I noticed recently that Nima Arkani-Hamed was giving a talk at Cornell, with the title Three Cheers For “Shut Up And Calculate!” In Fundamental Physics. No idea whether or not video is now or will become available.
From the abstract one can more or less guess what sort of argument he likely was making, and it’s one I’m mostly in agreement with. “Shut Up and Calculate!” is pretty much my unspoken reaction to almost everything I read purporting to be about foundational issues in quantum mechanics. I have in mind in particular discussions of the measurement problem, which often consist of endless natural language text where one struggles to figure out exactly what the author is claiming. An actual calculation showing what happens in a precise mathematical model of a “measurement” would be extremely helpful and likely make much clearer exactly what the problem is (or, sometimes, whether or not there even is a problem…). Such calculations are all too few in a huge literature.
Over the last few years, while teaching and writing a book about the mathematics of quantum mechanics, the tedious exercise of trying to get all signs right in calculations has sometimes turned out to be quite illuminating, with tracking down a mysterious inconsistency of minus sign leading me to realize that I wasn’t thinking correctly about what I was doing. I’m all too aware that this kind of calculational effort is something I too often avoid through laziness, in favor trying to see my way through a problem in some way that avoids calculation.
On the other hand, I’m not quite ready to sign up for “Three Cheers”, might just stick to “Two Cheers”. For a perfect example of what’s wrong with the “Shut Up and Calculate!” philosophy, one can take a look at the forthcoming Workshop on Data Science and String Theory planned for Northeastern in a month or so. They have a Goals and Vision statement which tells us that they plan to:
treat the landscape as what it clearly is: a big data problem. In fact, the data that arise in string theory may be some of the largest in science.
About being the “largest”, I think they’re right. The traditional number of 10500 string theory vacua has now been replaced by 10272,000 (and I think this is per geometry. With 10755 geometries the number should be 10272,755). It’s also the case that “big data” is now about the trendiest topic around, and surely there are lots of new calculational techniques available.
The problem with all this is pretty obvious: what if your “data set” is huge but meaningless, with nothing in it of any significance for the problem you are interested in (explaining the Standard Model)? This is not a new project, it’s an outgrowth of the String Vacuum Project, which I wrote about here, here and here. This started with a 2005 funding proposal, ended up getting funded by the NSF during 2010-2014. From the beginning there were obvious reasons this sort of calculational activity couldn’t lead to anything interesting, and as far as I can tell, nothing of any value came out of it.
For an opposite take to mine on all this, see the paper Big Numbers in String Theory, by Bert Schellekens. It contains an odd June 2017 preface explaining that it was supposed to be part of special issue of Advances in High Energy Physics devoted to “Big Data” in particle and string phenomenology (“all the ways we use high performance computing in addressing issues in high energy physics, and (in particular) the construction of databases of string vacua”). This issue was cancelled “as requested by the Guest Editors”. I wonder what the reason for this cancellation was, in particular whether it had anything to do with part of the topic of the special issue being considered by some to be obvious nonsense.
I think you’re right that string theory is based on an incorrect physical idea, but I don’t think that’s what Smolin is talking about. He’s pointing out that the physical idea isn’t even talked about anymore, and hasn’t been for a long time. It’s just a bunch of physicists spitting out meaningless calculations, many of which aren’t even properly done. Continuing to do that, which is what Arkani-Hamed is suggesting, won’t get anyone anywhere. And unfortunately I don’t think physicists (well, string theorists) think that the calculations they’re doing show they’re on the wrong path, precisely the opposite. Look at all the possibilities! It’s not that there isn’t interesting math, there is, but it’s not the physicists who are doing it, and it has no relation to the physical world. I should give credit to the physics folks for pointing out an interesting direction for mathematicians to look in, I think honestly much of that was Witten, who has remarkable mathematical insight, there’s a reason he won a Fields.
I don’t think it’s meaningful or helpful at this point to globally say “string theorists are doing X” and criticize that, there’s a huge range of things of different value being worked on by “string theorists”, most of which have nothing at all to do with the original physical idea of quantizing a 1-d extended “string”.
To me one of the most bizarre aspects of the current situation is the inversion of attitudes among prominent string theorists (note that Arkani-Hamed has never really been a “string theorist”, he comes from a different background) over the past decade. Back when our books came out in 2006, the criticism from string theorists of Lee and some other people doing LQG was basically that they were woolly-thinkers, going on about Einstein and how you just needed some vague revolutionary “physical” idea, while string theorists had a real theory and could do real calculations. Nowadays, many of the most influential people among “string theorists” (of those who haven’t completely abandoned the subject for field theory and condensed matter) are engaged in pursuit of a vague but impressive sounding “physical” idea that somehow a theory of quantum gravity can be found based on general ideas about entanglement and quantum mechanics. This looks to me like exactly the sort of thing Lee’s string theory critics had in mind a decade ago when they were criticizing him and the LQG people (those LQG people as far as I can tell are now mostly doing detailed calculations…)
One thing I’d love to know is who exactly Arkani-Hamed had in mind when he decided that he should give this talk. Who does he want to “Shut up”? I suspect Lee was one such person, but what does he think for instance of Susskind, who doesn’t appear to be calculating anything anymore? What does he make of “calculations” like “ER=EPR” or “GR=QM”?
For those interested in some recent thinking about quantum measurements, which includes calculations, I suggest having a look a What happens in a measurement? by Steven Weinberg, Phys. Rev. A 93, 032124 (2016).
Thanks for the Weinberg reference. Working along similar lines with the Lindblad master equation, Apoorva Patel and Parveen Kumar have produced what appears to me a spectacular new result, with realistic prospects for experimental tests:
In the style of Einstein’s comment on science and religion,
Calculations without physical idea is blind; ideas without calculations are lame.
