Why String Theory Is Both a Dream and a Nightmare

Yes. The primary issue with string theory I have is that it seemed for a long time to suck in large numbers of very talented theoretical physicists because it was a fashionable field - I saw many contemporaries of mine who went into theoretical physics go down this route.

It's not just experimental physicists and the big budget costs that comes with particle accelerators. Theoretical physicists of course cost a lot less than the LHC but they are quite a scarce resource nonetheless - especially very talented ones - and I wonder what could have been achieved had many of them been working in different areas of physics or even in different fields altogether.

Which experiments have turned up nothing?

Do you mean SUSY not showing up at the LHC? Do you think physicists can materialize particles that do not exist in Nature? Finding that the mechanism of electroweak symmetry breaking happens to be the Higgs, discovering it does exist and measuring its properties is small potatoes? Getting the actual data to discard the alternatives is worthless?

All this is very confusing. You do have a theory about what's going on in HEP and pretty much how an experimental science such as physics works. Unfortunately it doesn't match what people do in HEP nor why they do it, so we've come full circle apparently.

BTW, any first year graduate student can answer those questions. It takes about 5 years if we're starting with Physics 101.

> Would it make sense to scrap the planned particle accelerator, and instead give $1 million to every person who actually understands the Standard Model and current research to devote their lives to something else?

No. That's just the talk of some "theorists." In reality, before any new discovery is made, there have to be some useful measurements that go beyond previous limits. To learn anything fundamentally new, we have to have new measurements. It's as simple as that.

There is a whole bunch of people/interest groups that try to sell the public that the basic research (as in experiments) is not needed. Unfortunately, they get a kind of support from some theorists that raise doubt in most of big investments in experiments (including particle experiments). But such arguments are very biased, when not plainly deceiving.

Whatever the limits to our measurements are now, they are at the same time the limit of what we can know. Investing more in pushing the boundaries is always something worth doing, and that includes doing further experiments with newer accelerators.

Moreover, at the time the technologies are first time developed, nobody can promise you that they will be used for the benefit of people. But they are used exactly so, often using them in totally unexpected way.

Take as an example MR scanners -- the complex super strong magnets, in the style as they were earlier used in the accelerators, actually change the orientation of the molecules in your body, and we can measure all that and actually diagnose the diseases.

Or take PET scanners: the substances given to the patients before these scans are actually produced in the cyclotrons. Do you think first cyclotron (also physics) researches could even dream that such devices will be used to help diagnose cancers?

http://www.bccancer.bc.ca/our-services/services/pet-function...

I can go on and on.

No theory can replace experiments. We need all the experiments we can do, and all that could go beyond the current limits.

Want the example from computer-related areas? All the modern achievements in AI came from the possibility to process more data, much faster. It was again pushing the limits, and that was simply necessary to move anything. Correpsonding software development under the limits of, for example, eighties technologies would not bring us to that what we have now. It's always about achieving the new limits. Theory for the sake of theory, without the experiments and practical implementation simply can't make the breakthroughs achievable by reaching new limits.

A better word is "unfalsifiable".

String theory predicts numbers other spatial dimensions. (at least 11) You can test for these. However, these extra dimensions can be arbitrarily small. You can prove that no extra dimensions greater than 1e-30m (or whatever) exist. This does not disprove string theory. Well, now you have to test for smaller extra dimensions. Let's say you prove no dimensions larger than 1e-40m exist. Still haven't disproven string theory: they could be smaller. So on and so forth.

These extra dimensions have to curl up. The way in which these extra dimensions curl up constrain the masses and properties of the various particles. There are a stupidly large number of different ways these extra dimensions can curl up. Something like 10^500. So there's no way to measure the various properties of physics and the universe in a way that falsifies string theory. Regardless of what experiments show, there are still going to be an immense number of curlings that are consistent with the measured reality.

IMHO string theory is still science, but it's depressing, fruitless, dystopian science. It lacks... hope.

