Why I am REALLY disappointed about string theory

[tom.stoer]

So with my statement that

in ST ... the web of approximate dualities is perhaps too weak to be called a theory. It is perhaps not sufficient to stay with this web of approximate dualities w/o being able to identify the underlying unifying structure. The initial idea and program of M-theory is still waiting for completion!

and your agreement in the sense that you

prefer to view ST as a framework that generalizes QFT [to QFT+gravity]

we can now conclude this discussion, can't we?

 

[atyy]

An interesting review.

String Theory as a theory of quantum gravity: a status report
Matthias Blau and Stefan Theisen
http://www.blau.itp.unibe.ch/stqg.pdf

They conclude "String theory is a very promising (and fertile) framework for a consistent theory of quantum gravity. However, we still appear to be at a rather preliminary stage of our understanding of this theory. In particular a non-perturbative formulation of the theory and uncovering its symmetries are important open issues. Matrix models and AdS/CFT correspondence mark recent success on the former issue, while hyperbolic Kac–Moody algebras might be the right language for the latter. Thus, far-reaching statements on either side of the string theory debate, proclaiming either the imminent demise of string theory or the ultimate unavoidability (and virtue) of the anthropic/multiverse scenario, appear to be pre- (and im-) mature, and should not distract one from trying to better understand profound quantum gravitational issues to which string theory presumably holds the clue."

 

[marcus]

An interesting review.

String Theory as a theory of quantum gravity: a status report
Matthias Blau and Stefan Theisen
http://www.blau.itp.unibe.ch/stqg.pdf
...

It is interesting! Thanks for posting the link. When one sees a review that is largely non-mathematical reportage, interpretation, and opinion its helpful to know who are the reporters. So I checked here for Prof. Theisen. I don't know of him but it speaks well that he is at the Einstein Institute (MPI Potsdam) which also has LQG and Group Field Theory teams.
http://www.aei.mpg.de/english/php-Skripte/quMembPage/index.php?%20personKey=theisen
I also looked for Stefan Theisen's publications
http://arxiv.org/find/grp_physics/1/au:+Theisen_S/0/1/0/all/0/1
His most heavily cited papers are from the 1990s, especially mid-90s
http://www.slac.stanford.edu/spires/find/hep/www?rawcmd=FIND+A+THEISEN%2C+S&FORMAT=www&SEQUENCE=citecount%28d%29

I must also try to find Prof. Blau...yes, here!
http://www.blau.itp.unibe.ch/blau.html
http://arxiv.org/find/grp_physics/1/au:+Blau_M/0/1/0/all/0/1
http://www.slac.stanford.edu/spires/find/hep/www?rawcmd=FIND+a+Blau%2C+M&FORMAT=www&SEQUENCE=
http://www.slac.stanford.edu/spires/find/hep/www?rawcmd=FIND+A+BLAU%2C+M&FORMAT=www&SEQUENCE=citecount%28d%29
Very respectable. Obviously these people are quite well-known (just not to me).

Likewise Blau's most highly cited papers are from 1990-2002. Now at Bern, but then he was mostly at Trieste ITP, part of SISSA (?), I think for instance Percacci is at Trieste. Both Blau and Theisen started publishing in the mid-1980s, like 1984-1986.

 

[Fra]

I have to conclude that I dislike A.N. Schellekens's reasoning and his reasoning with regards to the gedanken experiments of constructing a anthropic measure that would yield "viable" windows of life in this "theoretical theory space" - which he defends, but he rejects entropic selection and compares the computation of unique statistical distributions with unique theories and thinks it would be a mystery why the statistical prediction would happen to be inside his viable windows of "theory space".

I agree with one thing and that is that it's obvious that the notion of "statisics" needs to be defined, in such a contex, as it's obvious that there is no observer, or scientist that could ever collect and verify this statistics in "theory space".

My opinon is that this, including his insisting on a mathematically obvious "theory space" (gauge theory place) is founding this problem.

If we instead constrict ourself to only realizable measures, the only way is to let an inside observer physicall construct these new measures that would eventually answer his question of "what does statistics and probability mean in this case". That would even apply to any antrhophic measure or viable theory paramteers as even this measure needs to be constructed from the inside (and not in external mathematical spaces); then these two measures should conincide - this there is no "conicidence" that entropic predictions match the viable windows, because they are the same thing if you accept that they must be intrinsically constructed.

