Not Even Wrong

There’s a new philosophy of science book out, Richard Dawid’s String Theory and the Scientific Method (available online here if your institution is paying Cambridge University Press appropriately or if you have a credit card). It comes with endorsements from string theorists David Gross and John Schwarz, with Schwarz writing:

Richard Dawid argues that string theory plays a novel role in the scientific process that has been neglected by philosophers of science. I believe that this book is a valuable contribution to the philosophy of science, which should interest practicing scientists as well as those who are more interested in the methodology of science.

Dawid is a particle theorist turned philosopher, and as you might guess from the endorsement, he approaches string theory from an enthusiast’s point of view. The fundamental question addressed is how one can reconcile string theory’s failure in terms of conventional methodology of science with its continuing hold on at least part of the physics community. He explains how the book came about as follows:

This book has been on my mind ever since I left physics and turned to philosophy in the year 2000. A core motivation for making that step at the time was my feeling that something philosophically interesting was going on in fundamental physics but remained largely unappreciated by the world outside the departments of theoretical physics – and underappreciated even within. Twelve years of grappling with the specification of that general idea have considerably changed my perspective on the issue but left the overall idea intact. This book is the attempt to present it in a coherent form.

Reading the book in an odd way reminded me of my recent experience reading Gordon Kane’s reissued book on supersymmetry and string theory written in 2000, especially Witten’s essentially unchanged introduction. The degree of self-confidence of string theorists at that time was much different than now: AdS/CFT was a new idea, with a solution to QCD on its way, SUSY a sure thing at the LHC if not at the Tevatron or LEP, and at least half of the new jobs in the field going to string theorists. No landscape or multiverse pseudo-science was around to sow dissension in the ranks. No failure of SUSY to show up anywhere. No discouraging numbers like those for the last two years which show, in the US at least, 9% of jobs going to string theorists, with more jobs going to lattice gauge theorists in 2011 than to string theorists. And of course, a uniformly positive press, with no naysayers like Smolin and Woit causing trouble.

In Dawid’s description, string theorists are still partying like it’s 1999:

String theory has attained a pivotal role in fundamental physics and has been treated as a well-established and authoritative theory for quite some time by the community of string theorists and by physicists in related fields. As we have described above, large parts of fundamental physics are influenced by string theoretical analysis. The string community is one of the largest communities in all of theoretical physics and for many years has produced the majority of the field’s top-cited papers. Moreover, many string theorists express a remarkably strong trust in their theory’s viability.

For the actual list of last year’s top-cited papers in HEP, see here, and “remarkably strong trust in their theory’s viability” seems to me more 2000 than 2013. He does go on to mention skeptics, but to him a majority of the field is behind string theory, with the skeptics only coming from outside particle theory:

On one side of the divide stand most of those physicists who work on string physics and in fields like inflationary cosmology or high energy particle physics model building, which are strongly influenced by string physics. That group represents a slight majority of physicists in theoretical high energy physics today. Based on an internal assessment of string theory and the history of its development, they are convinced that string theory constitutes a crucial step towards a better and more genuine understanding of the world we observe. On the other side stand many theoretical physicists of other fields, most experimental physicists and most philosophers of physics. They consider string theory a vastly overrated speculation.

Dawid’s main thesis is that string theory critics fail to recognize that a new paradigm of scientific methodology is now needed:

String theory thus should not be taken to announce an end of science but rather to represent a new phase of scientific progress. In this new phase, progress in fundamental physics is no longer carried by a sequence of limited, internally fully developed theories, but rather by the discovery of new aspects of one overall theoretical scheme whose general
characteristics identify it as a candidate for a final theory, yet whose enormous complexity bars any hope of a full understanding in the foreseeable future.

What is the reason you should accept this final theory that no one can understand? Obviously the lack of any empirical support is a problem, so Dawid turns his attention to a detailed study of the subject of “non-empirical theory assessment”: how do you assess scientific progress absent connection to experiment? This is a real and serious problem, which Dawid studies in detail, although from a point of view which just naively accepts all arguments made by string theorists. He considers three main reasons for studying a theory with no empirical support:

• The No Alternatives Argument. This is the best argument for string theory: there aren’t a lot of viable unified theories out there. Of course, the way science progresses is that there always are unsuccessful ideas with no good alternatives, until the day someone come up with a better idea.
• The Unexpected Explanatory Coherence Argument. This is the idea that if a theory holds together better after you start studying it and understand it better, that’s a good thing. Dawid repeats uncritically claims of some string theorists that this is the case for string theory. I think you could make an equally good case for string theory unification becoming a more and more dubious idea as it became better understood (see, the Landscape).
• The Meta-Inductive Argument. Here the idea is that if a theoretical research program worked before, so will a similar later one. Dawid claims that the string theory research program is just like the research program that led to the Standard Model:
  

  Given the entirely theoretical motives for its creation, the lack of satisfactory alternatives and the emergence of unexpected explanatory inter-connections, the standard model can be called a direct precursor of string theory.

