You can catch up on some of the earlier Scenes by looking at the posts listed at the end of this one. Through the course of doing those posts I’ve tried hard to summarize my views on the debate about the views of Smolin and Woit – especially hard to emphasize how the central point of their debate that is worth some actual discussion actually has nothing to do string theory at all. Basically, the whole business of singling out string theory as some sort of great evil is rather silly. If the debate is about anything (and it largely isn’t) it is about the process of doing scientific research (in any field), and the structure of academic careers in general. For the former matter, Smolin and Woit seem to have become frustrated with the standard channels through which detailed scientific debates are carried out and resolved, resorting to writing popular level books that put their rather distorted views on the issues into the public domain in a manner that serves only to muddle. On the latter, there is a constant claim that string theory and its proponents are somehow brainwashing and/or frogmarching young people into working on that area to the exclusion of all else. The authors seem oblivious to some simple facts to the contrary there: (1) that you simply can’t do that to genuinely smart, creative young people; (2) that even students who have string theorists as their Ph.D or postdoc advisors often work on non-string theory research topics (3) that they’re doing an excellent job of either driving young people away from working on some of their favourite alternatives – or from pursuing theoretical physics altogether – by failing to clearly explain their merits and by using the press to help turn this into a distorted spectacle.
I’ve summarized a lot of what I think in the latter part of this post.
There are two major problems with how live debates take place in the public sphere. One is that the average person listening to the debate cannot know whether much of what Smolin and Woit claim as facts are right or wrong (or anyone on the other side of the debate, for that matter). When someone disputes a claim that Smolin makes, he either rapidly degenerates his arguments into frustratingly technical (and often missing the point) details, or lists 10 papers to read that seldom turn out to have shown what he claims they have shown, and anyway totally derails the discussion in hand. You can find several examples of this in the discussion threads of the blog posts below and especially over on some debates on Jacques Distler’s blog. With Woit, there’s hardly ever any evidence that he has command of the issues he claims to know about (and upon which he bases very strong statements such as claiming that string theory can never make any contact with nature), and when pressed for detailed arguments seems always to be suddenly busy, dismissive, or (disturbingly often) deeply offended that he was asked to give concrete arguments at all. (You can see examples of this in the comment threads of many of the posts of a similar title below, such as IV, and VI.) To those who know the field, it is clear that there’s a failure to present credible detailed arguments, which somewhat undermines their entire position, but to the general public it seems like it might be a balanced discussion between proponents of equally well established and well developed alternatives. It is frustrating, but that is the beauty of their ploy of turning this into a public “David vs Goliath” attack. The press love that sort of thing (it is one of the few ways they care to present a science story), and the representatives of the Goliath or so-called Establishment position can’t help but come off as complacent at the very least.
The other major problem is that in many cases where there is a public debate in the media, in the presence of someone who would like to put the case in favour of string theory research, the case they present is nothing like as accusatory as when they have the platform to themselves, or on blog debates. What you get from them in those public live debates has mostly been a very reasonable set of obvious statements that nobody can disagree with: There should be diversity in research, freedom to pursue alternative ideas, better support structures for young people working on harder problems that may not be part of the mainstream, and so forth. So the listener is left with the impression that if there is a big debate or controversy, it must be that the string theorists are somehow against this, which is of course ridiculous. So the string theorist present says that they agree with the sentiments expressed, and there’s nothing left to talk about. (See an example here.)
Lee Smolin is especially good at that trick, and somehow manages to present himself as remarkably reasonable on the one hand, while on the other hand people who have read the book come away very frustrated by the attacking emphasis (and sometimes plain inaccuracies) within, and a very unpleasant characterization (some would say demonisation) of the string theory community. When it suits him, Lee’s even taken great pains to even distance himself from things in his own book at times, starting with the somewhat damning title and going forward from there.