I think the main problem is the idea that there exists a unquestionable an observer independent and static theory space. This in itself IMHO an unscientific notion that I find hare to make sense out of. It comes out to me as a mathematical idea; having no clear relation to phyiscs.

/Fredrik

 

[mitchell porter]

Just on a very basic level, I now think the most exciting thing happening is the development of worldvolume theories for M2-branes and M5-branes - the fundamental membranes in M theory. There was a "membrane minirevolution" for the M2-branes a few years ago, and work on the M5-brane is already leading to spinoffs like Witten's "Fivebranes and knots". In the end I think this should give us something like a new fundamental equation for M theory.

 

[tom.stoer]

In the end I think this should give us something like a new fundamental equation for M theory.

I hope so.

I thought about restarting this discussion for a while but concluded that this is not possible. There is one problem when discussing string theory, namely that (as we agreed) there is currently no fundamental formulation available. But w/o such a formulation I do not know which questions I should ask.

 

[MTd2]

The only thing I found with "Fivebranes and knots" was this August 17th talk:

http://online.kitp.ucsb.edu/online/duallang_m10/witten2/

Is there anything else?

 

[Fra]

In the end I think this should give us something like a new fundamental equation for M theory.

Can someone explain what and if such equation imply for the landscape and ambigousness problem? Does it improve the predictability issue?

Is the idea that we count the solution space to M-theory so that we get at best a probability measure? or would ST still have to resort to strage anthropic reasoning?

/Fredrik

 

[tom.stoer]

Can someone explain what and if such equation imply for the landscape and ambigousness problem? Does it improve the predictability issue?

Is the idea that we count the solution space to M-theory so that we get at best a probability measure? or would ST still have to resort to strage anthropic reasoning? /Fredrik

This is not the issue. The main problem today is that you cannot say what string theory really IS; you cannot describe the whole solution space with one fundamental formulation. Having such a formulation at hand you may be able to address new or currently unsolved problems, you may find new symmetry principles, you may be able to identify a selection principle = something that restricts the vacuum degeneracy etc.

 

[Fra]

fundamental formulation. Having such a formulation at hand you may be able to
...
identify a selection principle = something that restricts the vacuum degeneracy etc.

That _was_ my main issue here, thanks.

But I'm not sure I understand the rationality of the method; since one can not "find symmetries" in the observational sense this this is all speculated abstractions. So, let's suppose we CAN describe the full solution space. The question still remains, what methodology to use to find the right constraints. In principle one can imagine that up to "experiments" using random or linear search given that you can count the solution space, but that sounds unlikely to be a good searching method.

(To me "string theory" certainly isn't a theory in the sense of there beeing an equation where you set the input and get the output, in my eyes it's a way of approaching an open problem (ie. a strategy or inference construction), that starts by postulating the FORM of the microstructure where information is coded; and then attempts to exhaust all possible constructible actions and tries to find in that the actions of the SM. The question is HOW this leads to increased predictability rather than disproportionally inflating the set of possibilities. I see this from the pure inference perspective. But these was also already concluded in the thread; that it's a 2framework etc". I actually have not "problem with this" - ie the fact that it's a framework and not a definite theory; MY concern is wether it's the RIGHT framework at all. )

Edit: As far as I understand, ST theorist find motivation for this framework in part by an ad hoc original idea (STRINGS instead of points) and then combining it with QFT and assuming it makes sense to extrapolate QM to cases where it's nto verified). Then I guess it's possible to be comitted to the way of reasoning, even though the goal isn't in sight, if it seems like the only rational option.

/Fredrik

 

[tom.stoer]

Look at QM. In became a fundamentally well-defined theory by changing the formalism from ad-hoc quantization a la Bohr-Sommerfeld to canonical quantization in Hilbert spaces a la Dirac etc. In doing that it was NOT the case that all questions were answered, INSTEAD some of them simply disappeared. I guess something like that must happen in string theory as well.

 

[marcus]

...
(To me "string theory" certainly isn't a theory in the sense of there being an equation where you set the input and get the output, in my eyes it's a way of approaching an open problem (ie. a strategy or inference construction), that starts by postulating the FORM of the microstructure where information is coded; and then attempts to exhaust all possible constructible actions ...

It also has another aspect---not so much a "way of approaching" as a directionless floundering in differential geometry. The mathematics of differential manifolds is astonishingly rich---a fertile womb of abstraction that brings forth wonders of every kind.

