Monday, September 19, 2005

Lisa's op-ed

Lisa Randall has an op-ed in the New York Times:

She explains that the communication between the scientists and the public is more difficult than it could be in the ideal world because of

  • terminological confusions
  • complexity of science
  • uncertainty of the scientists themselves

The first, terminological theme is documented by the terms "relativity", "uncertainty principle", and "theory" that are abused by moral relativists, uncertain anti-scientific postmodernists, and intelligent designers, respectively. It is possible to abuse them because the words mean something little different in science than what they mean in the everyday life.

As you know, while I agree with her viewpoint on creationism, I completely disagree with Lisa's evaluation of the climate science, especially the statement that its insights have been "underplayed" (unfortunately, I've checked that this was no typo); with her comments about the intrinsic aptitude of sexes, and their sociology - including the questions where is the source of the confusion; what are the true reasons that prevent anyone from finding the truth about these questions, and so forth. But I think that she has done a good job anyway.

Others about the op-ed

The op-ed is also discussed by Peter Woit and Sean Carroll. Sean seems to agree with Lisa - and he conjectures that the topic of the op-ed shows that Lisa must be reading Cosmic Variance; I think that Sean's conjecture is wrong.

Peter says more useful as well as more incorrect things: he explains that the title is a pun (a "dangling participle" is an example of faulty grammar). Peter also shows that he does not know himself what the word "theory" means in physics. Why do I say so? Because Peter seems to believe that the term "string theory" is a misnomer.

Does the word "theory" refer to a wild speculation, as the creationists want you to believe, or a cheap and immediately testable set of equations that must avoid all concepts from string theory, as Peter Woit wants you to believe?

"String theory" is definitely not a misnomer; more precisely, if it is a misnomer, it is because the first word (it is not just a theory of strings), not because of the second word (theory). A theory is a coherent and consistent set of ideas, concepts, and equations that have the capacity to make predictions about a larger set of observations or experiments than the set of its assumptions, axioms, and parameters. The word "theory" does not have to mean that the theory is correct or exact as the description of the real world. We say "Newton's theory" even though we know many profound reasons why this theory is not quite the right description of reality. We also use the words "Kaluza-Klein theory" and "Little Higgs theory" even though we know that at least one of them will probably be proved incorrect.

String theory is definitely a theory in the scientific sense. It is the most coherent, consistent, and rigid set of ideas, concepts, and mathematical equations we have ever seen. It has no adjustable continuous parameters whatsoever; and it is clear from what we know that it contains all the features that are required to describe all observed phenomena in the real world. It also admits many unrealistic classical solutions (or "backgrounds"); in many of them, we can calculate almost everything with no input, and the character of these calculable quantities follows the template of the previous theories (such as the S-matrix). And Peter Woit shows that his misunderstanding of the word "theory" may be compared to the ignorance of the creationists.

Models vs. theories

Also, it would be completely crazy to replace the term "string theory" by the "string model" or something like that. The word "model" has been used to represent one of hundreds or thousands of conceivable "small theories" or "systems of a few equations" of a certain kind. Even in the context of string theory, the word "model" represents one of numerous classical solutions or backgrounds whose detailed physics may be very different; in which we make a lot of choices.

Models are usually constructed according to a template rooted in a well-known framework: models in quantum field theory are constructed by choosing the gauge groups, the matter representations for the fermions (and scalar) and their couplings.

A purely stringy example is a free fermion heterotic string model in which particular choices of the GSO projections (and the corresponding allowed boundary conditions) for the fermions have been made. The word "model" belongs to the model builders whose task is to produce thousands of models; virtually all of them will be proved inconsistent with observations soon or later. Some model builders use the intuition from string theory, others don't.

Depth vs. explicitness

Another but related difference between a "model" and a "theory" is that a model should be much more specific. It does not have to be terribly deep, but it must immediately allow the calculation of certain things. The model's being too specific usually means that the model is less likely to be true; less profound; less general; and it does not deserve to be called a "theory". This is another reason why the evolution theory or string theory can't really be called "models".

String theory as we know it is definitely not "another model". It is the unique theoretical framework of its kind. Even after several decades of attempts, it is the unique known framework that reduces to quantum field theory at low energies but goes beyond at short distances (and can include quantum gravity). I am convinced that this general insight is extremely important and those who think that the word "string model" would be more appropriate for "string theory" have misunderstood what the last 20+ years in theoretical physics were all about; no doubt, this set of people includes several heroes of theoretical physics and Nobel prize winners for physics.

String theory may have started as the "dual models" in the late 1960s before the people actually understood their meaning and their mutual relationships. Today we know that the consistent "models" follow from "superstring theory" and all their Hilbert spaces and dynamical properties are connected into a single master theory called "string/M-theory".