The beginnings (the first that I heard, at least) of breaking away from these frustratingly lame debates came when Lee Smolin debated Jeff Harvey (University of Chicago) on a Chicago radio show. I blogged a bit about it here and here and here. Sadly, although Jeff Harvey did a good job, given the circumstances, the debate did not really get as far with the content as would have been desirable before the show ran out of time. (We did have some excellent extended discussion on the latter two threads, including substantial contributions from Jeff Harvey and Jacques Distler, and on this thread, with substantial contributions by Mark Srednicki.)
Well, I heard a recording of an excellent live debate yesterday that I’d like to point out to you. It turns out that the format was just perfect. It is part of a series of lectures and debates hosted by the The Royal Society for the Encouragement of Arts, Manufactures & Commerce (RSA), in London. (Audio on this page.)
Two things were different here. The first is that the audience in attendance and listening “at home” is not assumed to be afraid of a little sophisticated scientific discussion, so real content can be included – not technical details, but some actual content nonetheless. The second is that the format was better: There were three panelists who each got 20 minutes to state their case, followed by some back and forth discussion between the panelists, and questions from the audience as well.
The panelists were the physicists Lee Smolin (Perimeter Institute) and Michael Duff (Imperial College, London) and the philosopher Nancy Cartwright (London School of Economics (LSE)). It was moderated by the physicist Chris Isham. The other key thing here (and it was the most promising to thing about this to me) is that Mike Duff is a plain-speaking no-nonsense [Yorkshireman] Lancastrian who simply does not take any bullshit. I had high hopes that this would not be another lame debate.
I was worried during Smolin’s presentation, though. He gave a fairly sensible summary of the usual obvious points that nobody would disagree with – diversity in research, opportunities for young people, etc. It looked as if it was going to go the usual way. Mike would just have to agree with him, and then we’d be done. But he didn’t. He started out by saying something like (I paraphrase): “the trouble with physics is that there are two Lee Smolins. The reasonable one who we heard from just now, and the one who wrote the book”. And from there it just got better, as he brought along several specific things Lee said in his book and disputed those in his own very direct style, entirely sidestepping the “truisms” offered up for debate in Lee’s opening remarks. (I should say that Lee later cleared up one or two misinterpretations of things said in the book, and not for the first time distanced himself from some other things in the book by in turn blaming the copy editor, the publicists, or saying that they were not in the UK editions (as though that matters!). There were very many more major points of substance that Mike brought up about his claims, to which Smolin presented no answer, however.)
The philosopher, Cartwright, spent twenty minutes using a lot of fancy-sounding words and phrases (like “pessimistic meta-induction”) in very long sentences to say what was really pretty simple and could have been said in five: (1) Yes, it is very difficult to make objective value judgments about theories in a time when there are no experimental checks. (2) Both sides should be careful since the search for unification of the physical laws may be a red herring.
There was a lot of good humour throughout the entire event that made it especially good to listen to. I will not say that any points were made that you can not have read on this blog in the posts I’ve done or in the (often heated) discussions that followed, but I would say that it’s a rather good one to listen to for a summary of the two opposing views, and for the fact that Mike Duff does not pull his punches, giving the listener a rare chance to hear a senior person in the field make a relatively full case in rebuttal, for example making some very good points about how science actually proceeds, as opposed to how Smolin and Woit would like it to proceed. It should be put alongside Joe Polchinski’s excellent and detailed guest post on Cosmic Variance on the same subject (to which -strangely- Lee Smolin never responded, as far as I can tell).
Overall, I’d say that debate this is worth your time to listen to, which is why I’ve brought this tired issue up all over again. Here’s the link to the RSA’s page of lectures, you can easily find the audio for the debate there. It took place on the fifth of March.