One can be so surprised by a few uncanny coincidences that one becomes superstitious.

 

[tom.stoer]

In the meantime I became confused about the fact that string / M-theory deals with strings and/or certain branes.

My feeling is that if we assume that spacetime is somehow quantized on a fundamental level, it does not make sense to use smooth objects as basic buildung blocks. If one looks at LQG then the lesson is that smooth objects are useful at an intermediate stage but that in the very end they will disappear from the theory. I guess that the same applies to string theory. In that case strings or branes would emerge as effective degrees of freedom valid in a certain domane, energy range or something like that. The question then is which principles derived for world sheet geometry, conformal invariance etc. could survice this reformulation and become principles of the fundamental theory.

 

[marcus]

... it does not make sense to use smooth objects as basic building blocks...

this sounds like it could be the verdict of History.

 

[suprised]

In that case strings or branes would emerge as effective degrees of freedom valid in a certain domane, energy range or something like that.

Exactly - that's how many people look at it. It is not at all clear whether there must be any "more fundamental" theory behind, perhaps all there is are effective theories (in the sense that they make sense only in certain regions, never globally).

 

[atyy]

Exactly - that's how many people look at it. It is not at all clear whether there must be any "more fundamental" theory behind, perhaps all there is are effective theories (in the sense that they make sense only in certain regions, never globally).

If string theory is only effective, then does one need 10 dimensions still?

 

[suprised]

If string theory is only effective, then does one need 10 dimensions still?

Translate 10 (rather 6 of them) to "internal, non-gravitational degrees of freedom", and then the answer is yes. And obviously this is a desirable feature.

The meaning of "effective" might have been confusing here - it was not in the usual sense of "low energy effective action" which arises from integrating out massive states; and which is incomplete and ill-defined above some energy scale, so that some other, more fundamental theory would take over above this scale.

I meant "effective" in the spirit described earlier: namely as analog of coordinate patches on some abstract manifold, each of which captures only part of it and there isn't a set of globally defined "master" coordinates what would be well-defined everywhere. In other words, there would not be any "master" theory that would display more fundamental degrees of freedom. On the contrary, by all what we know so far string theory seems complete, there is no indication that any states would be "missing" (for example, to account for the entropy of black holes).

To what extent "emergent gravity" is an indication to a more fundamental theory, or not, is unclear to me and certainly a matter of debate.

 

[tom.stoer]

Are there attempts to identify these fundamental (discrete) building blocks?

A couple of years ago people focussed on matrix theory, but in the meantime it became quiet about that topic. What about an algebraic approach, e.g. hyperbolic Kac-Moody algebras or 3-algebras? Do they have the potential to provide a more fundamental formalism or are they just another couple of "coordinate patches"?

To be honest: I understand the idea of these "patches", but I dislike it. It sounds like surrender.

 

[suprised]

Are there attempts to identify these fundamental (discrete) building blocks?

Of course, in the corner where strings are well tractable (10/11 dimensional theories), this is a solved problem. This is where these ideas came from, they didn't come out just out from the blue.

A couple of years ago people focussed on matrix theory, but in the meantime it became quiet about that topic. What about an algebraic approach, e.g. hyperbolic Kac-Moody algebras or 3-algebras? Do they have the potential to provide a more fundamental formalism or are they just another couple of "coordinate patches"?

Same thing - coordinate patches. Matrix theory deals well with the high-dimensional theories, but not with the low-dimensional ones. Algebraic approaches never lead to anything concrete so far.

To be honest: I understand the idea of these "patches", but I dislike it. It sounds like surrender.

Well.. it's not a matter of liking or not. This is simply the situation as it appears for the higher dimensional theories, as a result of definite computations. The question is whether this picture applies generally, by extrapolation, to all string theoeries. And this again, in principle, is a matter of computation.

This is what I counter-criticise here. Often people have opinions about what a fundamental theory of nature should or should not be like. Whether emergent, relational, based on "inference", fundamentally discrete, background independent, etc. All this is fine, as are political or religious opinions. But what should be realized that theoretical physics is about actual computations and their results, rather than ideological beliefs and wishful thinking. Every fool can cook up some half-baked ideas, but to make it actually _work_ (or at least make some concrete sense), is almost infinitely more difficult.

