Wednesday, December 14, 2022

Dynamical Models do not Generate Free Will

Physics professor Scott Aaronson's blog attracts trolls, but a couple of comments there are so glaringly wrong that it is worth showing why.

SR Says:

As pointed out by Mateus Ara├║jo above, QFT is local, so it seems to me that the only ways to reconcile this with the nonlocality of standard QM evidenced by the Bell test are: (1) QFT being totally incorrect in an easily-measurable way, (2) Many-Worlds, so that the appearance of non-locality is only due to our observable universe residing in a slice of the “true” wavefunction, (3) superdeterminism.
If those were really the only consequences of the Bell tests, then the Nobel citation would have said so. It did not.

Fred writes:

“It’s very weird that we feel that we have the free will to perform experiments and the consciousness to feel like we understand them, when, to our best understanding, everything is predetermined by the laws of physics.”

What’s even weirder is that determinism isn’t even the core culprit here! You only have two ingredients at each end of the spectrum: perfect determinism (e.g. a bunch of balls moving around and hitting one another… i.e. every event is caused by a prior event), and pure randomness (everything is uncorrelated noise, i.e. no causality, things happen without a prior cause… very weird too when you think about it).

And then you can mix those two in various amount, on a continuous scale. QM is a mix of determinism and randomness, somewhere in the middle of the scale. MWI + consciousness also seems to lie in the middle of the scale (the wave function of the universe is determined, but my place as a conscious being on that structure seems random, from my subjective point of view).

When it comes to free will: sure, determinism seems to obviously exclude it… but randomness seems to exclude it too! For the general idea of free will isn’t exactly understood by throwing a dice at every moment a supposed “decision point” happens. He is right that our dynamical models do not generate free will. That is why it is called free will.

He responds:

You might as well call it pixie magic dust then?

A dynamic system is a system which state evolves with time (a parameter).

My point is that dynamic systems either evolve following causality (current state is derived from prior state) and/or randomness (current state is independent of prior state), and then any degree of mix of those two things (where events depend partly on prior events and partly on some randomness).

Note that randomness is non-determinsm, meaning an event without any cause within the system. Whether that randomness is pure (appearing magically within the system) or is a dependence on causes external to the system is basically the same.

That’s it!

What other ingredient would there be?

No, randomness is not an event without any cause within the system, it is an event without cause in the model. Something is random if your model cannot predict it. There is no such thing as pure randomness.

Free will is a form of randomness that no one is even trying to model.

He moved on to the latest scientific hoax:

Looks like fusion energy supremacy has been demonstrated!

Scott Says: Only “scientific supremacy,” not supremacy where you account for the actual energy cost of the lasers which is still 100x too high. Still good though!

They refer to this story:
Scientists with the U.S. Department of Energy have reached a breakthrough in nuclear fusion.

For the first time ever in a laboratory, researchers were able to generate more energy from fusion reactions than they used to start the process. The total gain was around 150%.

"America has achieved a tremendous scientific breakthrough," Energy Secretary Jennifer Granholm said at a press conference.

The achievement came at the National Ignition Facility (NIF), a $3.5 billion laser complex at Lawrence Livermore National Laboratory in California. For more than a decade, NIF has struggled to meet its stated goal of producing a fusion reaction that generates more energy than it consumes.

To make this claim, they ignore the energy required to run the lasers, and to power the confinement structure. As Scott notes, they are really putting 100x energy in.

The analogy here is that this new experiment is supposed to show that it is possible to generate fusion energy, even if the overall efficiency is lousy. Quantum supremacy is supposed to show that it is possible to generate a super-Turing computation, even if it is completely useless.

Update: I added this comment:

@fred: You say that events are either caused by the previous state (determinism), or independent of it (random), and magic pixie dust is the only other possibility.

Consciousness is the other possible cause. My conscious state appears inscrutable to you. You cannot model it, and my decisions appear unpredictable. You as might as well call them random, if you cannot predict them. I cannot really explain it either, except to say that I am more sure of it than any of my perceptions. Cogito ergo sum. (I think, therefore I am.)

Readers here might say I could be a troll or an AI bot, and hence not to be believed. Fine, decide for yourself. Do you really think that modern physics has explained causality so thoroughly as to rule out human consciousness? Or do you make decisions everyday that science cannot predict or explain?

Update: Another comment:
They got a breakthrough on Nuclear Fusion. No one now can say quantum computing is the nuclear fusion of the 60s.
Not sure if this is intended to be sarcastic. Quantum computing is more like nuclear fusion than ever, with both fields making big splashes with supremacy claims of no practical significance.


  1. The entire 'omg it's gonna be nuclear fusion just you wait' schtick is growing old. It's so old, that it's over eighty years old now, it's always been 'fifty years away' since they first started blathering about it. I have reasons to believe it will not be anything more than another financial transfer to people wearing lab coats. I've heard no explanation whatsoever of how a fusion reactor would actually be used (talk about global warming) and how any kind of magnetic containment vessel could withstand fusion levels of heat continuously...which would be required if you were going to produce energy with the damn thing.

    In addition, consider that we already have the means to produce nuclear energy with new designs of power plants that are far safer and cleaner with todays technology. The primary reason they are not being built is the same reason nuclear fusion will never 'be a thing' even if they mange to pull it off, certain folks (like the president and those he answers to) don't want there to be cheap energy, they want it to be more expensive, not more affordable for various malthusian reasons...and they want to be the ones who centrally control said energy and profit off it with bullshit taxes that serve no purpose other than to entrench rent seekers without any added value.

  2. Dear Roger,

    > "No, randomness is not an event without any cause within the system, it is an event without cause in the model. Something is random is [if] your model cannot predict it. There is no such thing as pure randomness." [Added correction: is -> if]

    Put very well!

    My additional comments:

    Pure randomness does not metaphysically exist. But it can be defined, in suitable contexts, as mathematical concept, via a limiting process. Randomness refers to the degree of absence of order [of a certain kind specified in a certain context].

    Pure randomness [in that context] therefore is the absence of every form of mathematical relation that could possibly capture orderliness [in that context]. Thus, the meaning of "pure randomness" in the context of one theory or model is radically (roots-wise) *different* from that in the context of some other theory/model.

    I love randomness. It's a handy tool. It's a tool of a certain kind of mathematical abstraction. In fact, you can use this tool to develop a certain higher-level description of a system that's not only completely deterministic but also within the means of calculations.

    For instance, if a punching machine has a rotor mounted with 5 different punches, and if in a single cycle of operation, all the punches only operate in a definite sequence, you can still say: the probability of a particular punch hitting the work-piece is 0.2 (in the limit of infinite number of punching cycles). That's the beauty of this concept: It allows you to size up the system in a certain way -- a way in which any order is [deliberately] disregarded out of consideration. But of course, disregarding out of consideration is not the same as saying not existing.


    As to the title: "Dynamical Models do not Generate Free Will"

    Pretty neat.

    Actually, it can be expanded... a lot! Dynamical models do not generate soul, life, or consciousness, and hence, not free will either.

    Essentially, from a physics perspective, free will plays the role of supplying an additional set of auxiliary data (initial, boundary, and continuing conditions). For example, in a computer game, when you grab an object using the mouse or so, and "throw" it (in the game), you are imposing this *additional* set of auxiliary conditions. The origin of this extra set is in your free will. You are a dynamical agent (because you have a body too, not just the mind), of such a nature, that you are able to *generate* such auxiliary data that it can affect the evolution of the game. And notice, every object in the game follows the equivalent of physical laws at *all* times. Also you (the special dynamical agent with soul, life, consciousness, and free-will).