-cvj
Some Related Asymptotia Posts (not exhaustive):
- Masterclass
- More Scenes From the Storm in a Teacup, VI
- More Scenes From the Storm in a Teacup, V
- More Scenes From the Storm in a Teacup, IV
- More Scenes From the Storm in a Teacup, III
- More Scenes From the Storm in a Teacup, II
- More Scenes From the Storm in a Teacup, I
- The Anthropic Approach To String Theory
- Punch and Judy Science Coverage
Didn’t the “rediscovery” of a non-zero cosmological constant historically occur after the supernova data indicating that the universe was undergoing an unexpected acceleration in expansion? I am fairly certain that this was in 1998-1999 as it was the subject of much discussion at the Pritzker Symposium on inflationary cosmology at the University of Chicago in early ’99. Up until the most simple inflationary models had a cosmological constant = 0.
Elliot
(rediscovery in the sense that Einstien had originally proposed a non-zero CC then backed away from the claim.)
Dear Clifford,
Of course I am not claiming to have been part of the discovery that string theory has a vast number of compactifications, or ground states. Nor was I the first to think about a multiverse cosmology, Linde’s eternal inflation was prior to my thinking about this. Nor was I the first to think about natural selection as a model for how to explain how the laws of nature of our universe were chosen, the philosopher Charles Sanders Pierce suggested this in 1893, although I did not know of this when I wrote the first papers on cosmological natural selection. I do claim that I was one of the first, if not the first, to propose that there would not be a dynamical vacuum selection principle in string theory leading to a unique vacuum state and instead to take seriously that the output of the theory would be a vast landscape of theories. I believe I was the first to define the problem of how to make falsifiable predictions from the landscape of string theories, and I believe that so far I have given the only solution to it.
Of course part of the reason for my insisting on this is personal: it would be nice to get credit for the notion that the connection between string theory and experiment must go through the landscape, especially given that so many people insisted I was wrong about this then, and especially since the term I introduced for this is the one that is now in common use.
But part of my reason is for the science, it is incredibly frustrating to have thought through the possibility that the AP be combined with the string landscape in the early 90s and to have understood why it, and more generally static, random distributions, on the landscape can’t work, to explain this in a book, and then to watch very smart people waste years playing with the ideas that one has already considered and understood cannot work. For example, it was nice to see the discussions of the difficulties of minimizing potentials on the landscape due to issues of computational complexity of Denef and Douglas, but this was no news to anyone who was familiar with the literature on evolution on landscapes in theoretical biology. The problem of evolutionary biology begins with the question of how nonetheless, in spite of the computational complexity of the fitness landscape in biology, nature produces organisms which are so fit.
Dear Mark,
Even if there were a unique theory of quantum gravity one would still need experiment to show that this was relevant for nature. This is because there is an alternative proposal, which is that gravity is not quantum mechanical but that instead quantum theory is modified or itself is only a low energy limit of a very different cosmological theory, applicable only for small subsystems of the universe.
But having said this, given that in LQG there is both a rigorous existence and uniqueness theorem for the quantization of diffeomorphism invariant gauge theories (the LOST theorem) and given that there are now calculations of the graviton propagator, there is a lot more evidence than there was five years ago that LQG does provide a consistent, finite quantum theory of gravity with a good low energy limit. There is also evidence accumulating that CDT has a good low energy limit in which spacetime emerges in 3+1 dimensions. Thus, while there is more to do on both approaches, it is looking less and less likely that string theory is either the unique or the best approach to quantum gravity,
Thanks,
Lee
“The problem of evolutionary biology begins with the question of how nonetheless, in spite of the computational complexity of the fitness landscape in biology, nature produces organisms which are so fit. ”
Dear Lee, It is not clear what precisely you are referring here to. Are you saying that what nature produced is so fit to the extent that the evolution process cannot be simulated, even in principle, by a computer?
“What if, someday, someone proves a theorem that shows that string theory provides the only mathematically consistent framework for quantum gravity?”
You will have to define first what you mean by “quantum-gravity”, and certainly you would get the desire theorem if the hypothesis is: “string theory is the only TRUE consistent framework for quantum gravity”, although the whole thing would become a logic puzzle.