 

[tom.stoer]

This is what I counter-criticise here. Often people have opinions about what a fundamental theory of nature should or should not be like. Whether emergent, relational, based on "inference", fundamentally discrete, background independent, etc. All this is fine, as are political or religious opinions. But what should be realized that theoretical physics is about actual computations and their results, rather than ideological beliefs and wishful thinking. Every fool can cook up some half-baked ideas, but to make it actually _work_ (or at least make some concrete sense), is almost infinitely more difficult.

Look, the problem is simply this: string theory claims to be the fundamental, unified theory of nature (and I agree that it is the only candidate as I see no other theory which has comparable same unifying ambitions). But unfortunately string theory is less developed regarding its fundamental formulation (degrees of freedom) compared to other, less ambitious theories (LQG for example has identified its fundamental degrees of freedom, QCD succeeded as well; of course there are a lot of open questions in LQG - just as in ST; of course QCD is focussed on string interactions only; but that's not relevant for my example). So there is a theory which claims to be the fundamental theory of nature, but at the same time is not able to explain what its fundamental buildung blocks ARE.

Several conclusions are possible:
- there are no fundamental building blocks, but coordinate patches and "effective" degrees of freedom only
- string theory is "work in progress" with one outstanding problem - what are its fundamental degrees of freedom
-...

I cited Gross here several times; and I think this is again a good point to come back to his central questions WHAT IS STRING THEORY? What is the fundamental formulation of string theory?

 

[suprised]

I cited Gross here several times; and I think this is again a good point to come back to his central questions WHAT IS STRING THEORY? What is the fundamental formulation of string theory?

Well clearly no one has an answer for this, all that I am doing is playing with the idea that a more fundamental set of degrees of freedom might not exist. Only if there were a regime (say at ultra-high energies) where those new degrees of freedom would "liberate", it would make sense to talk about them; non-oberservable quantities should be avoided. But there are indications that when going up in engergy, beyond the Planck scale in particular, nothing of that sort happens (see eg the recent papers by Dvali & Co).

 

[tom.stoer]

I understand you speculation, but my conclusion is different simply because of the rather immature status of the theory; or worded positively because of its huge but still latent potential. In a rather well-developed theory I would agree, but as there are continuously new and surprising (and hopefully promising) research directions I think it's too early to give up.

As this thread originated in my disappointment regarding (the status of) string theory I have to admit that abandoning they idea of identifying one fundamental formulation would be another disillusion.

 

[suprised]

Well it is probably obvious that things are not developed enough to draw a conclusion on these matters, time will hopefully tell.

 

[Fra]

In that case strings or branes would emerge as effective degrees of freedom valid in a certain domane, energy range or something like that.

Exactly - that's how many people look at it. It is not at all clear whether there must be any "more fundamental" theory behind, perhaps all there is are effective theories (in the sense that they make sense only in certain regions, never globally).

As someone have a different approach and strategy to understanding physics, I still can agree that that understanding the various more or less dual "choice of microstructure" of ST (ie different strings, branes etc) as an effective or "relative" in a different context is the one way where some of ST can make sense to me as well. But then, as far as I imagine this (for me strings would then be coninuum limits of more combinatorically discrete microstructures) the new constructing principles from which a continuum string or brane might be "emergent" in the high complexity limit (with high complexity I refer to the AMOUNT of information thta is in fact encoded in a continuous memory strucutre; even if there is redundancy there, the apparent information is still massive) would be so different from the constructing principles of ST that I'm not sure if it would make sense to associate it with string theory as we know it today? With this I mean that it is equally possible that other programs; say LQG with their spin-networks ideas, can expand this to encode also internal information and matter and that it can be developed to the same limit.

It is not at all clear whether there must be any "more fundamental" theory behind, perhaps all there is are effective theories (in the sense that they make sense only in certain regions, never globally).

I symphatize with this and I think this is the case. I think the idea of fundamental (fixed observer independent) microstructure for information encoding is an idea that isn't verifiable. I think too that all there is are effective degrees of freedom (except I prefer to think in terms of discrete complexions, not continuum structures). Ultimately each observer, "sees" not only different states, but different state spaces. This is also why I insist interpreting the string background choices as choices of observers, and to me the non-uniquess is simply due to the non-uniquess of observer. Thisi s notrhing strange, but I just object to how ST treats this. To me this also applies to tha laws of nature, there are not objetive laws that a give observer can infer with certainty. Only effective laws are inferrable. To infer that the effective law is in fact fundamental is not possible as far as understand. Wether that's politics I don't know, but for me it's a quite clear point.