I suppose the original physical idea behind strings is at fault here. Once the failure became apparent, around 1990, the solution from the string establishment was more calculations. There were further physical insights (like AdS, conformal symmetry, etc.). But these were tangential to the original unification program that gave rise to the popularity of strings in the first place, in the 1980s.
Perhaps a better way to say it is that a certain style of “math jock” theoretical physics, combined with the previous successes of symmetry-based unification, led to the feeling that just driving further in that direction — without careful study of the map — would soon lead to the “end of physics.” This path was never devoid of physical ideas. But string theory did not incorporate a new and deep breakthrough physical principle that could be turned into a coherent theory. Very different from gauge theory, chiral symmetry, spontaneous symmetry breaking, or the spawn of superconductivity and the Goldstone-Higgs complex. The latter ideas were “pretty,” but mainly, they were simple, general, and powerful. Granted a limited pool of assumptions, you can go very far with these — hallmarks of the best scientific ideas.
Searching my memory of graduate school in the 80s, I’m trying to remember what it was that entranced us about strings. (I never worked in strings directly, just around the periphery, back when theory included phenomenology, an essentially dead subject now.) The first memory I have is “anomalies cancel! anomalies cancel!” Within the following year or so, we learned that strings incorporated higher dimensions (like Kaluza-Klein), groups that extended unification groups we already knew (like E8 and SO(32)), and supersymmetry and supergravity. These were already well-known topics.
The funky string diagrams were fun. But what convinced us was the fact that strings seemed like an extension of a familiar road that we’d been traveling on for a decade+ anyway, just with some weird and unexpected twists. The fact that it did NOT propose a radically new physical idea seemed sensible and attractive.
re: the deep learning talk from last page, the difference between ‘deep learning hype’ and ‘big data hype’ is that *deep learning actually works*, at least sometimes. Yes, there are major unanswered questions about many aspects (such as interpretability), but all that means is there’s a big gulf between experimental results and theory, a situation that should not be too foreign to people reading this of all blogs.
You say that some mathematicians think they understand quantum mechanics because they (think they) understand Schroedinger’s equation. When I was an undergraduate 40 years ago, I knew that I did not understand quantum mechanics, because I knew that I did not understand Schroedinger’s equation. This has not changed: I still do not understand quantum mechanics, and I still do not understand Schroedinger’s equation, although I am perfectly able to shut up and do the calculations if I want to. What *has* changed for me in the last 40 years is that I now know that nobody else *really* understands Schroedinger’s equation either, although many people think they do, or at least say they do.
23 Oct 11.11pm. I agree. Physical ideas lead to math, you do the math, it doesn’t work, you revise the physical idea, and/or you revise the physics-math interface, and you go round again. And again. And again. Fifty years of ignoring the basic misconceptions and calculating forever in the wrong model is a complete waste of time. New physical ideas are two-a-penny, but mostly they are counterfeit, and the genuine ones are very rare and precious. Distinguishing the two is the hard part, and most are thrown away before being properly examined, so it is likely that there are some genuine ones out there that are not recognised as such. Mathematics is also cheap, but again, choosing the right mathematics is very, very difficult and expensive. It seems to me that the current strategy amounts to going to a car boot sale and picking up a bit of mathematics or any old junk in the hope that it will solve the problem. The chances are zilch. I have no solution to the problems, but I know where you have to start: get real mathematicians involved in the search. It is essential to have a genuine two-way dialogue between the physical ideas and the mathematics that expresses those ideas. But choosing the right mathematicians is also very very hard. The evidence of the past thirty years is you’ve consistently picked the wrong ones. Hahaha! “You’re very clever, young man, but it’s turtles all the way down!” Throw the dice, start again, pick some clever people at random, give them tenure *before* they’ve proved themselves. That’s the only way you’ll not only get the really innovative ideas you need, but also get someone committed and able to work on them. You’ll give tenure to 99 people who after 40 years come up with nothing, but you may also have hit on the one person who is worth 100000.
This is venture capitalism on the most extreme scale of risk, and it is necessary to approach it in that spirit. Build a team of 3 or 4, or 6 or 8, extremely clever and open-minded people, spanning particle physics, relativity, cosmology, mathematics, and leave them alone and give them everything they need for their entire careers, to argue among themselves until they resolve their differences. Treat them like a jury, and refuse to let them out until they come to a unanimous decision. Above all, do not interfere in any way in their deliberations, and do not give them a time limit. And maybe repeat the exercise a few times in parallel, up to the limit of the desired investment. Ah well, utopia is a beautiful place, today’s political reality is a very different kettle of fish.
Robert A. Wilson,
As someone who has done (or tried to do) research in particle theory for a long time, I am convinced that there are no fixes of the sort that you recommend.
If it were simply a matter of building the right sort of team or looking for specifically qualified people, it would have already worked.
There is no simple path to creative achievement. Creative people have to find the way themselves. Good ideas are a matter of luck, talent and knowledge in that order. In my limited experience, planning is an obstacle, not an incentive!
You are right, of course, I was merely venting my frustration with the current system, not really proposing an alternative. “Luck, talent and knowledge in that order”, definitely in that order. You can’t control luck, and you can only vaguely spot talent, so the system is based largely on assessing knowledge. And that is where the systematic biases cause the most damage, because you cannot know in advance what combination of knowledge is going to trigger the key insight that leads to the big breakthrough. You need to allow for input from unexpected directions, and I admit I don’t know how to do that. But you are surely right that less planning rather than more, is the way to go.