Nature in the other hand is a different story. I regard mathematics (and ultimately logic) as a fantastic tool to help us trying to understand the universe, and possibly the best we can get. However, mathematics itself isnt physics, and [most] mathematicians dont cranck their heads trying to come out with simple leading ideas or models to explain data and make predicitons. One of these new ideas may lead to an experiment which will show us that QM isnt the final word and will make us rethink what a quantum theory of gravity could be.
I like what mathematicians say about their theorems, they are as useful/strong as its assumptions. And these are ultimately what makes the difference.
I’ll give u a simple example. Let’s say we find Lorentz Invariance is broken, or evolution violates unitarity. Then we would have to go back and change our assumptions and start all over, falsifying ST along the way perhaps?
I like philosophy (perhaps too much) as well, and truly believe it can give us hints and lead our path towards understanding nature, but we all want to do ‘Science’ at the end of the day, dont we?
G
And what if someday somebody proves a theorem that bodies heavier than air cannot fly?
XXX wrote, “And what if someday somebody proves a theorem that bodies heavier than air cannot fly?”
No one ever claimed to have such a theorem. In 1895, Lord Kelvin declared that “heavier-than-air machines will never fly.” No proof was offered. And he was immediately challenged with the counterexample of birds. To which his only answer was that birds were different because they were alive.
Lee: is your position that case (1) on your list, “Ours is one of a vast collection of universes with random laws”, could not possibly be true? That was the conclusion I drew from what you wrote in your book, and I am left unclear by what you wrote above.
Garbage, XXX:-
Mark clarifies what he means in this comment. If you wish to contribute something sensible to the discussion, please reply to him on that thread, to avoid confusion.
(Update: I see Mark replied here too. Well, ok, carry on. But do read his comment on the other thread.)
cheers,
-cvj
Gina asks, “Are you saying that what nature produced is so fit to the extent that the evolution process cannot be simulated, even in principle, by a computer?” Certainly there are many numerical experiments in simple systems modeling natural selection from which a lot has been learned. But when we come to the real thing, with millions of different species co-evolving all at the same time, one faces a problem of enormouosly greater complexity than evolution of a single species on a fixed fitness landscape. There are claims that it may be impossible to create a static landscape to represent something like the space of all possible combinations of all possible viable biological species on a planet. There are also results like the “no free lunch theorem” which states that no single optimization procedure works better than random search on all possible landscapes, that imply that unless you know a lot already about the landscape you cannot design a good optimization procedure for it. This means that no single computer program can both anticipate all possible landscapes that might be produced by coevolution of a large number of species and simulate evolution on it efficiently.
A related issue is “exaptation”-the fact that the selective advantage confered by a trait can change unpredictably through evolution by the discovery of a new strategy or niche,never before used or occupied. There are claims in the literature that there is no fixed procedure that can anticipate such novel shifts in fitness. This implies that the contribution to fitness of a given attribute of a species can change unpredictably as a result of changes in the environment brough on by the co-evolution of many other species. This means that there is a limit to the validity of studying evolution on fixed fitness landcapes.
Imagine that you list all the traits that contribute to the fitness of bacteria. Could you then infer from these the traits that confer fitness on multicell creatures? On the fitness landscape of bacteria there is no meaning to “flying” or “having a song which communicates health and vitality”. Yet all life evolved from bacteria. The claim that you could simulate all evolution on a computer requires anticipating in advance all possible traits that contribute to fitness in all possible biospheres and knowing how these are coded into instructions to make proteins.
One source for these issues are the last two books of Stuart Kauffman.
Thanks,
Lee
IN relation to the landscape. This is not what stood out when I went to look at Lee Smolin’s reference to chapter 5 with regards to comment #148 I hope this shift is okay for posting?
Just drawing attention to the dates of publication and comparison of views. I was thinking of “Benchmarks” in terms of the progressions, that could have been marked as successes, and help one to realize that there was still a process unfolding?
I thought these two views countering one another?