/Fredrik

 

[tom.stoer]

Frederik,

I think you are mixing up two different issues.

I roughly understand your ideas regarding observer-dependent descriptions, but I think we should keep them separate from the discussion regarding a unique formulation of string theory.

One can have a unique framework which allows for observer-dependent descriptions. In quantum mechanics you have on Hilbert space on which you can select different sets ob observables, related by unitary transformations, reflecting observer-dependece; e.g. a two-particle system as seen from the the center-of-mass frame. In quantum gravity with the holographic principle one expects something like boundary Hilbert spaces defined by observer-dependend Dirac observables, nevertheless there is one unique mathematical framework (certain aspects are visible in LQG when introducing horizons and horizon-degrees-of-freedom from Chern-Simons theory). This is what I expect to be incorporated in a unique formulation of string theory as well.

What you are saying is that in addition the mathematical framework itself becomes observer-dependend. I understand your idea but I would like to stress that this is a second step beyond the more basic discussion we have here. It depends how you interpret the situation in string theory as of today. Does it reflect the impossibility of an observer-neutral unique framework? Or is it simply due to the limited understanding of the theory? I guess it's the latter.

Look a quantum mechanics: wave mechanics (Schrödinger) and matrix mechanics (Heisenberg) have been proven to be strictly equivalent in a broader picture provided by fully developed quantum mechanics. Whereas in the early days of QM it appeared mysterious how two different formalism could generate the same predictions, today it is clear that they can be derived from a fundamental theory using the notation of abstract Hilbert spaces; today you have the choice which description to use - depending on the problem you are trying to solve. I expect something like that to happen in string theory as well.

So for me me the different "coordinate patches" or "theories" are not fundamental entities due to some deep principle but simply due to the limited understanding of what string theory is. So instead of accepting these losely coupled patches I would like to insist on their unification.

-------- edit --------

@Fra: let me ask the other why round: what is the reason that in string theory it is not possible to identify a unique fundamental formulation whereas in other (less ambitious) theories like QCD, LQG it is possible?

 

[Fra]

I think you are mixing up two different issues.

Yes you're right, I do, but it's not due to my ignorance. I'm making it into a point.

From my perspective at least (which is of course as biased as anyones) consistency of reasoning implies that they do mix. The physical process of one observer inferring the interaction rules for two other observers or subsystems are subject to the same rules as the process whereby the two first observer interaction from their inside perspectives. What I am suggesting is the deeper idea that they are mixed by nature, rather ME "mixing them up" due to confusion. I'm just trying to acknowledge this "mix" as a fact of nature, and try to use it to understand it.

One can have a unique framework which allows for observer-dependent descriptions.

Alot of people, includiding rovelli, would agree with you on this, and I certainly know what you mean.

But your "point" here is what I call a form of structural realism. I'm suggesting that this realism is not something we need, and that this form of realism still is a source of confusion.

Let me put it like this; I am not saying this form of objective descriptions are impossible, I'm just saying that it's not possible for a physical inside observer to DEDUCE the observer-independent status beyond the effective level. An inside observer can and DO infer EXPECTATIONS of such observer invariance - this is the way I explain the current objective descriptions or QM, GR and SR. But in the perspective I suggest this is special cases.

This is also part of similar to part oF smolin/ungers argument against timeless laws. Smolin makes similar argument but my understanding is that unger takes this to a deeper level than smolin currently appreciates.

What you are saying is that in addition the mathematical framework itself becomes observer-dependend.

Yes.

I understand your idea but I would like to stress that this is a second step beyond the more basic discussion we have here.

This is probably true, but I do not see the value in try to aim for an obviously flawed intermediate next step. I do agree with you, that improvements probably is possible, that is in line with the common structural realism. But to have this as a target, when it seems clear that it's still flawed is not rational.

If the reason is that it would be "easier" to make this smaller step first, then I am not so sure I agree about that. I think we have reached the point with regards to the nature of several open problems that a more radical new thinking is needed.

Does it reflect the impossibility of an observer-neutral unique framework? Or is it simply due to the limited understanding of the theory? I guess it's the latter.

