Monday, December 12, 2022

Impossible or Fundamentally Impossible

Here is Scott Aaronson's main argument for quantum computing:
I notice that you never responded about a fault-tolerant QC running Shor’s algorithm. Do you believe that that’s fundamentally possible, or not? If not, what physical principle is going to come in and prevent it? Will you agree that the discovery of that principle would be a revolution in physics?
He draws a distinction between what is impossible, and what is fundamentally impossible. I am not sure the distinction makes any sense.

In other words, QC might be impossible, but it is not fundamentally impossible unless some law of physics forbids it. We have not found that law of physics. The Extended Church-Turing Thesis forbids it, but he says it is not fundamental and it "is still on thin ice."

A response:

From an engineering point of view, there are often unforeseen limitations emerging from complex interactions of different domains such as physics of materials, chemistry, thermodynamics, mechanics, economics, etc. Those different knowledge fields are themselves at a much higher conceptual level compared to the underlying basic physics they all share, so their own laws/heuristics only hold in specific domains with specific assumptions, and all those various models (often highly non linear) just don’t overlap.

There’s nothing in the basic laws of physics explicitly saying that you can’t build a stable stack of quarters from here all the way up to the edge of space.

But do you believe it can be done? Given an existing stack of quarters, it’s trivial to just add one more quarter to it, and then by recursion assume the stack can be arbitrarily high. But that’s not how system scalability works in practice: at some point, what works for 100 quarters won’t work for 1000 quarters, because new problems are introduced: e.g. the wind will screw things up, and if you build your stack inside a tube with a vacuum, you’re now facing another totally different engineering challenge (create a 100 mile-high tube that can contain a vacuum). And, even without air, you’d have to deal with the effects of tides, plate tectonics, strength limitations in alloys, etc.

There’s also no specific law of physics telling us whether building room temperature super-conductors is impossible.

Same about building a stealth bomber that can travel faster than mach 5 at sea level.

It also goes the other way: a hundred years ago, it would have seem impossible (given the technology of the day, but given pretty much the same laws of physics) to build a gravitational wave detector that could measure changes in distance around 1/10,000th of the diameter of a proton, between two mirrors separated by 4km.

So, for the vast majority of hard engineering problems (and building a QC *is* a hard engineering problem), the fact that there’s no clear black and white basic principle saying it’s impossible isn’t really helping much at all. It wouldn’t be the first time we set up to build something, and then it never happens because various requirements just can’t be met within the same system (often it’s quietly killed because money runs out and people move on to other things because some new engineering progress makes an entirely different problem more exciting to work on).

So maybe QC is like that gravitational wave detector. It seemed impossible for a long time, until some huge technological advances were made.

In the same thread, Aaronson ridicules the idea that Einstein might have thought that ER=EPR. He did not even believe in either wormholes or quantum mechanics. It is not clear today if anyone really believes this wormhole entanglement nonsense. Somehow Einstein did inspire a lot of bogus physics thinking.

Sabine Hossenfelder has a new video on Quantum Uncertainty Simply Explained. She correctly describe the uncertainty principle as an essential part of quantum mechanics, but also explains that all waves obey an uncertainty principle. The uncertainty principle is a way of saying that electrons have wave-like behavior.

Update: Quanta has an article by Robbert Dijkgraaf saying There Are No Laws of Physics. There’s Only the Landscape. That is because he is a string theorist who studies mathematical abstractions that have nothing to do with reality.


  1. I find this little bit hilarious:

    "It also goes the other way: a hundred years ago, it would have seem impossible (given the technology of the day, but given pretty much the same laws of physics) to build a gravitational wave detector that could measure changes in distance around 1/10,000th of the diameter of a proton, between two mirrors separated by 4km. "

