Why I am REALLY disappointed about string theory

[marcus]

Why is it so important for the features of string theory (susy, extra dim) to be realized independently in nature? Maybe they could just be internal machinery of the theory? Quantum mechanics uses Hilbert spaces. Have any of you guys ever seen a Hilbert space? Or maybe you could argue that string theory is wrong since it uses the identity 1+2+3+...=-/12, which we all know is false. Leave the possibly internal stuff out of our universe.

If string theory can compute things correctly, that's all we need. If it uses susy, apples, or sand, why does it matter if we don't see those things as we would naively expect? The only question which we need to ask is : can it be used for anything?. And the answer seems yes.
...

can it be used for anything? is a reasonable question to ask about whatever line of mathematics. And sometimes it's desirable to push ahead even if there is no positive certainty.

 

[ccdantas]

Why is it so important for the features of string theory (susy, extra dim) to be realized independently in nature?

You mean then that those features are not necessarily physical, and hence their eventual non-observation has no impact whatsoever on corroborating or not the theory? Then what is your criteria for considering a feature physical within a theory, or in other words, what would make it "important" or not (to the point where one could use it to falsify the theory)? Or would it be acceptable that everything would just be internal maths that magically gives a correct observable output?

If string theory can compute things correctly, that's all we need.

So again, what exactly do you want to calculate? Correctly to what precision (in order that you find that a given theory is acceptable)? Suppose someone gives you a black box that you cannot see inside what it calculates, and you see the output of the box which matches an observable up to a given precision. Would you be happy with that and finish your business? Then suppose that you are allowed afterwards to open the box and see that the internal calculations use some concept that have been falsified, or that something ad hoc was put in by hand, but nevertheless gives good results at some level. Would you still be happy with that and finish your business?

 

[tom.stoer]

String theory is crowding out other promising research programs, i.e LQG, and may not be physically correct (i.e 4D, non-SUSY)

That COULD be right, but it's definately NOT the discussion I wanted to start in this thread. Sorry.

 

[tom.stoer]

Quantum mechanics uses Hilbert spaces. Have any of you guys ever seen a Hilbert space? ...

If string theory can compute things correctly, that's all we need. If it uses susy, apples, or sand, why does it matter if we don't see those things as we would naively expect?

I used the same example with Hilbert spaces in quantum mechanics. The difference is that with qm certain things in nature became calculable the first time - this is not the case with strings. And as I said there is another difference: SUSY is directly visible in the physical spectrum - but we do not see SUSY to be realized in nature.

 

[tom.stoer]

@ccdantas: there are a couple of things which should be calculated from a fundamental theory; I listed some of them is post #1.

And again I repeat my questions from post #1

• What are the major achievements of string theory?
• Are there predictions subject to (accessable to) experimental verification / falsification both in principle and in practice? Are there physical phenoma which (once observed) would kill string theory?
• Are there predictions specific for the string theory context (nothing that may follow from SUSY as SUSY could be true even w/o string theory)
• What are the short-term / long-term research programs?
• What are the major obstacles inherent to string theory preventing the theory from delivering on its promises?
• What will be the final theory in terms of strings - a theory, or a framework to create theories?

Sorry for insisting on that. The idea I had in mind when starting this discussion was to let string theorists tell us more about their theory, their specific achievements and issues - instead of always explaining them what we (outsiders) think about it.

 

[ccdantas]

@ccdantas: there are a couple of things which should be calculated from a fundamental theory; I listed some of them is post #1.

Yes, thanks, I know. I was just questioning negru in order that he would clarify his rationale.

But since I am not a string theorist and do not wish to contribute diverging from your interesting thread, I will just be following with no further comments.

Thanks.

 

[ccdantas]

Just one thing: I wonder who are really professional string theorists known to contribute at PF??

 

[tom.stoer]

Yes, thanks, I know. I was just questioning negru in order that he would clarify his rationale.

But since I am not a string theorist and do not wish to contribute diverging from your interesting thread, I will just be following with no further comments.

Thanks.

@ccdantas! This was not to forbid you to speak - sorry for that!

 

[tom.stoer]

Just one thing: I wonder who are really professional string theorists known to contribute at PF??

Very good question. Everybody is welcome. I would also invite members who are in contact with profession string theorists to contribute.

 

[ensabah6]

Very good question. Everybody is welcome. I would also invite members who are in contact with profession string theorists to contribute.