't Hooft's definition of a theory

Also, string theory undoubtedly satisfies the expectations that Gerardus 't Hooft (as quoted in Peter Woit's article) expects from a "theory": it comes together with instructions how to deal with it to identify the things that one wishes to describe - the elementary particles - and how to define, at least in principle, the rules to calculate the properties of these particles and make predictions about them.

(These rules are not known universally; but they are known in various approximations, e.g. the perturbative expansions in the string coupling, expansions around AdS spaces, etc. A theory does not have to be known non-perturbatively from the scratch for it to be called a theory. QED was only known perturbatively, yet it was the most accurately verified theory we have ever had. A difference between QED and string theory is that QED is non-perturbatively inconsistent due to the Landau pole while string theory is, according to everything we know, non-perturbatively consistent.)

The detailed physics as predicted by the evolutionary theory or string theory is just much more complex than in some simple models - for example in the model of creation of species by God in 7 days (that allows you to calculate that it took "7 x 5,040" minutes which you may check by comparing it with hundreds of editions of the Bible) or in the model of spin networks (whose prediction that there exists no macroscopic space can also be compared with hundreds of papers about loop quantum gravity, and most of them agree).

(The paradigm that a "model" in theoretical physics does not even have to agree with the apparent existence of space around us is often called background independence; this term used to be meaningful in theoretical physics until it was twisted in this bizarre, loop-gravitational fashion.)

This complexity and generality is another reason why the word "theory" is much more appropriate for evolution and for string than the word "model". The word "theory" is closer to a "framework" and long-term research projects; the word "model" is closer to a small set of ideas that can be "completely" calculated within a very short time period.

String as chairs

Comparing string theory to a "chair without legs, seat, back, and armrest" is just silly, and the Nobel prize given to the author of this statement does not make it any more reasonable. If you allow me to underplay the importance of string theory a bit, string theory is the most perfect kind of chair produced out of platinum that we have ever had. ;-) Identifying the exact solution or a stationary point of the effective action may turn out to be difficult, much like the identification of the precise history how all the species evolved. But this fact of nature does not change the fact that the evolution and strings are "theories".

Strings and evolution: two soulmates

I am deliberately comparing evolution and strings because I find this analogy natural, and because both of them have been used in the previous articles. In both cases, the theory is the only known solution (and probably also the only possible solution) that is consistent with very basic insights about the relevant issues.

The evolution theory is the only known theory consistent with the Earth that was created 5 billion years ago, with the apparent absence of divine forces surrounding the biosphere, with the observed difference between the offspring and their parents, with the striking similarity between biochemistry of all species, and with the fact that organisms with certain "negative" features are more likely to die.

String theory is the only known theory consistent with the existence of gauge fields and chiral fermions coupled to each other at low energies in agreement with the laws of quantum field theory; with the existence of gravity that apparently follows the laws of general relativity to very good accuracy.

In both cases, we just don't have any alternatives and don't seem to have any choice. Of course that one may continue to invent ever more intelligent theories of intelligent design ;-) or ever more convoluted theories of spin networks. But at some moment, science should realize that some attempts to reconcile these basic insights into a coherent framework have probably failed. It is not easy to give precise, quantitative predictions of evolution theory or string theory for this particular Universe, but it does not diminish the fact that the logical reasoning that has led to these two theories is very robust. This robustness justifies us to use them as the intellectual framework for further reasoning.

Speed is not the truth

You know, there can be faster and slower periods in evolutionary biology or any other field. But whether or not a period in science is fast or slow can't be enough to determine the fate of a theory. A theory can only be abandoned once some demonstrable contradictions (internal or with reality) are found and another theory doing a better job is found.

This is a reason why evolution is - using Glashow's language - more or less permanently safe, despite some possible difficulties in reconstructing the history of genes (unless we are making a silly error and missing a better theory of the origin of species). And string theory is almost permanently safe and it will only be abandoned and replaced once a better choice emerges - which, I believe, will never happen.

This safety of the status of a theory or a framework as a leading explanation of certain phenomena is of course something completely different from the safety of individual careers or grants; the latter should depend on the speed of progress, of course. But I hope that the readers can distinguish that the money and the truth are not necessarily the same thing.

1 comment:

  1. Dear Lumos,

    You take issue with Peter Woit and Gerard t'Hooft on the definition of 'theory'. How can you do this, when t'Hooft writes:

    ‘It had been an audacious idea that particles as small as electrons could have spin and, indeed, quite a lot of it. … the "surface of the electron" would have to move 137 times as fast as the speed of light. Nowadays such objections are simply ignored.’ – Professor Gerard t’Hooft, ‘In Search of the Ultimate Building Blocks’, Cambridge University Press, 1997, p27.

