Saturday, March 18, 2017

Hossenfelder sensibly critical of our "simulated" world

Sabine Hossenfelder writes a lot of wrong texts, especially about issues that depend on some nontrivial calculation. But she is often reasonable when she discusses certain conceptual issues, including the general properties of quantum mechanics (and the absence of non-local influences in QFT etc.).

The latest example of the penetrating texts is
No, we probably don’t live in a computer simulation
I've discussed the proposals that "our world has been programmed by our overlord, Ms Simulator" in 2011, 2013, 2016, aside from other moments.

But let's look primarily at the comments by Hossenfelder and her readers – who surprisingly seem to agree.




First, she mentions a philosopher named Nick Bostrum who loves the idea that we've been programmed by Ms Simulator and we should be worried that she presses the button and turns us off. (She may also selectively turn those people off who don't read TRF or have never... Oops, I removed the joke about the corrupt simulator and donations after I received two. It wasn't really my goal! I have no evidence the simulator is selectively turning people off.) Brian Greene was among those who have promoted these ludicrous ideas. Hossenfelder describes the intellectual motivation of the fans of this concept as follows:
The simulation hypothesis, as it’s called, enjoys a certain popularity among people who like to think of themselves as intellectual, believing it speaks for their mental flexibility. Unfortunately it primarily speaks for their lacking knowledge of physics.
Even though I can't reliably see into their skulls, I tend to completely agree. Fans of this simulation paradigm want to prove that they are flexible – and it may be true even for the likes of Brian Greene or Neil deGrasse Tyson. They have opposed religions and they internally agree with some of their critics who say that the atheist proponents of science are "spiritually narrow-minded". So they picked this replacement religion to show that they can also "think big", like the believers, and they're not narrow-minded, after all.

A problem is that their "simulation paradigm" isn't just "big". It suffers from the very same problems as most of the religions.




As physics, the simulation hypothesis sucks, we read at Backreaction, because the purpose of physics is to explain the observed phenomena and patterns in them and the simulation hypothesis just doesn't explain any. It's worse than that: the hypothesis sends us back to the age of mythology. It wants to replace the conventional physics in the very questions where physics seemed relevant – but it pays no attention to the arguments and insights about the laws of Nature that physics has actually accumulated. I completely agree with that.

Now, when you actually try to use this hypothesis to produce predictions, you should admit that you create predictions such as the "deja vu cats" – typical programming glitches – and the violation of the seemingly continuous symmetries in Nature. If the simulation runs on a "computer" with discrete information – whether it is classical bits (totally hopeless to simulate a quantum world, Hossenfelder agrees) or quantum bits – you should observe violations of the continuous symmetries. It's very hard to avoid this prediction and the general simulation hypothesis really predicts that those violations shouldn't be small. They are more likely to be of order one. But we haven't observed any.

If we were doing physics, the theory would be abandoned. However, the champions of Ms Simulator tell us:
It doesn't matter that this is predicted. Ms Simulator has arranged things so that we don't notice the manifestations of the discreteness, computer glitches, or other contradictions.
Fine. But this quote really says "let's not pay attention to the empirical evidence, all of it may be fake". And once you stop paying attention to the empirical evidence, you stop thinking as a scientist.

This point isn't just about some arbitrary definition of what is science and what isn't science. The problem is that science, as conventionally defined, is the only way that works how to find out something about the rules underlying the world we inhabit. When you decide that you don't pay attention to the empirical data, you will simply not find any justifiable insights about the world. What you will think about the world will be pure prejudices.

I always find it important to add that even if some divine entity were deliberately obfuscating some deep truths about our world, if She did it really well, the moral duty of the scientist would be to exactly find out what She wants us to think. If some deep features of the world can't be found even in principle, then they literally don't exist according to the scientific definition of existence. If it always looks like the statements \(A_i\) are true in our Universe according to all observations, then a scientist should agree that the statements \(A_i\) really are true. If Ms Simulator masks things perfectly, so that the hidden features and unusual predictions are unobservable even in principle, then she manages to change the truth, indeed. If you disagree with this notion of the truth, then you think as a non-scientist.


Kids have the duty to dance in front of cameras to amuse grownups. Original BBC footage from an expert in Japan. In the same way, scientifically inclined folks have the duty to pursue the scientific method to amuse Ms Simulator, even if She knows that they're finding things that aren't fundamental in Her world. (She doesn't know that Her world is a simulation, too LOL.)

So you should better pay attention to the predictions done by the simulation hypothesis (just like any hypothesis in science) and compare them fairly to the observations. And they just don't seem to agree with anything we observe. The hypothesis predicts deja vu cats, there aren't any. It predicts things will generally be discrete and not rotationally symmetric or Lorentz-invariant. But they are continuous, and so on. The comparisons with the experiments say Bad, Bad, Bad. Again, if you don't care, you're not thinking as a scientist.

One aspect of the simulation hypothesis is that the data on which "we run" are fundamentally discrete. As I mentioned, it seems to contradict the continuous symmetries we seem to know in physics and other manifestations of the fundamental continuity in Nature. But even if the contradiction didn't exist, I would find these people's assumption that "discrete is better" to be a totally irrational prejudice. Criticisms of discrete physics on this blog are numerous.

For example, these people may want to assume that the simulation needs a finite number of bits and a finite number of binary (or, in the case of the more sophisticated proponents, qubit) operations per second. But why would they assume such a thing? The only rational reason to make a similar assumption is that we know that if we consider machines we can create in this world, they have this property. We need to do some particular piece of work if we want to add one bit to a RAM memory chip, or to make a CPU capable of manipulating with an extra bit etc. And we know that we may only make a finite amount of work in our finite lives so everything we produce will only manipulate with a finite number of bits etc.