There's Lubos Motl

 

[ccdantas]

@ccdantas! This was not to forbid you to speak - sorry for that!

There is no misunderstanding It's just that it's really more appropriate to read how professional string theorists will address your questions than I write/question anything general for the moment. My concern with negru's comments is somewhat outside this thread.

 

[tom.stoer]

There's Lubos Motl

 

[marcus]

I got interrupted while I was writing this, and had to be away. I want to continue from here, and try to make some points related to what Negru said:

can it be used for anything? is a reasonable question to ask about whatever line of mathematics. And sometimes it's desirable to push ahead even if there is no positive certainty.

But there seems to be a lot of free-floating defensiveness. I'd like to understand that better. Who is supposed to be the enemy?

One has to distinguish between criticisms of the mathematics itself, and criticisms of the program (direction, emphasis...)

The most trenchant criticisms I can remember from recent times were from Nima Arkani-Hamed (November 2008) and from Murray Gell-Mann (I will try to find the links).

Gell-Mann was talking about the direction of the program (avoiding hard fundamental questions of principle in favor of increasing elaboration) and Nima was talking about what he suspects are mathematical limitations (not to expect it to say anything new about high energy physics, but maybe about gravity). I was surprised, a bit shocked, by both statements.

But we are told repeatedly about imagined bogeymen. "Armchair experts" who apparently were calling for a complete halt to string research 10 to 15 years ago!

I do see changes going on within the string research community (shifts in the makeup of new publications, the annual conference etc., the actual research focus of those traditionally considered top people, citation patterns...)
Surely transparency is a good thing and these trends should be reported.
I'm not sure that these changes should be considered problems or troubles. No matter what happens there will still be thousands and thousands of string theorists, many unable or disinclined to do any other kind of research.

If string is having trouble, it is surely not due to popular books by "Smolin and Co."
One should try to be serious. It is silly for real and interesting shifts going on in research to be blamed on "Smolin and Co." And it only deflects people's attention. A kind of noise--like banging on pots and pans. Maybe it allays some people's anxiety to focus their attention on an exaggerated image, but it doesn't make the real situation go away.

Personally I wouldn't label the changes going on in the stringy world as "trouble". I don't consider them a problem, just very interesting---something to observe and try to understand.

And it is way way overly dramatic to talk about "death" or "kill". Those words have been used in this thread. The "armchair experts" imagined as relentless, uncomprehending enemies, are supposed to desire the death or rejoice in the killing of string.
At most all we are talking about sociologically is a small percentage adjustment in the departmental pecking-order. A tiny adjustment in the prestige and self-importance index of a few academics. With thousands of theorists still on board and working world-wide.

 

[tom.stoer]

OK, sorry for using the word "kill". Let me explain: you can "kill" a theory in many different ways: cutting budgets, malicious gossip, ..., falsification. I would like to stress that I was always talking about falsification, nothing else. Sorry again.

 

[marcus]

Tom I understood the word when you used it! You simply meant empirical falsification, ruling out.
It was other people I thought were responding hysterically---with even a bit of paranoia.

My take on the empirical test issue is that we are looking at a large and versatile body of mathematics.
Freeman Dyson had a wonderful perspective piece on string where he talks about birds and frogs.
Anyway a great, admirable, complex, manyshaped, myriad-minded body of mathematics that has grown
(suddenly but in a sense naturally) out of the differential geometry and algebraic topology I learned about in grad school
back in the day.

You do not falsify that kind of thing. It is a self-supporting form of human creativity. Possibly even an addiction.

Maybe I'm wrong or my attitude is somehow inappropriate---if so I'm more than happy to retract what I just said.
And you may be right in asking for a unified coherent testable theory at some specified scale of interaction.

But what I am suggesting (at least right at this moment) is that we aren't dealing with a physics theory---something based on enunciated general principles---with a central main equation or two---that predicts new phenomena and you can compare with various critical future experiments.

What we are confronting is, instead, a vast mathematical grab-bag, which seems rich and applicable in several quite different areas of physics.
It somehow doesn't seem fair to ask it to be falsifiable.