As a Lay person I was thinking of the word “ad hoc” in Lee’s statement, and wonder if this is still reflected in his views of today. This was a build up and precursor to the statement about string theory in question according to Lee’s book statement??
Unravelling String Theory,by Edward Witten 29 Dec 2005
Have we moved past this today and “all” in agreement?
The Weinberg-Witten theorem is ironically evaded by the celebrated ADS/CFT correspondency, again, a theorem is as powerful as its assumptions…
In the other end, it is like saying ST predicts gravity…
I have a great deal of respect for ST, it is actually kinda fun and some of its results suggest it is probably in the right track. Now, from there to even imagine is the ONLY way there is a huge gap. It seems like an intelligent design type of thing: Let ST be…
By the way,
“… and that the interactions of these states is always consistent with a string interpretation.”
Even so, that doesnt mean at all it is the ONLY consistent theory of QG in any sense, and one more time (sorry to bother repeating myself), it is always based on the assumptions.
G
ps. I wonder actually, how is it possible to have a theorem which will *prove* from scratch that the *only* consistent world is 11 dimensional…
I wonder actually, how is it possible to have a theorem which will *prove* from scratch that the *only* consistent world is 11 dimensional…
E8 is a lesson on diversity of symmetry, and “the 11th” the movement to branes?
iMagine what goes on in this abstract thinking? Even to me it is bewildering. Yet the brane diagrams help. How did we get there? The “complexity” quantumly involved, would have to be mind boggling?
Dear Lee,
Thank you very much for your answer.
I asked “Are you saying that what nature produced is so fit to the extent that the evolution process cannot be simulated, even in principle, by a computer?â€
To which you answered:
“But when we come to the real thing, with millions of different species co-evolving all at the same time, one faces a problem of enormouosly greater complexity than evolution of a single species on a fixed fitness landscape. There are claims that it may be impossible to create a static landscape to represent something like the space of all possible combinations of all possible viable biological species on a planet. There are also results like the “no free lunch theorem†which states that no single optimization procedure works better than random search on all possible landscapes, that imply that unless you know a lot already about the landscape you cannot design a good optimization procedure for it. This means that no single computer program can both anticipate all possible landscapes that might be produced by coevolution of a large number of species and simulate evolution on it efficiently. …”
Your suggestion, Lee, that the fitness achieved in our world is not something that can be simulated (even in principle) by a computer is quite far-reaching!
(Of course, we are not talking about a simulation that will lead to precisely the same outcomes, since those depended on a lot of randomness and relied on unknown parameters. But we are talking about comparably successful processes.) This opinion resembles in some general terms Penrose’s claims about the human brain.
While being naivly somewhat appealing, as far as I know ideas about proceesses in nature which are capable to achieve optimization beyond the capacity of computers seems contrary to emerging insights about computation. It is not clear to me if by “computational complexity” you refer to the same thing as the people in computer science. (I think the paper of Douglas and Denef that you mention and regard as “no news to anyone who was familiar with the literature on evolution on landscapes in theoretical biology” is really talking about technical terms from the theory of computer science.) But if you do talk about the computer science notion of “nor feasible” , I do not see what can be a possible interpretation of your ideas on this matter. What can be a mechanism to achieve fitness that produced something computers cannot match?
Claims of similar nature and the reference to the “no free lunch” are prominent in the intelegent design literature. Of course, advocates of intellegence design may have or may quote serious scientific ideas. But in this particular case, I do not see any reasonable scientific interpretation of claims which attribute to any optimizatin process in nature, such as evolution, computational powers which are superior than the powers of computers. (I also very vaguely remember that the “no free lunch” argument is considered rather weak.)
Gina:
I don’t want to put words in his mouth, but I believe Lee was referring to the improbability of evolution approximately repeating itself, were it even to start with a same progenitor ‘bacterium’.
This is usually referred to in evolutionary biology as the problem of convergence. How likely is it that any ‘similar’ structural or functional, features produced in one evolutionary lineage, will be able to arise, completely independently, (using different DNA sequences, let alone the SAME sequences)? This obviously depends at least upon the complexity of the feature, the time available, and the properties of the environment (both physical and biological).