    TIt is clear that a theory can be contradictory and survive if objections are ignored as being irrelevant. This was the point I was making in an earlier post about Special Relativity, it can be mathematically correct without being gospel truth in the way it is commonly presented:

    ‘According to the general theory of relativity space without ether is unthinkable.’ – Albert Einstein, Leyden university lecture ‘Ether and Relativity’, 1920. (A. Einstein, Sidelights on Relativity, Dover, 1952, p. 23.)

    ‘… with the new theory of electrodynamics [vacuum filled with virtual particles] we are rather forced to have an aether.’ – P.A.M. Dirac, ‘Is There an Aether?,’ Nature, v168, 1951, p906. (If you have a kid playing with magnets, how do you explain the pull and push forces felt through space? As ‘magic’?)

    Now the Special Relativity theory denies absolute motion. But you get absolute motion everytime you accelerate, so you feel a force and can tell. And you can't start moving or stop moving without acceleration, which SR can't deal with. So you need GR, in which the principle of relativity is then the equivalence principle.

    It is clear that gravity is the major problem in physics. In Dr Randall's Warped Passages, for instance, on page 6 we read: ‘A tiny magnet can lift a paper clip, even though all the mass of the Earth is pulling it in the opposite direction. Why is gravity so defenceless against the small tug of a tiny magnet? In standard three-dimensional particle physics, the weakness of gravity is a huge problem [the ‘hierarchy problem’, the differences in the strengths of the fundamental forces]. But extra dimensions might provide an answer.’

    However, Dr Randall's suggestion that a dimension is stretched out does not offer a quantitative prediction of the coupling constant for gravity, a factor of only 10^-40 or so of the electromagnetic force strength.

    If I understand the situation correctly, in 1985 Witten and others showed that a 10-D string theory with the 6 extra dimensions curled up into a Calabi-Yau manifold would model the standard model, preserving supersymmetry and yet giving rise to an observable 4-D spacetime in which there is the right amount of difference between left and right handed interactions to account for the parity-violating weak force.

    In 1995 Witten proved that 10-D strongly coupled superstring theory is equivalent to 11-D weakly coupled supergravity.

    If we accept that this is a reasonable approach to take, are we supposed to give up on quantitative predictions and mechanisms? I can't understand why string theorists try to claim that a self-consistent model is the only way forward.

    It is obvious from the sort of input you can put into string theory that you are not going to get the coupling constant for gravity, 10^-40.

    When I say ST is at a dead end I'm referring to the approach by which you hope to get something useful out of an energy balance using the existing cosmology.

    With 90% of the mass of the universe undetected if classical cosmology is right, plus the issue that distant expansion is not being slowed down as classical cosmology predicted, it looks like a dead end.My "great idea" comes from LeSage, who suggested gravity is a pushing effect in 1748 and used it to predict the nuclear atom (because the force would have to penetrate atoms to act on every particle of matter, not just on the outer surface area of a planet): George Louis LeSage, Lucrece Newtonien, Nouveaux Memoires De L’Academie Royal de Sciences et Belle Letters, 1782, pp. 404-31. It is online at

    In CERN preprint EXT-2004-007 and in two Electronics World articles I showed that Feynman's ideas on presenting general relativity as a compressing force of the spacetime fabric were equivalent to the Lorentz contraction: the spacetime fabric pressure when moving shortens objects in the direction of motion, and the contraction term in general relativity supplies the same effect for gravity.‘… the source of the gravitational field can be taken to be a perfect fluid…. A fluid is a continuum that ‘flows’... A perfect fluid is defined as one in which all antislipping forces are zero, and the only force between neighboring fluid elements is pressure.’ – Bernard Schutz, ‘General Relativity’, Cambridge University Press, 1986, pp. 89-90.

    ‘It was proposed that a mechanism of gravity should be developed to rigorously test all of the consequences of the physical fluid model for the fabric of space… The success of this model for gravity has implications for the unification of fundamental forces via quantum theory.’ – Nigel Cook, ‘Solution to a Problem with General Relativity’, CERN Document Server paper preprint EXT-2004-007.

    The paper is at and shows that for the correct mechanism of gravity due to LeSage, the critical density is exactly .5e^3 (or about 10 times) higher than the true density. Hence most of the dark matter is eliminated, enabling an energy balance to become feasible.It is interesting that your reaction is so similar to Peter Woit's and also Quantoken's, who both dismissed it as 'nonsense'.

    Best wishes,