I would say that string theory as it stands now, does NOT imply this. So I think you are right that it's an incomplete understanding of ST. BUT, my reasons for believe in impossibility of observer independent fixed framework comes from a totally different direction! I just project an understanding on the string landscape from a different direction.

So if we eventually understand that such objectivity is not possible, I think it is certainly not thanks to ST as I see it. ST seems more due to a conincidence due to, like marcus also notes, it's nature of mathematical fiddling (but guided by some questinonal extrapolations of QFT principles and extra ad hoc assumptions)

@Fra: let me ask the other why round: what is the reason that in string theory it is not possible to identify a unique fundamental formulation whereas in other (less ambitious) theories like QCD, LQG it is possible?

I can't provide the correct string theory answers (ie what the reason is from the point of view of a string insider); this is what I'd like to hear the insiders view, and it's why I appreciate surprised contribution here.

But as I understand it (ie. projecting it onto my reasoning) is that, loosely speaking:

The original starting point of string theory, where one postulates the microstructure of where information is encoded (ie the continuum STRING, the background space) and the string ACTION which encodes the inference machinery contains too many degrees of freedom that are unknown, and the "consistency" principles ST uses from QFT and so on, to constrain the possibilities are simply too weak and you still end up with an extremely large set of "possibilites" that simple originate from the massive amount of information that is encoded in the continuum strings and the choice of background spaces. There is also an unclear understanding of WHY the correct action correspond to quantized classical string.

Alot of this is IMO traced to the ambigoty if starting with using continuum structures as coding structures, beause it's ambigous howto count and quantify information in these structures. This ambigouty itself yields a landscape of possible "limiting procedures".

From my inference perspective, it is not sensible to start out with an uncountalbe set, and try to apply reasoning since the only way to make any sense of that requires to view the uncountalbe infinite set as a limiting case of something if not finite, at minimum countable.

So the "gigantic landscape" of string theory, in my view corresponds (If at all!) to a highly evolved very complex and massive observer. And the evolution history would hold the answer to WHICH structure is the "right one".

/Fredrik

 

[Fra]

The physical process of one observer inferring the interaction rules for two other observers or subsystems are subject to the same rules as the process whereby the two first observer interaction from their inside perspectives. What I am suggesting is the deeper idea that they are mixed by nature, rather ME "mixing them up" due to confusion. I'm just trying to acknowledge this "mix" as a fact of nature, and try to use it to understand

An illustrative point of disagreement here with Rovelli is his RQM paper, he first makes some excellent point, such as there beeing no objective states and that there are only relative states or relations between observer and observed, and no absolute relations. The only way for two observer to compare their observations is by means of interaction = communication.

So far, it's all good. But the coherence of his reasoning when he flatly assumes that all communiation follows QM structure. He further explicitly avoids discussing the meaning of probability etc.

Here rovelli resorts to structural realism and faith in the structure of QM, in a way that I find inconsistent. His reasoning is not consistent.

Instead of assuming that observer independent structures are forcing, I'm considering observer democracy where the observer indepdepent laws are evolving constraints, where this evolution contains darwinian and unpredictable elements.

The string landscape somehow seems to attempt to describe the complete statespace, instead of considering it to be evolving. It's both too small and too large! An evolving statspace in evolution is always sufficiently large to be adaptive, but not too large to stall progress.

/Fredrik

 

[tom.stoer]

Thanks for the long reply. I think we mostly agree, especially about the difference between the radical change in science you are proposing, namely to abandon structural realism in which all scientific theories are deeply rooted, and the more technical details regarding string theory.

I disagree with you that it makes no sense to "aim for an obviously flawed intermediate next step". String theory isn't fully developed ad there is a huge potential regardless if one agrees to your reasoning or not. In addition I do not agree that your conclusion is "obvious". And I think that regarding string theory it is not true that "we have reached the point with regards to the nature of several open problems that a more radical new thinking is needed".

I am perfectly aware of the fact that I am limiting myself to standard reasoning, structural realism etc - at least here in this discussion regarding string theory.

Your argument against string theory using smooth structures and eventually deriving fundamental discrete ones is not valid. LQG did exactly this: start with smooth structures and derive discrete ones. You need not love LQG but you should agree that reducing smooth structures by large symmetries to descrete structures seems physically reasonable. Anyway - we agree on the basic problem.