There is a great deal of evidence for convergence in the evolution of features and functions like flight and swimming adaptations. The case for higher levels of complexity…like the evolution of a nerve, of light sensing organelles, and of eyes is less clear cut (see my Science and Physics Today discussions of these matters: and ).
10^500 may seem big, but the combinatorics of possible DNA messages dwarfs, that magnitude into complete insignificance. Given infinite time, biological evolutionary processes MIGHT be approximately ergodic, but not likely over as little as a few hundred billion years. In this sense (rather than “in principle”), “The evolution process can not be simulated…by a computer”.
Hope this helps.
Dear Len,
I see, thanks. (But for reading your papers please give a precise reference/link.)
The issue you refer to (if I understand you correctly) seem to apply even for one or a few different species co-evolving together. (Not just for the “real thing”.) I realize that you cannot write a computer program repeating precisely or even approximately an evolutionary process that depends on unknown parameters and may have chaotic components. I thought Lee meant something stronger. (Maybe you mean something stroger as well but in this case I do not understand what it is.)
Specifically, Kauffman’s argument that Lee mentioned based on the “no free lunch theorem” is supposed to say something stronger about impossibility – in principle – to optimize in the context of evolution. Some people I talked to were very skeptical that the no free lunch theorem (whatever it is; I am only vaguely familiar with what it is, but they were,) says much about optimization in general, and were especially skeptical about any relevance to the theory of evolution.
Gina
Somehow, the 2 links I provided, were deleted. They are:
http://www.pipeline.com/~lenornst/Life.html and
http://www.pipeline.com/~lenornst/SETI.html
You note,”The issue…seem(s) to apply even for one or a few different species co-evolving together”. It applies for many different species evolving together, in the sense that the problem gets to be even more complex.
If you need infinite time to solve a problem, then in the real world the best you can get is a good approximation. And if you can’t expect even a good approximation in a few billion years, the problem is hard enough to fit under “no free lunch”; although that also is only an approximation.
Maybe this provides some clarification?
Layman need markers from which to progress.
The Landscape – For Real This Time
Clifford:
See here for link to article.
Look familiar?
How many time since then have we seen “the familiar” dressed up in what another may say of the landscape?
I am a little confused about another thing. Is this evolution stuff really related to physics and the questions regarding quantum gravity?
Gina:
I brought evolution back in to clarify(?) Lee’s position. It also has some relevance to at least Steve Weinberg’s ‘justification’ for the Anthropic Landscape, e.g., see:
http://www.pipeline.com/~lenornst/Anthropic.html
I know most have moved on.
String Theory Landscape
See here
AS a layman I needed to understand the differences of String theory Landscape and Fitness landscape? Are others having the same problem?
Fitness landscape
See here
Always still a puzzle to me.
Thanks for your patience.
Concerning the analogy of evolution and high energy physics: If Smolin and Weinberg find it useful and it led Lee to his ideas about the landscape it is hard to argue against it. (Philosophers of science make a distinction between dicovering and justifying and point out that there is no clear methodology for the “discovering” part.) But still we can be somewhat skeptical about how far this analogy goes.
Concerning the superior computational power of evoluion/emergent of physics rule, There are three levels. One extreme level is that like Penrose’s claim about the human mind also evolution and the emergence of the law of physics exhibit superior computational power. An intermidiate step would be that since we do not have understanding of the mechanisms, solving the tentative models we create is computationally unfeasible. An even more mundane statement (which seems rather obvious) is that the randomness involved and unknown “input” prevents us from giving a precise simulation. I do no know where Lee’s opinion lies on this spectrum.
And as for the “no free lunch” theorem that Lee mentioned and its relevance to optimization and to evolution, I found this paper by Olle Haggstrom.
Some one tell me that your website is very good !I do think so!