  Honestly, this I just find bizarre, and have no idea what he’s talking about, with the history of the Standard Model and the history of string theory two radically different subjects.

In one crucial respect, this book is very different though than Kane’s. Kane is well aware that the idea of an inherently experimentally untestable theory is something he can’t sell to his colleagues and the public, so he devotes his book and its argument for string theory to supposed experimental tests. More savvy string theorists than Kane though are seeing the writing on the wall: no SUSY at 8 TeV means almost surely no SUSY at 13 TeV, and thus no prospects for experimental evidence for SUSY during any of our lifetimes. To prop up the string theory unification program past SUSY null results from the 13 TeV LHC in 2016 is going to require relying on Dawid’s “non-empirical theory assessment” and convincing people that string theory and the multiverse represent a new paradigm for how to pursue fundamental science. This book will be welcomed by those pursuing such a goal.

Update: For a different take on the book you can see Lubos Motl’s review (as you might expect, he’s a big fan). The case of the most prominent string theorist blogger reminds me of one of the funnier things in the Dawid book that I forgot to mention, this footnote:

It should be emphasized that physicists on both sides of the divide are aware of the slightly precarious character of the “non-physical” arguments deployed in the debate. Lee Smolin has applied the concept of groupthink to the community of string physicists (which, incidentally, seems a quite accurate representation of what many critics of string physics do think about string physicists) but is careful not to present it as a core argument. String theorists, when entering a discussion with their critics (see e.g. Polchinski in his reasoning against Smolin), try to keep the debate at an entirely physical level.

Back in the year 2000, Gordon Kane published Supersymmetry: Unveiling the Ultimate Laws of Nature, a popular book promoting supersymmetry and string theory. The thrust of the book was that there was already indirect evidence for SUSY, with confirmation by discovery of superpartners due to come soon from LEP (which was running at energies near 100 GeV/beam) and the Tevatron (where Run II at high luminosity and nearly 1 TeV/beam was to start in 2001). The LHC was also discussed, mainly as the place that would confirm and extend the LEP/Tevatron superpartner discoveries.

Thirteen years later, with no hint of SUSY showing up as promised, not only at LEP/Tevatron energies, but also at the much higher energies and luminosities of the 8 TeV LHC, Kane has a new popular book promoting supersymmetry and string theory, entitled Supersymmetry and Beyond. It includes his claim to have predicted the Higgs mass using string theory (see Matt Strassler’s take on this here, mine here). Much of the book though consists of exactly the same text as the 2000 version.

How does Kane handle the detailed failed predictions of the 2000 edition in the new 2013 version? Basically by editing them out, with no indication to the reader that this has been done.  What’s the right word to describe the result of an Orwellian exercise like this? You can make up your mind about that yourself, since I’ve gathered together here some examples of the book text, showing the edits that were done to create the new version.

Pages xvii/xviii

Supersymmetry is still an idea as this book is being written (mid-1999) in late 2012. There is considerable indirect evidence that it is a property of the laws of nature, but the confirming direct evidence is not yet in place. That is not an argument against nature being supersymmetric; rather, the accelerator collider facilitiesy that could confirm it (the LHC) are  is just beginning to cover the region where the signals could appear (Chapter 5)[about LEP and Fermilab].

Pages 2-3/3

If we understand supersymmetry and its implications correctly, direct experimental evidence for supersymmetry will be found in the next few years – possibly soon after this book is published (or, with great luck, even before).

Pages 13/16

Only now are colliders and detectors at laboratories are now achieving the energies and luminosities (amounts of data) and sensitivities needed to explicitly detect the superpartners explicitly, at least if our thinking about their properties is more or less right.

Pages 70-71/77-79

The manner in which supersymmetry explains the Higgs physics is elegant and has important consequences for how we expect to test supersymmetry experimentally. It is rather technical. A more detailed description is given in Appendix B; here I will give a short version. There are three parts…

Therefore, the supersymmetric Standard Model explanation of the Higgs mechanism would not make sense unless the some superpartner masses were not much larger than the Standard Model masses they explain. That gives us an estimate of the masses we should expect the superpartners to have as we search for them, and it tells us at what stage we should question the validity of the theory if the superpartners have not been detected. Such estimates are only approximate, but luckily the expected masses are small enough that they imply the superpartners should be detected soon.