You argument against the landscape isn't fundamentally enough. The landscape is only the space of solutions of a set of theories related by dualities. The argument should be against the missing uniqueness of the (formulation of the) theory itself, not against the missing uniqueness of the solutions (look at the standard model: the sun, iron, my laptop, a pint of beer, ... all solutions of the SM - by no means unique).

My conclusion is that I am still willing to rely on structural realism and that therefore the major issue for me is to identify objective, structurally simple and phenomenologically viable laws of nature. As long as this seems to be possible I am willing to accept string theory as a (candidate of a) physical theory.

 

[Fra]

Thanks. Yes I think we understand each other and have a partial agreement.

Given that you're willing to accept strucutral realism at least for the moment, your other arguments may fit in with that. I could comment on your arguments but those arguments again go back to the general reasoning, because in my perspective "solutions" and theories are a result of similar proceses; the reason for missing FUNDAMENTAL uniquesss in both cases are to me similar, but living at different levels of the inference hierarchy.

The inference has hierarchies and parts. A given theory + initial conditions may give a prediction, or more correctly an expectation - which usually is in the form of a proability distrubution of possible results. But even the theory is also a result of an inference at a depper level, the history of interaction has made us EXPECT these laws, and it is thus determining our actions and expectations.

What should in my view give the uniquess you seek; is when you single out an observer. Only once you consider a specific observer, should be expect unique theory (ie an inference machinery with produces expectations of the future from the present); the theory itself encodes somehow the history, and the theory itself is nothing more than an expectations as well. IT's just a far more confident one, than the specific outcomes OF the theory + initial conditions.

The fact that two different observers may end up with inconsistent predictions is in fact not really a physical inconsistency. It merely reflects that their expectations are different. The normal way to "restore" this is to consider a second observer that can find transformations between the observer, to restore consistency - but this consistency is restore relative to a third and usually MORE complex observer. So from the inside view a lot of the big symmetries remain broken. I know this is a different way of thinking than strucutral realism, and does present some own novel problems, but I find it easy to convince yourself that these are features of nature we should understand. I used to have faith in structural realism as well, but have come to abandon it. My main argument used to be that if the laws of physics aren't observer invariant, then what's they worth? That may sound like a reasonable point - it did to me, but once you see that laws are in a different light, there are different ways to see this.

/Fredrik

 

[Fra]

Your argument against string theory using smooth structures and eventually deriving fundamental discrete ones is not valid. LQG did exactly this: start with smooth structures and derive discrete ones. You need not love LQG but you should agree that reducing smooth structures by large symmetries to descrete structures seems physically reasonable.

Maybe I'd need to expand on this to explain. In my view, which is inference, the choice of microstructure and the choice of action is connected. But I don't want to destroy this good thread with expanding too much on this. I'll just note that yes continuum models can produce unique discrete stuff, but that's not enough for me. LQG is not constructed from intrinsic inference (like I seek) it also uses actions that are either ad hoc or inspired, borrowed from GR.

I'm thinking of a pure natural inference (essential that ALL interactions are entropic), and there computing gets important, in evaluating hte action. If both the action form and the microstructure contains continuum ghosts then it sure is possible that they somehoe match and cancel. But everything gets far more complex and you have to spent a lot of work to make the expectations cancel properly. This simply won't happen in the program I envision. On the contrarcy do I expect thta continum models may "fit" as extrapolations on the discrete structure, but I think it's easier to understand the logic of the dynamics and emergence of these continuum models if you look at the distinguishable structure.

/Fredrik

 

[tom.stoer]

The discussion was interesting, but I think we somehow lost focus.

I asked whether there are candidates towards a reformulation of string theory from which other "effective descriptions" can be obtained. Unfortunately both the standard string theories and SUGRA, as well as matrix models and other algebraic approaches were classified as "patches valid in certain regimes" but not as candidates based on fundamental degress of freedom.

Some weeks ago I asked regarding string field theory. Something more to say about that?

 

[marcus]

I don't wish to interrupt and so will only comment briefly that I find the discussion instructive and this last question quite interesting. Structural realism has several times been mentioned as a key idea here. I understand your last question to express that concern: Are there any mathematical models which a structural realist could take seriously? If I understand your meaning, a structural realist stance requires models of how really how the world is, not mere "patches" with limited applicability, or ad hoc schemes for calculating.

I may not understand this position or criterion, in which case you would need to spell it out.

