I was planning on writing something about the field with one element, but Lieven Le Bruyn has done a better job of it than I would have, linking to all of the recent news on this subject I was aware of, and more.
Today’s New York Times has an article entitled Dark, Perhaps Forever, which is summarized as “Scientists are beginning to despair of explaining the universe”. It is about the recent dark energy symposium in Baltimore, and focuses on Witten’s talk, which was discussed previously here. To an account of the talk itself, it adds this quote from Witten:
As for how I feel personally, I am not sure what to say… I wasn’t terribly enthusiastic the first, or even second, time I heard the proposal of a multiverse. But none of us were consulted when the universe was created.
There’s no mention of the crucial issue that Rachel Bean implicitly confronted Witten with in a question at the end of his talk: if the landscape inherently can give no testable insight into physics, why should a scientist bother with it?
Other speakers at the symposium discussed possible future experiments to measure dark energy and their funding prospects. One worry is that such experiments may do little more than give a somewhat more accurate dark energy number, providing no further insight into the problem of its origin.
Why do you say this is a worry? Unless they find strong evidence that dark energy is inconsistent with a cosmological constant, this is a given. NASA’s upcoming joint dark energy mission (either SNAP, DESTINY, or ADEPT) is banking on finding something interesting, where interesting means the equation of state (w) can be determined to be something other than a cosmological constant (w=-1, with no change in w over time). It’s terribly boring, but there’s really no other option other than to look for this.
Either way, we still have no idea how to account for either a cosmological constant or something different from a cosmological constant.
Ethan,
I was just summarizing the remarks of Steinhardt and Krauss at the end of the article. In any case, everyone agrees that there’s not much to do except go do the experiment and see.
Wiltshire proposes an alternative to dark energy in arxiv.org/abs/0712.3984 , namely the interpretation that it is a measurment artefact due to the inhomogeneity of the universe. When integrated over 13700 million years, the inhomogeneity yields acceleration and all other modern observations. What has to be thought about this sort of work?
John
I wish more people understood math. Then they’d realize what the really cool part of this blog entry was!
Thanks for the link, Peter – I hadn’t seen it.
350 scientific instruments now on line:
http://instrumentenzaal.teylersmuseum.nl/
Following up on John’s comment, I should mention that one of the best links to follow from Lieven’s posting is this one:
http://math.ucr.edu/home/baez/week259.html
Hi Peter, this is off-topic but relevant to your campaign against hype in general, you might consider a post on it.
The word over here in Europe is that the cold fusion controversy is back – yet more headlines on the fantastic promise of cheap energy, this time backed by an experiment at Osaka University.
I have a posting on my blog, or you can red the original story at http://physicsworld.com/blog/2008/05/coldfusion_demonstration_a_suc.html
Cormac
Cormac,
On the list of things I would like to ignore, and definitely don’t want to start a discussion about here, cold fusion is right up there…
Oops, sorry! Just wondered if you guys had heard..
>Wiltshire proposes an alternative to dark energy in
>arxiv.org/abs/0712.3984 , namely the interpretation that it is a
>measurment artefact due to the inhomogeneity of the universe.
Edward “Rocky” Kolb of Fermilab made a similiar proposal in 2005.
http://arxiv.org/abs/astro-ph/0506534
Peter what do you think of Susskind’s talk at the PASCOS meeting?
There area also some other string theory related talks.
Shantanu,
I took a look at the Susskind slides. I still don’t see any predictions of anything coming out of the multiverse scenario he is promoting.
“if the landscape inherently can give no testable insight into physics, why should a scientist bother with it?”
It is not true that the landscape “inherently can give no testable insight”. cf the work of Anthony Aguirre, Matthew Kleban, and many others on the possibility of collisions of universes in the multiverse. You have had this pointed out to you *many* times. Readers can draw their own conclusions.
vicar,
I have many times pointed out that, yes, it is in principle possible to construct multiverse models in which there are observable effects on our universe, for example the kind of thing that the people you mention are trying to do. There’s a huge difference though between showing that one can construct such models, and showing that they have something to do with our universe.
The models typically being studied by the string theory landscape people don’t have observable effects due to other universes, and it was those that Rachel Bean was asking Witten about in the question I referred to.
I should also point out that the context of the quote is about the question of whether the multiverse can provide insight into why we live in one “string vacuum” rather than another, and thus explain something about particle physics. The models you refer to don’t provide any such insight or explanation.
