Friday, 22 November 2013

Do not work in Quantum Foundations under any circumstances. Or do so, what the hell.


Hi rats,

Very often, people come and tell me*: “Rat, you are so magnificent! Here you are, a former magician-actor-tap dancer-model-ninja-vampire-master of the universe-Madonna. How come that someone with your obvious talents and –ehem!- sexy muscular body is wasting his time in foundational research? (blink, blink)”.

*Approximate reconstruction.

The answer is complicated. I think I’m not surprising anyone when I say that nowadays Foundations of Physics raises the same expectations as a new star trek movie. For most, Foundations is synonym of mediocre results, low-level mathematics and endless pedantic discussions. And it is true: most works in Foundations (and even whole conferences!) are just like that. One has to dig very, very deep to find that precious gem that makes everything worth it.

In this post, I will try to summarize the good and bad aspects of the field. That way, independently of what you choose to make of your scientific careers, you'll know what you can expect, or what you'll be missing.

Let's start with the cons: working in Foundations sucks when…

1) … someone proposes a lame semi-classical model for photon polarization that actually reduces to the definition of quantum separability when one tries to make physical sense of it. However, since the author holds a Nobel Prize for completely unrelated research, the “discovery” soon becomes a popular topic in Foundations that no amount of logical arguments can kill. Bravo!

A different take on this story, advocated in this note, is that the photon model may be, in fact, scientifically sound. When the laureate defines a model inconsistent with the notion that post-selection is a type of preparation, he’s not making a gross mistake: he’s proving his creative genius by "freeing himself" from this traditionally held assumption*. And when the laureate violently bumps his head against the floor, he’s not stumbling and falling: he’s estimating the density of concrete in public pavements.

*Indeed, how could I be so blind!? Why didn’t I consider the possibility that, after measuring a photon, the universe disappears, or all other photons turn into Toblerone bars?

2) … for the third time, the John Stewart Bell Prize, which is awarded “for significant contributions first published in the [last] 6 years” and “is not intended as a "lifetime achievement" award” (check the rules), goes to senior group leaders.

Dear committee members: Adán Cabello will renege on contextuality before you award the prize to a mere postdoc, so stop giving false hopes to junior researchers. Be honest, remove the six-year requirement from the description of the Prize and give it to Tsirelson. God knows that, if someone deserves the John Stewart Bell Prize, that is Boris Tsirelson, the man who invented quantum Bell inequalities. Like Bell, he had a deep vision that translated into breakthrough results. And, like Bell, his work was largely ignored by the Physics community at the time.

Rats, I say we owe Boris Tsirelson big time: a prize for Tsirelson, now!!

The John Stewart Bell Prize committee, deciding the fate of a postdoc nominee.

3) … people from serious* fields advance a foundational topic by an epsilon (OK, two epsilons), get their results published in Nature and everybody wets their pants. Meanwhile, all other relevant contributions rot.

*Here by “serious” I mean “socially acclaimed”. If you believe that quantum computer science is objectively serious, stop random pedestrians on the street and try to explain them why quantum complexity classes are much more important than, say, epistemic models. If they try to escape, hit them on the head with your gun**.

**You don’t have a gun!? Then, how do you get people to cite your work at introductions and review papers?

4) … two authors publish three times essentially the same result (and I'm not counting the review!).

Why stop there? From this blog I want to propose Colbeck and Renner new ideas to spread their message:
  • “Quantum mechanics is complete”, the coloring book (it’s already colored), and “Quantum mechanics is complete”, the jigsaw puzzle (one piece may be missing).
  • “nIv'e', yu'egh Qap, tugh Qo' (the wave function exists, soon you won’t)”, the Klingon Opera.
  • “Who moved my local part?”, the best-selling book that has helped millions find their true nonlocal selves and now it can help you, too!
  • “The Texas chain Bell inequality”, the independent motion picture directed by Lars von Trier, with Nicole Kidman as the statistical distance and Philip Seymour Hoffman as the random variable Z. All the characters are trapped between the pages of a prestigious journal!
5) … people resort to obscure foundational problems to motivate an elementary experiment, whose results are published in Science.

Seriously, does it make sense to conduct Wheeler’s delayed-choice experiment in the 21st century? In the same line, why not measure the speed of aether, or the weight of a soul?