 

[atyy]

The discussion was interesting, but I think we somehow lost focus.

I asked whether there are candidates towards a reformulation of string theory from which other "effective descriptions" can be obtained. Unfortunately both the standard string theories and SUGRA, as well as matrix models and other algebraic approaches were classified as "patches valid in certain regimes" but not as candidates based on fundamental degress of freedom.

Some weeks ago I asked regarding string field theory. Something more to say about that?

These guys http://www.blau.itp.unibe.ch/stqg.pdf hardly mention it. They point to http://arxiv.org/abs/0710.3017 . But they seem to think AdS/CFT and E11 stuff more promising for trying to figure out a non-perturbative formulation, and what the symmetries of string theory might be.

Incidentally, what if the question is analogous to asking for a single coordinate system to cover a sphere?

 

[tom.stoer]

Marcus,

I am not sure if I understand you correctly.

The discussion regarding structural realism was a detour only.

Simply speaking my expectation is that different observers may not agree on the details of their observations or experiments (different locations, different reference frames, different energy or length scale, ...) but they should agree on a basic framework (= the "theory") which allows them to translate or transform these different observations into each other.

Fra insists on abandoning this structural realism and allow even the framework i.e. the theory to become observer-dependent. That's why he is not worried about the "patches" one observes in string theory (as of today).

I do not ask these question because of the discussion regarding structural realism but simply because it always was and still is my expectation (since I started to study physics): that physics is able to unveil parts of "ethernal truth of nature". Physics over the last centuries identified phenomena on different scales, energy ranges etc. At the same time scientists always insistet on the same paradigm, namey that even so there are a plethora of phenomena, behind them always is a small set of eternal laws. So I simply stick with this approach and insist on the existence of eternal laws which we should partially uncover.

That's why I find it puzzling to be comfortable with a situation where there is a plethora of different formalism suitable for specific scenarios, domains, energy or coupling ranges etc.

I tried to find examples where (looking backwards in time) one may find a similar situation. E.g. in physics regarding the strong interaction one finds S-matrix, current algebra, chiral perturbation theory, non-relativistic quark models xyz bag models etc. Now we know that the existence of these theories is NOT due to observer-dependent theories, but simply due to limited knowledge regarding the strong interaction. Today we know that we can to a certain extend derive these theories as effective theories of QCD. We observe their limitations not only based on phenomenological considerations but we are able to understand both their successes and their failures based on QCD. Even so QCD is still not able to allow for the calculation for soft pion-proton scattering (which is a simple task within the Skyrme model) one is able to understand to what extend one can derive or motivate the Skyrme model from QCD. That's why I am absoluetly sure that QCD is "more true" than the collection of low energy effective theories I just mentioned.

As I am not an expert in string theory I follow the real experts' reasoning explaining to me that the theory is still in an early stage of development. If this is true (and it's not up to me to question this) then the natural conclusion is that we should insist on finding this "eternal law behind string theory". I want to make clear that "eternal law" need not mean that string theory is true. An SU(4) gauge theory of the strong interaction is physically wrong, even so we know its eternal law :-)

Regarding my first post where I tried to explain why I am REALLY disappointed about string theory: it is exactly this, namely that up to now nobody is able to explain what string theory fundamentally IS.

 

[Fra]

Just a last note on this

Now we know that the existence of these theories is NOT due to observer-dependent theories, but simply due to limited knowledge regarding the strong interaction.

From my perspective, there is no difference between the two possibilities you describe.

I like to quote Zurek here (although I take the meaning further than I think Zurek did):

"What the observer knows, is indistinguishable from what hte observer IS"

My point of intrinsic inference, means that there are constraints of what a given observer CAN know or CAN optimally infer; and this relates to the complexity and information capacity in my view. This means that it doesn't make sense to just dismiss it as ignorance in the ordinary sense, as this type of ignorance is due to how nature works. And this "ignorance" will influence the action of this observer. Because theories are also information and they neeed representation.

I know I'm comparing human level inference with physical inference between particles in the analogy here but I'm suggesting there is a deeper connection and the analogy is just to illustrate the principles. So in a sense there is a common framework, but it's not a theory like we know it. The framework is rather some constructing principles behind rational inference and action.

The distinction between "ignorance" or just observer dependence makes sense in the structural realism view. Different observers generally make difference inferences and thus end up with different information.

/Fredrik