And then there is the separate issue of a possibly dangerous fairytale: the anthropic multiverse. Essentially Susskind's 2003 reaction to the January 2003 KKLT paper.
Since that was excluded from Strings 2008 and played only a small role in subsequent Strings conferences, I am hopeful that Landscapism is now mainly for public consumption and that the community itself has avoided that route. String mathematics does not need Anthropics, it can flourish quite well without that IMHO. Anyone is of course welcome to correct me if I am wrong about that.

 

[MTd2]

I blame it on Marcus & Co.

 

[tom.stoer]

That's not the point: I had something in mind when I started this thread. I knew that it would become difficult. Therefore using "killing" was counter-productive, unrewarded and wrong.

 

[marcus]

I blame it on Marcus & Co.

Marcus & Co. loves and admires string. We have never criticized string mathematics. Why should we?
Personally I just report the news

But MTd2, let's not talk about each other! Tom has started a great thread. I am really interested in what people like Negru have to say that are actually embroiled in the string business. Maybe like Christine I will try to stay in the background and listen to others' opinions.

 

[marcus]

I had something in mind when I started this thread...

I will be sorry if my butting in somehow contributed to stalling the thread.
The initial series of exchanges between you and Negru was refreshing and constructive. Negru is a good spokesman: forthright honest un-defensive comparatively uncomplaining. I'm hoping to hear that conversation continue in some form, with the same or different parties, sometime soon.

There was also a mot juste metaphor about shampoo and justice,
a few words of which I copied and pinned to the wall of my computer space.

Of course I am prepared for answers regarding landscapes (Susskind) and mathematical universes (Tegmark). But frankly: I will never accept these arguments. This is regarding string theory, therefore I expect answers in the context of string theory (if my daughter has to go upstairs in order to shampoo I don't accept discussions regarding justice; that does not mean that I am unintersted in justice - I am - but not in the context of telling a six-year old girl to go upstairs in order to shampoo!)

 

[tom.stoer]

Hey marcus!

You like it? Good on you - and me :-)

Back to strings and to my intention. We had a couple of discussions here (and I was involved) where we discusses pros and cons of string theory. Unfortunately most threads ended with a variation of "string theory is unphysical" and "why can't you see that string theoriy is a great theory".

My impression was that were are (at least partially) unfair. We judged string theory from the perspective of completed or "working" theories although we should know that it isn't. (This become clear in this discussion a couple of times). So we should be fair and try to apply different standards. Nobody here knows how it felt 100 years ago when great physicists where looking for a solution of the quantum puzzle; maybe similar; one knew that there was something out there, but what exactly nobody could say.

So my conclsuion was that we (skeptics) should accept that string theory is work in progress and should listen to the argments of the string community carefully (we are no Woit's here :-).

But on the other hand the string community should accept some standards as well (at least for THIS discussion). Namely that one should not constantly change the rules of the game. That's why I was insisting on not talking about new standards in science, landscape and things like that. To be honest: it can very well be that string theory stays "unpredictive" in this sense; but (once proven) to overcome this situation is not a matter of string theory itself. It is really a matter of physics (as a whole). If string theory predicts that it cannot make predictions except for the prediction of the landscape, then we need a different and a larger context to deal with it. I do not say that I am not interested in this discussion, but not in the context of string theory - seen as an ordinary but still developping physical theory. Especially as I learn that string theory is work in progress and it may therefore very well be that there are clever ideas beyond the landscape to be discovered ... So I try to limit (focus) the context of this discussion - nothing else (and so far it worked, that's why perhaps it may have been better not to write this paragraph :-)

My conclusion was that it should be string theorists themsevles to assess string theory according to its value and its inherent problems. This is what I am still asking for.

 

[mitchell porter]

How I think about the string landscape:

First we need to find some vacua that do give us the particle-physics standard model and cosmological standard model (perhaps with some extra features). Remember that even that much has not been done! There are vacua which look like the MSSM, but no-one seems to have exhibited a construction providing the exact masses of the observed particles. It's all still work in progress.

Then, we need to embed this in a cosmological solution to string theory. I don't hold much hope for the idea of a unique, dynamically determined vacuum. The viewpoint of eternal inflation, with different regions of space exhibiting different particle spectra etc., seems like a natural cosmology for a theory with many different vacua. What's important is to show that this really is the natural cosmology of string theory.