Experimental demonstration that the pagan god Mawu does not manifest when the pentacle is open. Did Mawu know in advance that we were going to open the pentacle? To appear in Nature Communications.


All right, enough cons for today.

Let's hear the pros: working in Foundations rocks because…

1) … contrary to absurd claims, there IS a measurement problem*.

*The problem is to explain why measurements return a single outcome, not why we don’t see macroscopic superpositions. Decoherence advocates, cut the crap: you’re not advancing the problem at all!

2) … problems are interesting by themselves. Not because you can relate them to algebraic topology, not because solving them will prove everyone how clever you are* and not because Terence Tao has worked on the topic before.

*Electroshock, please.

3) … there is room for imagination.

OK, rats, this has been all. I hope that this post has inspired you to do something productive, like insulting me in the comments meditating on deep foundational issues. It has certainly inspired me to try to get invited to QCRYPT next year by republishing my 2006 hit on Optimality of Gaussian Attacks in CVQKD. What do you think, rats? Shall I submit it to PRL? Or should I try Nature Communications this time?

Yours truly,

Schroedinger’s Rat

26 comments:

  1. This comment has been removed by the author.

    ReplyDelete
    Replies
    1. Oops. Didn't mean to delete this. I think the internet deleted it for being such a boring comment.

      Delete
  2. A few thoughts... http://arxiv.org/abs/1308.5619

    ReplyDelete
  3. Almost certainly Colbeck and Renner are not alone with their approach for recycling results. Any other examples?

    ReplyDelete
  4. I'm glad the rat is still in business. But I'm not convinced the conventional wisdom is wrong to be excited about Reichardt-Unger-Vazirani. The McKague-Yang-Scarani paper is also interesting, but I think the key new improvement of RUV was getting the bound to work for n parallel CHSH games. This has led to applications that previously were elusive, like proving QMIP=MIP* and infinite randomness expansion (in http://arxiv.org/abs/1310.6755 ).

    ReplyDelete
    Replies
    1. (same anonymous as the last comment:)

      Maybe you view it differently because these are CS applications more than foundational ones. But from the CS perspective, their improvement makes this essentially the first useful self-testing result.

      Delete
    2. I do appreciate the Reichardt-Unger-Vazirani extension to parallel games. It is very infuriating, though, to hear from very good researchers comments like: ""self-testing"? What's that? Ah, you mean "rigidity"!".

      As for usefulness: how much CHSH violation do they require to make non-trivial claims?

      Delete
    3. What seems to be less appreciated is that you can only use the device once in this setting. Really weakens the result.

      Delete
    4. >As for usefulness: how much CHSH violation do they require to make
      >non-trivial claims?

      One of my co-authors has worked out the answer. Before scrolling down, close all your house windows, put some loud music and prepare to shout.

      The problem is: suppose that you violate CHSH by an amount of 2\sqrt{2}-\epsilon. How close is the state in your setup to a two-qubit singlet?

      Nowhere, nowhere in the RUV paper, or its Supplementary Material, or any talk I have attended about this result, you'll find a concrete answer. You'll just find expressions of the sort O(\sqrt{\epsilon}) for the trace distance; or, equivalently, 1-O(\epsilon) for the singlet fidelity.

      Well, as it turns out, if you connect different theorems in the Supplementary Material you do find that the singlet fidelity F equals 1-O(\epsilon). More concretely,

      F>=1-10^6\epsilon.

      In other words: in order to make a non-trivial prediction with the RUV results, you "only" have to improve the precision of current Bell experiments by four orders of magnitude. A very useful result, indeed!

      I wonder if the authors are also contemplating to write a paper on cryogenics, explaining how to reanimate a frozen body: "you just repair every single cell membrane, and -hop!- you're done!"

      Delete
  5. Well, the Rat is not as brave as I thought. He waits till just *after* the QIP decisions go out to to insult the PC chair publicly. This raises a point, though. Could we have a post discussing the QIP program once it becomes public?

    ReplyDelete
    Replies
    1. Two things:

      a) I am not insulting anybody.

      b) One of my co-authors asked me to postpone the post until the QIP decisions were made. Had I submitted my work alone, I would have published the post a month ago.

      Delete
    2. Come on. Do you think it's complimentary to say that someone has published the same result three times? In this case it appears to be true, but surely the authors of those papers would say they're at least slightly different ideas.

      Delete
    3. Scott: pointing out that your car is badly parked is not complimentary either, but it is definitely not an insult.