Appendix B was deleted entirely, it contained the text (page 156)

Therefore, the superpartner masses cannot be very much larger than the Z boson mass if this whole approach is valid. This is the only place where we can use the theory to relate the unknown superpartner masses to known masses, so on the one hand, it is a major test of the correctness of the supersymmetry explanation of the Higgs physics, and on the other, it is the most significant reason whey we expect the masses of the superpartners to have values that allow them to be produced at Fermilab or even LEP. This connection also suggests that if the superpartner masses are much larger than the Z boson mass, then the apparent success of the supersymmetry theory in explaining the origin of the Higgs physics of the Standard Model could be an accident.

Pages 88-89/89

Several arguments imply that some sparticles are within the reach of Fermilab the LHC. The strongest One of the most appealing is based on the explanation supersymmetry gives for the Higgs mechanism of the Standard Model, as described in the last chapter Chapter 7. Basically the qualitative argument is that because since supersymmetry provides the Higgs mechanism that accounts for the masses of W and Z, the some sparticle masses cannot be much heavier than the W and Z masses themselves. Fermilab has already produced and detected thousands of W’s and Z’s. When this argument is framed put in a technical form, it implies that gluinos and probably charginos and neutralinos and stops should be in the Fermilab within the LHC reach. If they are not, the impressive successes of supersymmetry listed at the beginning of Chapter 4 may be meaningless coincidences. There are some arguments, both theoretical and phenomenological, suggesting that squarks and sleptons will be too massive to produce at the LHC.

Chapter 8, on SUSY implications for matter/anti-matter asymmetry, proton decay, rare decays like mu to e-gamma, and CP violation has been deleted. Appendix D, on large extra dimensions, has also been completely deleted.

Witten’s preface has been edited:

Experimental clues suggest that the energy required to produce the new particles is not much higher than that of present accelerators. If supersymmetry plays the role in physics that we suspect it does, then it is very likely to be discovered by the next generation of particle accelerators, either at Fermilab in Batavia, Illinois, or Large Hadron Collider (LHC) or its upgrades, at CERN in Geneva, Switzerland.

Lee Smolin’s new book, Time Reborn, is out today. For more about the ideas in the book, see video of a talk here, and an interview here.

While I mostly vehemently agreed with what Smolin had to say in his last book, The Trouble With Physics, I find myself equally vehemently in disagreement with this one. On some of the topics covered, I’m indifferent to his arguments mostly as a matter of taste. While my views on human society are likely similar to Smolin’s, I’ve never found the scientific insights of fundamental mathematics or physics to have anything significant to tell me about this part of life. Similarly, while I’ve spent some time studying philosophy, I’ve mostly found this of little help in gaining deeper understanding of math or physics. Others though have a very different experience than me, and I’m not about to argue against people looking for enlightenment wherever they happen to find it.

On some of the scientific issues dealt with in the book, again I’m mostly just indifferent. Smolin accurately explains how the lack of predictivity makes typical multiverse models empty, but I’m not convinced that his favored alternative (“cosmological natural selection”) does much better. While I understand well the human appeal of wondering about what came before the big bang, I’ve yet to see any specific models of this that carry enough explanatory power about anything to make them particularly attractive or interesting.

Many of the ideas Smolin is arguing for are clearly labeled as what they are: speculative challenges from a very much minority point of view to some of the received wisdom of this kind of science. Unfortunately, parts of his argument that are most problematic are ones which are in danger of becoming the new received wisdom of the subject. The refusal to admit the failure of the idea of string/M-theory unification has left many of our most prominent theorists pushing the idea that fundamental physics is based on some new and very different degrees of freedom, with dynamics that just happens to be too complicated to allow them to find vindication by seeing how the Standard Model emerges at low energies. For his own reasons, Smolin signs on to a version of this point of view, writing:

I’m inclined to believe that just about everything we now think is fundamental will also eventually be understood as approximate and emergent: gravity and the laws of Newton and Einstein that govern it, the laws of quantum mechanics, even space itself…

A large part of the elegance of general relativity and the Standard Model is explained by understanding them as effective theories. The beauty is a consequence of their being effective and approximate. Simplicity and beauty, then, are the signs not of truth, but of a well-constructed approximate model of a limited domain of phenomena.

