nature: if earth would rotate, the missing observed parallaxe on the fixed stars implied their distance to be way beyond contemporary belief about the size of the universe (mainly influenced by aristotelian metaphysics). that was the most serious argument *against* heliocentrism.
in conclusion, i think this little piece of science history highlights very nicely the forward power of painstaking observation-phenomenology-modelbuilding as compared to the manifest backward trend inherent in metaphysical speculation.
Reading your post, I have to wonder whether the question should have been ‘Will physicists search for God’.
The Discover article is actually nicely written.
As an aside, there is this annoying ‘null-physics’ ad in the middle of the page. I first noticed that while reading an article about Einstein’s family, and the ad said ‘null physics – bring the science back into physics’ or something like this. I wrote an email to the magazine, complaining the whole advertisement is an insult to every physicist reading it (no matter what it actually advertises), and got the expected reply saying we have to get the money in etc. I don’t know whether there is a causal relation, but at least the slogan of that ad has changed into something less offensive. (I am occasionally convinced advertisements are the source of all evil in the world.)
Either way, to come back to the question of making predictions vs. offering explanations. There has been a decade or so where PREDICTIONS, testability, and falsifiable were the keywords. Now I have to wonder whether at some point people will just turn around, drop these efforts and say, no, we don’t need predictions, stop whining. It’s a bit scary to think about what this would do to science, starting from theoretical physics which is especially vulnerable in this regard, but when I read your blog it doesn’t seem impossible to happen.
You said “. . . this little piece of science history highlights very nicely the forward power of painstaking observation-phenomenology-modelbuilding . . .”
It’s true that Ptolemy’s theory was discarded for its metaphysical implications and for it’s inability to accurately describe planetary orbits, particularly as new observations were made.
I certainly didn’t mean to imply that an incorrect scientific (i.e. falsifiable) theory wouldn’t inevitably be proven wrong by further observation and discovery. In fact, that’s the notion I’d hoped to reinforce by mentioning how one such theory — which held sway with quite a few people for a good period of time — ultimately neither correctly predicted natural phenomena or explained the true nature of world, and was discarded.
In any event, a scientific theory has to make testable predictions. Carroll speculated that Weinberg might have been getting at this idea: “there may be formulations of the same underlying theory that look different but are actually equivalent; then the distinction truly wouldn’t matter.” Peter filled in the blanks saying “[a]ll Weinberg is doing is making the point that what is experimentally known about the gravitational force is consistent with either a theory based on Riemannian geometry, or with one that has a massless spin-two particle to provide the gravitational force.”
It’s easy to distinguish between theories that make different predictions that can be tested. But perhaps in the “hard” cases, it’s OK to lean a little toward the view that “explanatory” power takes on a bit more importance.
I definitely find the “we don’t need predictions, stop whining” arguments being made worrisome, but don’t think this is a big problem for science in general. It is a big problem for certain subfields of theoretical physics though, where the people pushing these arguments have influence. What is happening is that these arguments are being used to justify an unwillingness to drop research programs that have shown themselves to be inherently unpredictive. This behavior stops the field from moving forward, and destroys its credibility among potential young theorists, other physicists, granting agencies, other scientists, and the general public. I think the fall-out from this is already becoming visible.
I’m sorry, Peter, but could you explain why you said this:
“we don’t need predictions, stop whining” arguments being made worrisome, but don’t think this is a big problem for science in general.
Why isn’t lack of predictability “a big problem for science”?
Or perhaps you meant this in the context of, e.g., evolutionary biology, where we can’t “predict” the next species to evolve, but we do know that “something” will?
I thought Peter meant only that the attitude wasn’t as prevelant in most areas of science but is a growing problem in subfields of theoretical physics, but maybe I missed the point.
On page 24 of his talk, Gross says: “We are not sure what the final theory is and what the rules of the game are.”
Then on page 25, he says “The Strategy
Calculate & observe.
Well, if one does not know the rules of the game, one does not know the rules for calculation, which means the calculations are meaningless. All too often physicists substitute calculations for understanding.
General relativity, and the prediction of the existence of antimatter (Einstein and Dirac, resp.) are the only two examples I know of. Both revolutionized their respective fields, and were not based on previous experimental/observational data.
Yes, Michael is right. What I meant is just that I don’t see this as having any effect on scientists outside of string theory and associated areas in cosmology. It’s not changing their views on the necessity of coming up with ideas that have predictive power if they want to do science.
GR did make distinctive predictions (precession of perihelion of Mercury, bending of light). It’s an interesting question how seriously it would have been taken if there were no such predictions. Absent such testable predictions I doubt many people would have considered it as revolutionizing the field.
The Dirac equation makes all sorts of testable predictions, antimatter was just one more.
Gross has a point that the thing to do when you don’t know how to make progress is to calculate what you can and see what it teaches you. But if as you do more and more of these calculations you just find more and more evidence that the framework you are working in can’t do what you want, at some point you should stop and calculate in some other framework. I don’t see the relevance of “observe” to string theory at the moment.
I think Sean’s point was the same one Feynman made with Mayan astronomers. They could predict eclipses, using a system of beans, without understanding the solar system. Plenty of physicists today seem to feel calculations are the only things that matter, but the irony is that Weinberg has spoken out against that positivist viewpoint. In saying it doesn’t matter whether the underlying theory of gravity involves curved space-time, I’m sure he was only referring to the current experimental evidence.
New experimental evidence exists now, a rough measurement of frame dragging by Gravity Probe B, but it isn’t clear when we will be allowed to see it. Data collection finished two and a half years ago, but it was announced yesterday the program has been extended again — at least to the end of the year, and possibly to the year 2010. I wouldn’t bet the farm that general relativity has been confirmed.