Peter, Lets assume the very real possibility that that there are indeed many ways (which cannot be resolved at current energies) to UV complete our familiar low energy physics and gravity. Is there ANY way of doing physics in that context which will keep you happy?
The way I would do physics in that context is to construct models that contain well-known low energy physics (with extra stuff at higher energies that I might or might not add depending on my prejudices), and wait for the next round of experiments. Modulo the (very important) fact that our tools are not sharp enough to give us the full control we would like to have, this is precisely what the string phenomenologists are trying. I don’t understand what is the issue of PRINCIPLE that you seem to be arguing against.
There is a huge distinction to be made between understanding the multiverse, which presumably tells us about string theory, and understanding specific vacua, which could potentially tell us about OUR universe. Landscape statistics, measure issues etc. belong firmly in the first category. But model-building sits squarely in the second. The word anthropic makes some people turn to violence, but for me it is the same as fixing the vacuum from experimental input, which is what we have ALWAYS done in physics! There is nothing I understand in string theory which tells me that this is in principle impossible. The real problem is actually technical control, but stabilized string vacua are only a few years old, so its not hopeless yet.
Peter says:
“if the landscape inherently can give no testable insight into physics, why should a scientist bother with it?”
It is not the landscape that gives us predictivity, it is the individual vacua.
Let me emphasize this once more: if there are many UV completions consistent with low energy physics, then it would be inherently impossible to offer a single “prediction” for future experiments. The best one can do is to offer classes of predictions. You will need future (LHC?) experiments to pin them down further.
This is for example essentially what Vafa and collaborators did in the paper that you talked about couple of days ago, which you were unhappy with because it did not make “a” prediction.
How can a vacuum be a theory? It’s a solution, with some parameters. If you call it a theory, you’re admitting that there is no criterion for these to be the right parameters. All you have then is a fit. Are we seriously going to call a parameter fit a theory of nature?
If it is true that fits are physics, I propose (take a moment to pause in childlike awe of Nature’s works!) the Standard Lagrange Model of the universe. Give me any set of data points, I will use Lagrange’s method to find a smooth function fitting those points. How beautiful that everything in the world is reduced to simple algebra. And by these ingenious method we require no more parameters than numbers to fit (in this respect it is better than some approaches). Now we finally understand how everything works.
OK, forgive my sarcasm. But seriously folks, if you want a UV completion of the Standard Model without experiments, you need a good reason to chose this completion. Renormalizability tells you that any completion is equally good – and no completion is necessary! QUALIFICATION: well the Higgs needs fine tuning and most likely has a trivial S-matrix near the Planck scale, but we’ll probably never do experiments there.
>All you have then is a fit. Are we seriously going to call a parameter fit a theory
>of nature?
The difference is that for a fully stabilized vacuum to qualify as our Universe, it has to accommodate not just the finitely many experiments that have to be done in order to determine the vacuum, but also the infinitely many experiments that CAN be done afterwards. Thats how we know that we are not merely fitting curves. Incidentally, the standard model is fixed exactly in this way – except that fortunately this could be done already at currently accessible energies.
There are many crucial questions. Here is one: can we fix a vacuum in string theory by doing finitely many sub-Planckian measurements? Or do we need Planck-scale measurements? These and other questions have to be answered to make string theory useful, and string theorists are trying (or should try) to answer them. But these are serious questions and real work will be necessary to see how useful or not string theory ultimately is.
My point is that the mere existence of the landscape or multiverse or whatever you want to call it, does NOT make the theory unpredictive. That is merely a strawman argument that gets a lot of knee-jerk reactions. Almost all theories have landscapes.The weird thing is not that string theory has a landscape, but that string theorists were hoping against all reasonable hope for tens of years that there would be a unique vacuum.
>if you want a UV completion of the Standard Model without experiments,
>you need a good reason to chose this completion. Renormalizability tells
>you that any completion is equally good – and no completion is necessary!
The problem is we have something other than standard model at low energies: gravity. And string theory provides the UV completion there, but also tells us that there could be many ways to do it.
Peter O,
In a particular vacuum with all moduli stabilized , all parameters such as gauge, Yukawa couplings are fixed as these depend on the moduli. In addition, it is not possible to just construct any model you want as there are usually a host of consistency conditions that must be satisfied that strongly constrain the models. Thus, if one can find a model which describes exactly the SM in detail in its low-energy limit, then this would be a very significant discovery.
Somebody,
I’m probably not going to convince you of anything in my rebuttal below. But I hope you will feel it is worth knowing that there is a rebuttal.