      In this blog I may speak in very hard terms about publications, etc., but I try my best to avoid personal attacks. E.g.: I would never write that this or that person is a bad researcher, or that I wish him the worst. Such comments, if taken seriously, can destroy a person's ego. An attack on a publication, on the other hand, can only affect its perceived value.

      Sadly, many people in QI do not see the difference.

      Delete
    4. Scientists are very attached to their results, and usually identify with them so strongly that an attack on the result seems like an attack on them. You know this very well, I'm sure.

      Delete
    5. It seems we can learn how not to take work-related criticism personally from our fellows mathematicians: http://www.math.columbia.edu/~woit/wordpress/?p=6430.

      Delete
  6. (this is Anonymous 24 November 2013 05:27----I forgot to include my name)

    I wanted to address the Rat's comment, "Or should I try Nature Communications this time?" You should no that nobody aims to publish their paper in Nature Communications. It must first take "the path": Submission to Nature, Submission to Science, Submission to Nature Physics, and then reluctant submission to Nature Communications. Probably Nature Communications publishes the most thoroughly rejected papers in the world. They may actually reject a few, but not a whole lot.

    ReplyDelete
  7. Hi Miguel,
    You are putting the bar very high by your negative examples: there is much worse down there, which we fortunately do not see and better not dig out.
    Specifically, I don't think that Leggett's is a bad example of foundations. He had the idea back in 1976, but at that time he had better physics to concentrate on; when he finally published it, he put it (better: buried it) where it should, Foundation of Physics. But then, he shared his idea with Zeilinger's group and --- hey, that is a good work because it gives testable predictions! Their paper, and the hype that followed and to which I contributed, definitely benefited from Leggett being a Nobel Laureate, especially in press releases. But Leggett himself followed these developments with amazing detachment: how many of us would have requested to co-author some of those prestigious publication, given that we all discussed with him.

    ReplyDelete
  8. Now, let me give you another line. A model that definitely sucks is the one Antoine Suarez and I conceived back in 1997 http://arxiv.org/abs/quant-ph/9704038. It was nice because testable, but frankly a bit on the crackpot side. While the tests were being performed, it took me four years (OK, disturbed by an ongoing PhD on magnetic nanostructures) to figure out that the model is actually signaling, contrary to the very spirit in which the original paper was written (http://arxiv.org/abs/quant-ph/0110074). So, that's it...
    ... or maybe not: in the process of finding out that the model was signaling, I moved on to conjecture a way of disproving all forms of "hidden signaling" models. This line of thought lead to a very cute paper by Stefan Wolf (http://arxiv.org/abs/1102.5685) before it's completion by some friends, who were kind enough to put my name on the paper (http://arxiv.org/abs/1110.3795).
    OK OK, maybe all this sucks --- but I had fun ;-)

    ReplyDelete
  9. Some thoughts:
    You should work on a test for distinguishing quantum Foundations papers from noise. Maybe reading the abstract of the paper to a dead cat, and it it comes alive it's not noise. At least you know that everyone you mentioned, in both the pros and cons, is noise. How come none of the people you mention are working on ER=EPR. Is it because they are all too dumb, or is ER=EPR dumb?

    ReplyDelete
  10. Thank you! Please post more often!

    ReplyDelete
  11. As to 'delayed choice, maybe http://ptep.oxfordjournals.org/content/2013/4/041A01
    or http://arxiv.org/abs/1211.1574 or...?

    ReplyDelete
  12. Two problems with Foundations:

    1. Too many occasional writers
    [Solution to the problem of the black body radiation was regarded to be sufficient for the associate professor position.]

    2. How many important contributions should be made by a person? Stop making of the scientists the simple employees!

    ReplyDelete
  13. I agree in particular with point (2). These prizes are created and then monopolised by the old boys club. What we need is a "people's choice award", or "shadow bell prize". It would be easy to set up an Internet vote amongst some sort of lightly restricted population (e.g. active foundationalists under 45 who read at least 10 papers a year.) I suggest the annual winner receives a symbolic one dollar and a paper crown which they can wear at conferences for a year before it needs to be transferred to the next recipient. To rectify long standing injustice, we could kick it off by voting 5 lifetime achievers like Tsirelson.

    ReplyDelete
  14. Brilliant rant. Funniest thing I've read in a while.

    ReplyDelete