In this regard, Susskind's original landscape paper (hep-th/0302219, pages 4-5) has a simple and vivid description of a string cosmos containing regions in different vacuum states. It's 11-dimensional M-theory, compactified on a 7-torus, leaving 3+1 large dimensions. The fundamental 5-brane of M-theory couples to a magnetic 7-form field. If you consider these branes and fields wrapped on 3-dimensional subspaces of the 7-torus, you are left with a 2-dimensional object coupling to a 4-form field in the 3+1 large dimensions. These 2-dimensional objects (partly compactified M5-branes) are then domain walls interpolating between regions of 3-dimensional space where the 4-forms take different values.

I presume the cosmological reality is more complicated than that, but it offers a starting point for thinking about the big picture.

 

[tom.stoer]

I agree that one has to come much closer to the xMSSM; but in parallel one has to find xMSSM at the LHC, otherwise the theoretical progress in string theory has no value.

 

[humanino]

I agree that one has to come much closer to the xMSSM; but in parallel one has to find xMSSM at the LHC, otherwise the theoretical progress in string theory has no value.

What happened to the misaligned superstring ?

 

[friend]

Isn't string theory about how strings and membranes vibrate on an pre-existing spacetime background which is just assumed? If so, then string theory is not background independent. Is this the reason for the landscape problem?

 

[tom.stoer]

Isn't string theory about how strings and membranes vibrate on an pre-existing spacetime background which is just assumed? If so, then string theory is not background independent. Is this the reason for the landscape problem?

I was thinking about that - and I hesitated to include the background-dependency into my list of questions and open issues. We had this discussion a cpouple of times and we came to the conclusion that string theory may be background independent - but not in the sense as excpected in ART.

I have a different opinion on the landscape: I guess that a completely different structure will emergy once one is able to perform full non-perturbative calculations. But this is implicitly assumed when a talk about a final set of definitions of the theory - which is currently missing.

 

[BenTheMan]

tom, et al.---

Expect a longer response to your questions in the near future, but let me make a point quickly that most people don't acknowledge.

The MSSM has a landscape problem.

Let me clarify: when I say ``landscape problem'', I'm not referring to cc problems (which no body can solve). I mean a much more mundane issue, involving scalar VEVs. I claim that any theory in which couplings and masses are set dynamically will exhibit a landscape of solutions, if not at tree level, then certainly when one considers higher dimensional operators. Let me explain.

One of the ways to understand the landscape problem in string theory is to note that string theory is a very good way to get SUSY QCD theories, which have a consistent UV completion. But any SUSY QCD theory with chiral multiplets (i.e., fermions that aren't gauginos) has a landscape problem: the reason is because one can always write down a set of supersymmetric solutions to any theory, called F=0 and D=0 constraints. In a broad class of theories, when F=0 and D=0 solutions exist, they're not unique. Thus there are a continuum of possibilities that comes just from the presence of supersymmetry.

Now, let's assume that the supersymmetry isn't broken in our universe. What does the landscape look like? Well, we have all these scalars in our theory, called sfermions (squarks and sleptons) which obey some F=0 and D=0 constraints. These F=0 and D=0 constraints can be satisfied at points where the squarks get VEVs. This means that SU(3) is broken, which is ruled out by observation (to say the least).

Thus we live in a very particular vacuum of nature: that is, all scalar VEVs, with exception of the higgs, are zero. The fact that other vacua exist where the SM gauge group is broken is a fact that follows from field theory, and NOT string theory.

You can extend this argument to any theory with scalars. This is a particle physics statement, and has nothing to do with quantum gravity.

To emphasize this point, consider the Higgs. The higgs mass is the only dimensionful parameter in the SM, and it sets the scale for electroweak symmetry breaking, and all of the fermion masses. The higgs potential in the SM is ad hoc, and has no known origin. In fact, if we write down some higher dimensional operators, as would arise from some Wilsonian effective theory, it's not hard to imagine that we might find other minima for the higgs. So in the full theory (some GUT, say, or---at worst---some non-stringy QG model) the higgs potential can have many minima. So it's even possible that the SM has a landscape problem.

The point is, the string theory landscape is nothing new, and was even realized a long time ago. The only reason it's a problem now is that some people are jealous that their pet theories are being ignored by serious scientists. You know---it's always easier to sh1t on someone else's work than to convince people that your ideas are worth listening to, or even come up with your own ideas in the first place. And make no mistake---QG researchers are actively involved in the former, and Columbia University professors who don't have their own research program actively engage in the latter.