1. “There are many crucial questions. Here is one: can we fix a vacuum in string theory by doing finitely many sub-Planckian measurements? Or do we need Planck-scale measurements?”
Suppose you could do this. Isn’t this just a parametrization consistent with string theory? And if so, how does it prove anything? I like my Lagrange Standard Model (above) better. The nation that controls Lagrange polynomials controls the universe!
2. “The weird thing is not that string theory has a landscape, but that string theorists were hoping against all reasonable hope for tens of years that there would be a unique vacuum.”
I’m with David Gross on this one. Abandoning this hope means that you no longer have a theory.
3. “The problem is we have something other than standard model at low energies: gravity. And string theory provides the UV completion there, but also tells us that there could be many ways to do it.”
No, it doesn’t provide THE UV completion. It provides a UV completion. I happen to think (as I’m sure you do) that it is an interesting completion, but it’s definitely not the only possible completion. For example, instead of a string tension as a cut-off, we could use a lattice spacing (like Regge calculus) as a completion. I am certainly not claiming that this is right, but a Regge-type lattice with lots of other fields (and Wilson terms to remove Fermion doublers) is something you can’t rule out, any more than you can rule out string theory. So there are even MORE ways to do the completion.
My own view on string theory is similar to what some people said about N=8 supergravity when I was a student: String theory is worth learning because it’s interesting, not because it’s right.
Eric,
“In a particular vacuum with all moduli stabilized , all parameters such as gauge, Yukawa couplings are fixed as these depend on the moduli. In addition, it is not possible to just construct any model you want as there are usually a host of consistency conditions that must be satisfied that strongly constrain the models. Thus, if one can find a model which describes exactly the SM in detail in its low-energy limit, then this would be a very significant discovery.”
Well, there are a host of consistency conditions on any model you use to fit data. As I said to Somebody, let’s all finally admit that the world is just a Lagrange polynomials. Sorry, but fits are not physics.
somebody, etc…
I don’t think it’s possible to have a sensible discussion about this, except by looking carefully at actual attempts to construct string theory vacua. This is why I wrote the earlier posting about the claims to get predictions out of an F-theory construction. My argument is not that having lots of string theory vacua inherently makes it impossible to test, but that you have to examine what these vacua look like and see what can be gotten out of them.
Sure, it’s possible to imagine a string theory vacuum construction that, given some experimental input, is predictive and testable. Problem is, people have been trying to do this for almost 25 years, and have gotten nowhere. Take a look at the F-theory paper. Taking as input everything we know about particle physics, and engaging in 300 + pages of complex constructions, the end result is a vague claim about the neutrino mass scale, that it is more or less where we already know it is. 25 years of this kind of thing is exactly what you would predict would happen if people pursue a wrong idea and refuse to stop.
Similarly, going on about how string theory completely stabilizes all moduli, making, for each vacuum state, precise testable predictions about SM parameters is just pure hype. That’s a dream about what some people would like to be true, not the current state of string theory. If people want to pursue that dream, fine, but they should acknowledge that they are nowhere near this, and that the current state of the subject gives no reason to believe that even if you could do such calculations, they would turn out to be predictive and explain anything about particle physics.
“Similarly, going on about how string theory completely stabilizes all moduli, making, for each vacuum state, precise testable predictions about SM parameters is just pure hype. That’s a dream about what some people would like to be true, not the current state of string theory.”
Actually, this isn’t hype, this is in fact the current state of string theory, as moduli stabilization has been a very important area of research for the last 10 years or so. It is really only a matter of time before a specific model is found which completely describes the SM and for which all moduli are stabilized. Progress in physics goes on despite the Woit’s of the world.
Eric,
“Actually, this isn’t hype, this is in fact the current state of string theory, as moduli stabilization has been a very important area of research for the last 10 years or so. It is really only a matter of time before a specific model is found which completely describes the SM and for which all moduli are stabilized. Progress in physics goes on despite the Woit’s of the world.”
In other words, it is only a matter of time before string theorists find a fit. See my remarks above.
Peter O.,
No, once the moduli are stabilized there no free parameters. Thus, all physical parameters are determined dynamically. This is quite different than just finding a fit.
> In other words, it is only a matter of time before string
> theorists find a fit. See my remarks above.
I explained to you very clearly why it is not just a “fit”. Because once you fit a vacuum every new experiment is a prediction. Why do you not acknowledge this issue and keep repeating your original, misinformed statements again and again?