One interesting irony in all this is that Weinberg is in some sense the father of the landscape. I think it is his work
( http://prola.aps.org/pdf/RMP/v61/i1/p1_1 ) on using a landscape to solve the cosmological constant problem that provides the most compelling justification for considering the landscape.
“All Weinberg is doing is making the point that what is experimentally known about the gravitational force is consistent with either a theory based on Riemannian geometry, or with one that has a massless spin-two particle to provide the gravitational force.”
My two bits: In the context of the G&C textbook quote, I believe the point was not that there are two competing theories, one based on Riemannian geometry and one based on the massless spin two particle. Instead, there are two ways of looking at the same theory (GR), and one may compute the same answers for the same observables from either perspective. The issue is not that the experimental data is yet insufficient to distinguish between two theories, but that there is really only one theory. This situation can be contrasted with the case of say, GR vs. Brans-Dicke, where one is actually dealing with different theories that lead to different predictions.
You’re right. Weinberg’s approach to gravity is considered equivalent to GR, and has not provided any different predictions. (Though I wonder a little whether they are really 100% the same.) Some other non-geometric alternatives to GR do make different predictions for Gravity Probe B.
Pingback: Effects, causes, and science adrift « A Quantum Diaries Survivor
Peter, I absolutely agree with you. Chris’ challenge was to mention a fundamental advance not directly based on solid experimental data (like the BCS theory depended a lot on the discovery of the isotopic effect, and Hubble’s law was basically a fit to data (a fit of sorts, though…)).
As far as I know there are just those two examples. Both were experimentally confirmed about a year after being published, as you say.
There is already an assumption that getting beyond the Standard Model will give us the “final theory” – see Gross’s comment cited above.
Pingback: You had better find God before He finds you « Gravitas Free Zone Weblog
“Sean’s point was the same one Feynman made with Mayan astronomers. They could predict eclipses, using a system of beans, without understanding the solar system.
And that was the point I was trying to make referring to Ptolemy . . . . but I like this one better.
Please stick to the topic, I’m deleting various excursions into the history of physics which don’t add anything to the discussion.
There are plenty more examples of important work that was done without any experimental basis. There’s Chandrasekhar’s calculation of the maximum white dwarf mass and the subsequent predictions about compact objects. Einstein’s prediction of the relative rates of spontaneous emission, stimulated emission, and absorption was done before same could be measured.
Lots of important theoretical work on the foundations and structure of quantum mechanics was done without any experimental input. For example, there’s the Aharonov-Bohm effect and Bell’s inequalities, both of which introduced highly unexpected phenomena. In the latter case, experimental confirmation didn’t arrive until years after the prediction.
Do you think there is a sociological reason for the association of God in science with various theories like multiverse that fall into the category of “hard to falsify”? It seems to me that there is a common element of mysticism to both areas that allows for common ground to be woven. How do scientists turn the tide back to “predictability” and putting the focus on experimental evidence? Is it a lack or redefinition of what is being taught in schools about the scientific method, or is it just a periodic dark age that we just have to suffer through?
Pingback: In Search of the God Particle, une discussion avec Steven Weinberg at Whitedwarfs
The problem here is localized to certain subfields in physics, more particularly to some specific people in these subfields. There’s no dark age, the rest of physics and science in general is doing just fine, and I assume that virtually all school teachers continue to teach their students what the scientific method is, recognizing the string theory anthropic multiverse nonsense for what it is and ignoring it.
I am not quite sure what you understand by the word “testable”, but I was a little misled by it, and had to read the original to confirm that Weinberg had not lost his marbles. Weinberg does not use that word, and in fact explicitly states that asserting the existence of a deity is not a falsifiable statement, which to me means prceisely that it is /not/ experimentally testable. So “testable in the sense of string theory” means “not testable”, correct?
‘Weinberg [notes that the] hypothesis of the existence of God is testable (in the same sense that string theory is testable), since thunderbolts coming out of the sky and striking atheists dead would give strong evidence that He (or She) exists.’ – Woit.
Actually, anyone who looks at all the suffering and tragedy in the world can see that if God exists, then She/He either has better things to do than to monitor (or respond to) serious human problems, let alone to get in a huff about claptrap from atheists like Weinberg. It’s a big multiverse, so maybe God’s presence is seen elsewhere. Maybe He/She resides only in higher dimensions…
This is an important point and seems worth discussing briefly. All mathematical models have boundary conditions, and when you push the models close to or beyond those conditions, the models start to break down. A good example would be very high Reynolds numbers in the Navier-Stokes equation. NS predicts fluid dynamics very well until you get into the regime of turbulence (high Reynolds number). Then everything becomes increasingly non-linear and eventually chaotic. That doesn’t make the Navier-Stokes equation useless in dealing with fluid dyamics, but does impose limits on how much we can predict and how far into the future underconditions of extreme turbulence. In particular, when the equations become usfficiently sensitive to intial conditions, predictability vanishes.
We see the same situation with weather prediction, protein folding, and so on. Weather prediction works well locally a few days into the future. Globally and years into the future, it works less well. This doesn’t invalidate climatology or fluid dynamics as valid sciences: it does place limits on our ability to make predictions far into the future under extreme conditions, given the current state of mathematics.
The hope is that eventually some better mathematical methods will be developed for dealing with highly-nonlinear regimes of dynamical systems, viz., turbulent fluid flow, protein folding, and so on. Work continues in an effort to develop new mathematical methods to tackle these challenging problems. In the meantime, most of these disciplines, like fluid dynamics, give excellent predictions for regimes outside of extreme conditions.
I believe Dr. Woit’s point is that this is an entirely different situation from a “theory” which cannot be disproven by any possible observations, such as the idle speculations about 10^500 vacua.