    You can't. Anymore than you can measure cosmic dust polarized by the big bang over 14 billion years ago. LIGO is complete and utter bullshit. You can't measure (meaning with instrumentation AND/OR fancy publicity hyped interferometry) something magnitudes smaller than the diameter of a proton between two mirrors separated by 4km on the surface of the Earth. I know all about the CLAIMS that LIGO made, but I also know the very mirrors themselves, and every single atom of structure (the building, the earth, and attendant billions of people surrounding said structure) in between the two sensors is vibrating far more and at much closer range than the (entirely imagined and thus fabricated) signal they were supposedly looking for. Whatever the hyped claims, the signal to noise ratio is so astronomically absurd that even if they actually could screen out all possible extraneous sources (which they simply can't, and they even publicly admitted they had to stop walking around their own facility while testing or it could compromise their little 'test') and get past the wee bitty problem of 'space vibrating' (it can't even according to Einstein, curved space exists only as static geometry in a mathematical block universe that doesn't move and can't accommodate things that do move).

    In addition, would anyone like to calculate the comfortable odds of such a hypothetical detection the moment you turned the stupid thing on? Like really? What do you think those odds are? So much for remotely intelligent skepticism at Cal Tech.

    I also notice LIGO has been fairly quiet for quite some time (years). Where are all these miraculous gravity detections that were going to 'revolutionize' astronomy? The only thing LIGO actually found was even more funding.

  2. Dear Roger,

    As I indicated a few days ago in a comment right on this blog, no, the uncertainty principle is *not* an essential part of quantum mechanics.

    What it rests on *is*, however, an essential part of QM, namely, (i) support of the wavefunction over all space at every moment, (ii) measured states as eigenstates of a Hermitian operator, and hence, the crucial relevance of the Fourier theory.

    But people are unable to separate out this *base* from the ``uncertainty of knowledge'' etc. sort of bullshit. The latter is emphatically not even a non-essential part of QM, let alone an essential part of it.

    Any one who believes the latter is plain in the wrong, and any professor or pop-sci writer who doesn't isolate these two parts apart (the base and the unwarranted irrationality) is careless, to say the least.


    About the QC and ``fred'' # 166 on Scott's blog:

    He says:
    > ``There’s nothing in the basic laws of physics explicitly saying that you can’t build a stable stack of quarters from here all the way up to the edge of space.''

    Well, there is. Forget arbitrarily big height, even for a much smaller height (may be, like, a few km, perhaps even just a few hundred m), there would be issues. On principle. An explicit principle.

    The operative word is ``stable.'' So, you've to ask: Stable, against what? The universe is not static. It's dynamic --- there is time (in the universe). Even if you take care of all the *external* destabilizing factors, there still are the motions of the elementary particles inside each object (like a quarter). They would make not only maintaining stability impossible, but also just placing the actual quarters perfectly on top of each other. It would require controlling ~10^{20} initial conditions per quarter.

    And, controlling even just the ICs becomes important because of an explicit principle: SDIC (sensitive dependence on initial conditions).

    Counter-point: Chaotic systems too can maintain stability. Here are a couple of posts I wrote at my blog that are relevant:

    “Fundamental Chaos; Stable World”, August 2019. [ ]

    “Determinism, Indeterminism, Probability, and the nature of the laws of physics—a second take…”, May 2019 [ ]

    Application of the counter-point in the case of quarters: Friction at the macro-scale (van der Waal's at the micro, and bonding and tunnelling at even smaller scales) will provide the stabilizing influence.

    Overall: The principle (SDIC) is very broad, not sufficiently well studied: there is not a single model system which can be of a direct relevance to my iqWaves theory for example; and indeed, even for many systems that have been studied, all the dynamical regimes still aren't well isolated. Therefore, application of SDIC must proceed very contextually and carefully, on a case-to-case basis.

    As to the QC: As I've mentioned many times (including at Scott's blog): When it comes to the QC, I am a skeptic, albeit a soft skeptic. I don't think they are going to be able to build a *scalable* QC (one which breaks RSA, or even a much smaller QC) in any foreseeable future. ... May be they should take a course or two on mechanical vibrations *and* control (including through mechatronics). They would realize just how hard controlling any dynamical system is, especially one that involves vibrations. But anyway, to conclude, as far as I am concerned, it's *their* money...


  3. Neither you nor Scott understand the harmonic analysis of Boolean functions. That where the debate is happening regarding NISQCs.