What I find strange is that the critics are NOT attacking the weaknesses of string theory, which are real!!! There are two issues here. One is the construction of standard model-like vacua, which as Eric emphasized to you is a difficult problem, which has not yet been fully solved. The second is the issue of the landscape (which was what I was talking about), which is that we expect that there will be many vacua which can reproduce the standard model, once we construct ONE. Can you see all the leaps of faith employed by the string partisans here? Why are you not attacking those (relevant) issues?
Frustratingly, you seem to be not getting any of these subtleties and just keep repeating your irrelevant soundbites again and again. I apologize to Peter (Woit) if I sound a bit harsh.
Sorry Eric. I don’t get it.
You need to chose moduli, i.e. a vacuum. This is designed to fit your data. Then you want to use this vacuum to make predictions. If there is a unique choice of moduli (this strikes me as unlikely), you are fixing these moduli as parameters. If the choice of moduli isn’t unique, you will simply have more parameters than you need (but okay, maybe there are some unique predictions within this parameter set).
Here is my question. How is this better than using any method to fit data by some function. My joke about the Lagrange Standard Model aside, any fitting method will make predictions. Don’t say it’s because it comes from string theory. You have to do better than that.
Somebody,
“I explained to you very clearly why it is not just a “fit”. Because once you fit a vacuum every new experiment is a prediction. Why do you not acknowledge this issue and keep repeating your original, misinformed statements again and again?”
This is why we fit data by functions. Experimentalists do it all the time. Once this is done we can make predictions. For example, we might try to fit some data by a straight line using a chi-squared fit, or we fit data by a Lagrange polynomial fit. But it doesn’t mean we have a correct theory. I asked Eric above why choosing moduli to fit data is better. If the answer is “because we like this way of doing things, since it comes from string theory”, it just isn’t good enough.
Let me put it another way. You fit data by a string vacuum. I fit it by Lagrange polynomials. We can both make predictions, and since we have both parametrized the data by smooth functions, they won’t differ very much. Why is your method better?
Peter O,
The moduli are not ‘chosen’, they are determined dynamically by a potential which is generated by turning on fluxes or other non-perturbative effects. For a particular construction, the fluxes are generally tightly constrained by consistency conditions. For example, in Type II vacua the flux is constrained by tadpole equations as well as supersymmetry conditions.
> This is why we fit data by functions. Experimentalists do it all the
> time. Once this is done we can make predictions. For example, we
> might try to fit some data by a straight line using a chi-squared
> fit, or we fit data by a Lagrange polynomial fit. But it doesn’t
> mean we have a correct theory.
Of course it does! If the next experiment tells you that the data lies on the same straight line, your straight line theory was correct! The only difference in high energy physics is that the fit that you are trying to make is so involved that there is a whole subculture called “theory” that is devoted to it. This is ALL there is to theory. You might want something more grandiose, but there isn’t anything more grandiose.
The real subtlety is that the thing that you are trying to fit is NOT always a straightline. There are many comlicated things that you are trying to fit at the same time. But apart from that “technical” issue, the analogy works.
“The moduli are not ‘chosen’, they are determined dynamically by a potential which is generated by turning on fluxes or other non-perturbative effects.”
Wait a minute, the moduli are determined?…this negates the whole idea of the landscape…
But I think you mean that there are consistency conditions on the choice of moduli, which have to do with things like p-form fluxes. Just as a Lagrange polynomial used to to fit data is constrained by the assumption that it IS a polynomial. Any method used to parametrize data will have its own idiosyncracies. But the results will only be a little different, assuming Nature likes smooth functions. So I ask again: why is your method better?
> You fit data by a string vacuum. I fit it by Lagrange polynomials.
I would like to see you fit all the high energy experiments for for example the standard model, by a spline fit. Its not one set of data that you have to fit, but every possible sets of data.
“If the next experiment tells you that the data lies on the same straight line, your straight line theory was correct!”
No! A straight line is not a theory. It is just a fit (NOW am I getting through?).
“The real subtlety is that the thing that you are trying to fit is NOT always a straightline. There are many comlicated things that you are trying to fit at the same time. But apart from that “technical” issue, the analogy works.”
OK, so don’t fit with a straight line. Use a Lagrange polynomial. You can fit anything this way. Why is your way better?
“I would like to see you fit all the high energy experiments for for example the standard model, by a spline fit. Its not one set of data that you have to fit, but every possible sets of data.”
I’m sorry, but I can fit anything with a Lagrange polynomial, no matter how much data you give me (assuming I was a good enough programmer). Why are the moduli (constrained by whatever nonperturbative effects you want) better than the coefficients of this polynomial?