It is true that string theory was overhyped. But this is the tendency of the media to sensationalize results. Remember the hoopla involved with Garrett Lisi's E8 paper a few years ago? Discover Magazine had a list of ``10 Successors to Einstein'', and this paper landed Lisi on that list. Good for him, but you can ask him yourself if he believes he should be there. Where were all the stories pointing out the objections of the scientific community of his work? Did you see any rebuttals? I didn't. Journalists, by and large, are idiots. (If you need proof of this, read this article, published in a ``serious'' pop science mag.) They want headlines, and physicists want money. So it's an easy tradeoff to make---we get publicity, and they get a story. Anyway this is sociology and not physics, so it's not really worth talking about.

Let me say that I hope (as much as anyone) that a unique solution exists. But I also understand why the solutions may not be unique---I have spent my (short, and rapidly winding down) career as a string phenomenologist, and I can see the types of problems people tend to have in ALL string constructions.

Let me finally reiterate and old argument: Citing the landscape as a problem may just come from the fact that people are expecting too much from string theory. Who are WE to expect that Nature is unique? Does this bother you? At some point, people tried to derive why we have 8 planets around the sun, or why Earth was some particular distance from the sun. But there are trillions of suns and Earth's, so is it worth while to try and understand why THIS earth-sun system happens to be the right age, and have the right separation for liquid water and a comfortable temperature? If you are a creationist, then you might marvel at this fact and spend time thinking about it---that is, if you believe that there's some reason for it all, then you will waste time on this problem. But if you understand that anthropics is the reason that humans live on earth, you ask different questions altogether.

 

[friend]

I was thinking about that - and I hesitated to include the background-dependency into my list of questions and open issues. We had this discussion a cpouple of times and we came to the conclusion that string theory may be background independent - but not in the sense as excpected in ART.

So string theorists believe that there may be only one background consistent with the physical constants we observe?

 

[negru]

Had a longer answer eaten..so I'll summarize some points.

Like marcus said, string theory at the moment is huge. It has quite a few applications ( f theory, ads/cft, black holes, etc). I personally find it unlikely that you can just expect to compactify some 10D theory and get the SM, then use ads/cft to compute some qcd stuff, use twistor strings to compute tree level qcd (topological strings are just as much strings as M theory is), etc and never make a deeper connection between these things. How could the real world be just a holographic projection of higher dimensional string theory, AND at the same time actually made of tiny strings at high enough energy? I mean, if gluons are actually strings, via ads/cft, you get another string/string holographic duality or what?

As has been pointed out, we don't have a working definition of string theory (one which explains why these things work). How could it predict anything if we don't know what it is?

That's why questions like "is string theory falsifiable" or "if there is no susy does it kill string theory" or "did it predict anything yet" etc always irritate string people. Not because they're inherently silly (these are the questions I was asking myself when deciding what path to choose), but they're just not well posed. We don't know enough string theory to ask the right questions.

And if you'd like to get rid of string theory, you'd have to separately kill all of these seemingly unrelated applications, since they use different assumptions. Also, about susy/high D being "physical". Physical to me means only something we can measure. Internal machinery, like hilbert spaces, we can never measure. However this doesn't stop them from making predictions. So for example if we go to Planck energy scales, don't find any strings, this won't mean ads/cft is wrong.

 

[MTd2]

You know---it's always easier to sh1t on someone else's work than to convince people that your ideas are worth listening to, or even come up with your own ideas in the first place. And make no mistake---QG researchers are actively involved in the former, and Columbia University professors who don't have their own research program actively engage in the latter.

You mentioned 3 types of people but you just qualified 2 of them.

 

[BenTheMan]

You mentioned 3 types of people but you just qualified 2 of them.

I am glad to see that you really understood the content of what I was talking about. You know---I come to this forum so that I can occasionally put in my two cents about things that I am interested in, or have thought about. It is truly rare that I meet someone who really gets me, at such a deep level.

Are you single?

 

[MTd2]

Hmm. Well, today I am pretty tired, so I am really slow, so I had to ask. But I am detecting some irony now... But no, I am not single.

 

[suprised]

So string theorists believe that there may be only one background consistent with the physical constants we observe?

Since this is thread is about getting out of control as well, I still need to read through all the answers and elaborate on some answers to Tom. But right now I don't have the time, though this question stroke me as separately commentable.