The notion of an effective theory represents a maturing of the profession of elementary-particle theory. Our young, romantic selves dreamed we had the fundamental laws of nature in our hands. After working with the Standard Model for several decades, we are now simultaneously more confident that it’s correct within the limited domain in which it has been tested and less confident of its extendability outside that domain.

This notion that the SM is “just an effective theory”, with its fascinating and deep mathematical structures nothing but an artifact of low-energy approximation has become the reigning ideology of the last few decades. One impetus for this has been string/M-theory, with its conjectured very different physics at short distances. This has been put together with our modern understanding of renormalization, according to which non-renormalizable theories make perfect sense as effective theories. The argument is then made that this is all there is to the SM, neglecting to note that to a large degree the SM couplings are asymptotically free, meaning that (most of) the quantized geometric degrees of freedom make perfectly good sense at all energy scales.

Smolin’s view that the recent history of particle physics makes us “less confident of its [the SM’s] extendability outside that domain [where it has been tested]” is one I strongly disagree with. Despite endless “naturalness” and “fine-tuning” predictions based on the “nothing but an effective theory” argument, the SM has not only been vindicated at the LHC over a large new energy range, but the discovery of the Higgs has shown it to have just the right characteristics to make perfectly good sense up to extremely high energies, far beyond anything we can test.

I’ve been teaching a course this past year on quantum mechanics for mathematicians, emphasizing the role of Lie groups, unitary representations and symmetries in providing not only useful calculational methods, but governing the underlying structure of the theory. Smolin argues instead that, based on Leibniz’s “identity of the indiscernibles”, symmetries cannot be fundamental (although a footnote says this doesn’t apply to gauge symmetries):

Symmetries are common in all the physical theories we know. Several of the most useful tools in the physicist’s toolbox exploit the presence of symmetries. Yet if Leibniz’s principles are right, they must not be fundamental.

This applies to the very structure of quantum mechanics:

Quantum mechanics, too, is likely an approximation to a more fundamental theory.

since it is linear, and he bets thus just a linear approximation to some fundamentally non-linear theory. Again, mathematical simplicity is seen as an artifact of approximation, not indication of something fundamental.

Smolin ends with a vision that is pretty much the exact opposite of mine, one with a vastly diminished role for mathematics in understanding the nature of reality:

The most radical suggestion arising from this direction of thought is the insistence on the reality of the present moment and, beyond that, the principle that all that is real is so in the present moment. To the extent that this is a fruitful idea, physics can no longer be understood as the search for a precisely identical mathematical double of the universe. That dream must be seen now as a metaphysical fantasy that may have inspired generations of theorists but is now blocking the path to further progress. Mathematics will continue to be a handmaiden to science, but she can no longer be the Queen.

Unfortunately it seems possible that Smolin’s arguments about mathematics will resonate well with the current backlash against sophisticated mathematics that one sees at many physics departments in the wake of the failure of string theory. In a footnote he explicitly argues that the problem with string theory was too much symmetry:

Indeed, we see from the example of string theory that the more symmetry a theory has, the less its explanatory power.

I don’t understand this argument at all. The problems with string theory are something I’ve written about endlessly here, but too much symmetry is not one of these problems.

Smolin has been quite right to point out in recent years that fundamental physical theory is in a state of crisis, but I think his diagnosis in this book is the wrong one. Abandoning the search for a more powerful mathematical understanding of the world because the huge success of this in the past has made further progress more difficult is the wrong lesson to draw from recent failures (the nature of which he lucidly described in his previous book).

My own interpretation of the history of the Standard Model is that progress came not from finding more, larger symmetries, but from a deeper appreciation of the various ways in which gauge symmetry could be realized (spontaneous symmetry breaking, confinement, asymptotic freedom). The arrival of string theory pushed the study of gauge symmetry into the background, and these days one often hears arguments against its fundamental nature, such as this one from Arkani-Hamed

What’s as a misnomer called gauge symmetry, whose beauty is extolled at length in all the textbooks on the subject, is completely garbage. It’s completely content free, there’s nothing to it.

Smolin’s arguments against the fundamental nature of symmetries, even if gauge symmetry is let off the hook in a footnote, just reinforce some of the attitudes at the root of our present-day crisis. The problems that remain in fundamental theory are difficult, but denigrating now the powerful ideas that have led to success in the past won’t help find a way forward.

Update: For more about this, there’s a review in the NYRB, and a piece at edge.org (with responses).