Peter O., a theory is useful because it can fit all data. If there is a finite number of experiments you can do and then predict EVERY single new experiment in our Universe, you will have the final theory. If you can some up with it, by curve-fitting if you will, thats all you need. In any event, this discussion is beyond ridiculous now and Peter (W.) is probably going to get upset, so unless there is something new you have to say…
No, no, I said what I have to say. I am happier than a pig in &^%&%&^$# that you finally agree that what people are doing with the landscape is curve fitting. What we disagree about is whether that constitutes a physical theory. I say it doesn’t.
Ummm, maybe enough about this particular point, OK?
What I’ve found in discussions here is that abstract arguments tend to degenerate, because string theory advocates point to the most optimistically possible situation that doesn’t completely violate the rules of logic, and, by this measure, sure you can’t rule out string vacua scenarios. That’s not the measure usually used to evaluate a research program though, which is results achieved so far, and ones which seem plausibly achievable considering how things have been going recently.
Sorry to let this argument continue for so long Peter. I believe, however, that is goes to the very core of the value of the landscape and that somebody, Eric and I performed a useful service is illuminating the issue.
“Wait a minute, the moduli are determined?…this negates the whole idea of the landscape…”
No, the landscape emerges because there are many different models with differerent gauge groups and matter content for which the moduli may be stabilized rather than just a single model which is unique.
Eric, that’s exactly the same thing as saying that moduli can’t be determined. I.e. the reason you can’t determine string theory moduli (to allow checkable predictions to be made from string theory) using the 19 empirical standard model parameters is precisely because string ‘theory’ isn’t a unique theory, but is instead a collection of a vast number of different models. So the inability to determine moduli is the same thing as having a landscape. Any theory with a collection of parameters that can be empirically constrained to no less than 10^500 possibilities, then that’s identical to having a landscape size of that magnitude.
Er, Peter is going to mad at me. But Eric, so what? The acceptable moduli are discrete rather than continuous. But there are so many choices that they might as well be continuous. Just curve-fitting.
OK, this is my last word. Eric, blast away in my absence.
Sorry folks, anything else on this point will have to be insightful and brilliant beyond words for me to not delete it.
If you will pardon this interjection from a purely lay perspective, and focussing for a moment on the New York Times article, then is Witten’s perspective – or indeed, that of physicists collectively – really one of ‘despair’.
In reading Peter’s brilliant book, Witten comes across as a rare figure in the modern landscape; a man with an eclectically wide range of interests who uses these to synthesise new viewpoints.
In many ways this seems to me to be similar to Einstein’s world view; he considered the physical world from many disparate viewpoints in order to construct a new understanding of space and time.
If eminent figures such as Witten are now starting to consider the physical landscape from an anthropic perspective – and just setting aside for a moment the instinctive revulsion many people have towards this approach – isn’t this simply a sign of a brilliant mind exploring a different perspective simply to see if it yields results, rather than an abandonment of the scientific principle?. Certainly this doesn’t seem like an act of despair.
Although Peter Woit has – quite understandably – expressed negativity regarding this approach – it does, after all, smack of some kind of recursive navel-gazing – it does seem to lead to some interesting lines of inquiry.
Most obviously, we can take the 20-something arbitrary parameters of the standard model and adjust each one by a small amount to see what the outcome is with regard to the evolution of the universe (e.g nucleosynthesis etc). Some parameters will presumably be more sensitive than others. Suppose it turns out there are some fantastically huge number of possible universes, say, with no matter as we understand it at all, only photons, some lesser number with – perhaps – photons and leptons – a smaller number again with fermions as well. And so on down to the universe containing only hydrogen, etc. etc. (or must all possible universes contain all the fundamental families of particle in the standard model, an interesting question in its own right, I’d have thought).
Then – and I think this is the spirit in which Witten is proceeding – let’s assume our observable universe is potentially a ‘bubble’ evolving from some alternate universe in which the laws of physics differ. We assume some probable alternative configuration based on our adjustment of the standard model parameters and then we calculate, for example, how our universe might evolve and expand from its parent. (would it, for instance, be possible to show that such expansion would accelerate and if so under what circumstances?).
Now this could be dismissed as crackpot science but there are testable hypotheses here, are they not, so it sounds to me like science. Certainly it involves entertaining some rather speculative hypotheses as a starting point, but how is this different from Einstein imagining what an observer ‘sitting on a photon’ would see, which appears to be his starting point for his exploration of relativity. At the time, there must have been many of his fellow physicists who at first thought his views untenable.