One thing that never comes out clearly in such discussions is that there is not "the" string physicists as a block with one fixed opinion. In fact there is a lot if arguments going on within the community, and it is split especially on this one. So I can answer only for myself; and my opinion is very clear on that. Namely it is quite inconceivable that the extremeley narrow range of parameters that allows us to exist, fits precisely to the only consistent background (of any theory). This is a bit like believing that the DNA molecule that defines us, is the unique solution of some fundamental theory, rather than being just a particular solution to a unique theory (electromagnetism in this example).

Edit: I just noticed that I might have misread the question. Well be it as it is.

 

[tom.stoer]

As has been pointed out, we don't have a working definition of string theory ... How could it predict anything if we don't know what it is?

That seems to be the conclusion of the discussion. So demanding that the theory shall produce physically falsifiable predictions shall be postponed as long as there is no sound definition.

That's why questions like "is string theory falsifiable" or "if there is no susy does it kill string theory" or "did it predict anything yet" etc always irritate string people. Not because they're inherently silly ... but they're just not well posed. We don't know enough string theory to ask the right questions.

I understand that these questions can irritate people; but we should neither blame the questioner nor the question but the theory. Look: I tried to ask physical questions, independent of any specific technical detail. I did not ask for details regarding compactification on a specific M⁷ in M-theory. I did not ask regarding mathematical proofs for certain dualities. I did not ask regarding a non-perturbative definiton. I did not ask for the n-loop measure for the superstring amplitude.

I asked purely physical questions which are to a large extend independend from these internal technical details. Whether SUSY exists (unkroken at a certain energy) can be decided experimentally, so we expect the theory to be able to make reasonable predictions. Whether a certain particle has a specific mass can be determined experimentally, so again the theory shall be able to make such predictions; etc. etc.

These questions may be too complicated; it may be too early to ask these questions; maybe string theory is not (not yet, never, ...) able to answer these questions (uniquely), but they are certainly not "not well posed".

In QCD it was (over decades - and still is?) a riddle what exactly causes confinement and how one can determine the spin structure of the nucleon. In condensed matter physics as of today nobody is able to tell us the detailed mechanics responsible for high-temperature superconductivity. But these questions are certainly well-posed physically. They are just hard to answer.

 

[tom.stoer]

Again I would like to come back to my (reduced) list of questions

• Are there predictions subject to (accessable to) experimental verification / falsification both in principle and in practice? Are there physical phenoma which (once observed) would kill string theory? My (preliminary) answer based on our discussion is "not yet".
• Are there predictions specific for the string theory context (nothing that may follow from SUSY as SUSY could be true even w/o string theory) Again my (preliminary) answer based on our discussion is "not yet".
• What are the short-term / long-term research programs? Not discussed so far.
• What are the major obstacles inherent to string theory preventing the theory from delivering on its promises? Besides some technical details (omitted fortunately) the main issues seem to be that string theory is still in an early stage of development, that a unique framework has not yet been established, and that therefore physical predictions are still out of reach
• What will be the final theory in terms of strings - a theory, or a framework to create theories? Not yet discussed, but my (preliminary) answer based on our discussion is "more a framework than a single theory".

 

[negru]

I think I agree with suprised on this one. It seems implausible that the "theory of everything" will have a unique solution. But that doesn't mean that no predictions will ever be possible. For example, maybe we'll have a two parameter solution space. If fixing these two numbers leads to different values of the SM constants, the CC, etc., we would still technically have a landscape with anthropic selection, but it would be an impressive achievement nevertheless. So I don't see what the big fuss is about the landscape issue. We could still find some principle able to reduce the solution space enough.


About the specific list of questions.
My answer would be "too early to know" for most of them. To the first two, you should include the ads/cft part as well. If you can compute something on the cft side and it doesn't match, you've just falsified ads/cft. Also, there are objects on the cft side which are only explained by strings (some of them even by classical strings, so nothing fancy at all). This is as specific as you can get.

Short/long term goals? Well there is extending ads/cft, exploring the landscape issue, finding string solutions in various backgrounds, understanding what the M in M theory stands for... All open problems are being worked on as we speak. A really short term (and doable, yet indirect) goal in my oppinon should be just understanding N=4 SYM. There is just so much weird stuff there, with the dualities, the twistors, the wilson loops (stuff that eg arkani-hamed, maldacena and bern have been working on recently). All of this will mean something for string theory in the end.