Regarding (1), at this point I believe that String theory is still incomplete. It can make some broad predictions (parts of which might be more or less testable, but most will be in the Planck-energy scale), but so far it's proven difficult to fill in all the gaps and get something equivalent to the Standard Model. Basically you've got a few too many degrees of freedom and most of the predicted behaviour is well beyond the energy scale that we can reach experimentally.

> People say it's untestable so it's not science

In that negative viewpoint it's still science, just another branch: Philosophy (Metaphysics) and still Mathematics of course ;-)

No expert in String Theory, but the bottom line is that the strings are really small. (~ Planck Length) So directly measuring anything is not going to work.

It's more like 1), but known predictions assumed by Standard Model are baked into the parameters.

"rest of its predictions match up with all tested physics" is just curve fitting.

i.e. what's the difference between #1 and #2? the only predictions that matter is where it differs.

The way I'd put it is

3) In its current state, you can get string theory to "predict" just about anything.

The reason is the enormous amount of string theory vacua, each giving rise to different "laws" of low energy physics. You've probably seen the old estimate that 10^500 such vacua exist. That turned out to be a slight underestimate:

https://arxiv.org/abs/1511.03209

Even if you try to narrow down your search to vacua with the same rough features as the Standard Model (types of particles and interactions) you end up with numbers like 10^723:

https://arxiv.org/abs/1810.00444

(Disclaimer: what follows is a caricature meant to illustrate the magnitude of the problem, not a rigorous argument.)

So in principle you can imagine performing lots of precision experiments to measure a bunch of quantities which can be derived from string theory, and then run through all those 10^723 vacua to see which ones match your data.

It will take a while.

Can you derive precise parameters of low energy physics from string theory vacua at a rate of one per second? The number of seconds since the big bang is less than 10^18. You do the math.

And even if each quantity you measure can discriminate between two halves of the remaining search space, narrowing those 10^723 alternatives down to 1 will require more than 2400 such mutually independent quantities to be measured (10^723/2^2401 ~ 1).

Alas, once you fix the rough features of the Standard Model, it has only about 25 free parameters.

So, even given all the experimental data you need and eons to run your search, you will not be able to nail down one string theory vacuum as the one we are in (assuming that string theory is correct), if all you know about experimentally is the Standard Model. That only reduces the uncertainty about which string theory vacuum we are in to 10^723/2^25 ~ 10^715 different alternatives. You need to discover new physics at higher energy, to bring in more free parameters to use in your search. Something like 2376 (!) new free parameters.

Only then will you be in a position to finally say that future discoveries in particle physics will be able to decisively rule out string theory (or not). Until then, whatever comes out of an experiment will just tell you that some vacua are out, but you will have plenty left.

Doesn't that make what is or is not science somewhat up to chance?

Suppose you develop a theory in order to explain the known results of experiments K1, K2, and K3. You have tuned any tunable parameters to make sure you theory is right for K1, K2, and K3.

Then you calculate what results you would expect from your theory in other situations, S1 and S2. S1 and S2 were not taken into consideration when you developed the theory.

It seems strange to me for whether or not your theory is "science" to depend on whether or not someone happens to have already looked at S1 and S2.

Suppose on Earth the correct results for S1 and S2 were already known when you developed your theory. Then we discover there is intelligent life on Mars and their scientists have developed the same theory at about the same time, but no one had ever looked into S1 and S2 there until after the theory called attention to them.

So the theory is not science on Earth but it is science on Mars? Does it cease to be science on Mars when we inform them that we knew the correct results for S1 and S2 before anyone on Earth or Mars developed the theory?

If someone had come up with General Relativity in 1680, a few years before Newton's Philosophiæ Naturalis Principia Mathematica was published in 1687, would Newtonian gravity have not been science? (I'm assuming that at the time the small differences between Newtonian gravity and GR would be below what they could detect experimentally).

Archive version https://archive.is/HQIXO

fwiw only paywalled if you're not willing to turn off adblocker for Forbes