But while these limitations on the "finiteness of memory" etc. are justifiable for man-made objects, they are absolutely unjustifiable for "divine entities" that were creating the world. Those entities aren't necessary equivalent to us, or like us, so it makes no sense to assume that they have a trouble to work with continuous functions or numbers that in principle require an infinite number of bits to be reproduced or similar things. Instead, the evidence is rather clear that the fundamental laws of physics start with objects that may remember some continuous information – that would need an infinite number of bits to be really exactly simulated etc.

Just like we said that Ms Simulator Herself is in principle the same thing as God, just described with some would-be atheists' spin, we may also say that the discreteness assumed by these people is just a modern-era counterpart of the anthropomorphic appearance of Ms Simulator or God. Our man-made objects (computers etc.) seem to store the information digitally. But that doesn't mean that the same holds at the natural, fundamental level. We're not equivalent to the creators of the Universe so there's no reason to assume that we have the same characteristics. In fact, it's pretty obvious that we don't.

The crazy thing is that many advocates of the simulation hypothesis – including folks like Brian Greene – are self-described atheists who often tend to say that it's naive to assume that the creator of the world looks like a man who may fly on his cloud etc. But the assumption that He uses some gadgets that store "discrete information" is pretty much equally anthropomorphic. So if they were fair, they would mock themselves by exactly the same observations – to assume that God is anthropomorphic is childish – as they mock the religious believers. But many of these people think that because their religion is newer or because they have been called scientists by themselves or others, they're immune against being childish as the religious people.

Well, you are not immune at all. If you suffer from equivalent prejudices, commit similar logical fallacies, or equally fail to eliminate the hypotheses that disagree with the empirical data, you suck as a scientist to the same extent as the believers do. It's really the same and your pretending that you are better for some reasons is absolutely silly – and very arrogant, too.

Hossenfelder correctly divides the "simulation hypothesis" approaches to those that work with discrete classical information – I am sorry, Stephen and others, but this is just damn silly and I don't even want to spend a whole sentence on it – and those with qubits. But even the latter (Hossenfelder mentions proposals by Wen et al.) are almost certainly inconsistent with the observations. To admit that the information in our world is fundamentally quantum information is just one step. But another step is to realize that it is not in any natural sense parameterized in terms of isolated qubits. Qubits mean that all the Hilbert spaces have dimensions that are powers of two, everything is revolving around base-two number systems. There is absolutely no reason to think that something like that holds in the world around us. A spin \(j=1/2\) particle may have two states of its spin but a \(j=1\) particle has three and a spin \(j\) particle has \(2j+1\) of them. None of the values of \(j\) seems to be more "metaphysically real or fundamental" than others and starting from harmonic oscillators and atoms, we know lots of systems with infinitely many irreducible states.

Later, Hossenfelder spends some time with the proposal that Ms Simulator makes sure that we don't notice the contradictions and computer glitches. She observes:
If the programmer could predict in advance what the brain will investigate next, it would be pointless to run the simulation to begin with.
Exactly. Our thinking is almost certainly one of the most complex processes running in the simulation. If Ms Simulator knew in advance – because it was used in Her program – how to interpret the patterns in our thinking finely enough to figure out "what we're planning to find out", then the whole simulation would be useless. She could deal with the "psychological" problems she was simulating directly.

Again, you may see that an anthropomorphic character of a deity is being assumed by the proponents of the simulation hypothesis. Just like the religious people assume that the creator of the Universe may listen to our prayers or have similar moral values as we have, the proponents of the simulation hypothesis assume that Ms Simulator can think about our wishes to analyze a scientific question in the same psychological way that we find relevant. Just like I said before: One has double standards if he mocks religions but praises these "clever ways" to protect the simulation hypothesis from being eliminated because the "clever ideas" of religions and the simulation hypotheses are basically identical.

In the following paragraphs, Hossenfelder says that it's difficult to create simulated minds and nations will only create one or two artificial clever brains with consciousness and use them as presidents' advisers. Maybe the technical problem is hard, maybe not. I don't think we can predict it.

But what I am pretty much certain about is that the anthropic reasoning that "we have to be simulated" if it's easy to create artificial brains is simply wrong. Even if there will be trillions of simulated brains in the future, it doesn't mean that we should belong to this set, too. There exists no logical basis for the "typicality" reasoning that we have to be "generic". This anthropic reasoning is particularly bad if you assume that the future events (will?) affect the events now, e.g. the question "who we are and what is our past". Such acausal influences are impossible and all reasoning that derives facts about the world from such influences is just wrong. I discussed this ludicrous fallacy and many others e.g. in the latest assault against the Boltzmann Brains including Carroll's. ;-)

Returning to Backreation, Hossenfelder also correctly says that the simulation hypothesis indicates that we should only observe a part of the Universe, not a whole one, and the resolution on short-distance physics should be rather lousy. Why? Simply because Ms Simulator, whatever She is, is supposed to follow similar basic qualitative limitations as programmers in our real society. In particular, She has finite resources and She doesn't want to completely waste them by simulating a space of radius 14 billion light years with the accuracy better than \(10^{-19}\) meters if the main goal is really to figure out what some humans in the simulation will think.

Again, if we were doing science, we would say that the simulation hypothesis makes some predictions – the Universe is probably not much larger than the place with interesting events; and the shortest distance scale at which the QFT-like laws work is probably not much shorter than the size of the neurons that are needed for the interesting phenomena. But both these predictions and many others are falsified by the evidence, once again, and they are falsified rather brutally. So the hypothesis just doesn't work. The advocates may still say "it doesn't matter, Ms Simulator is just playing games with us". I've explained why this only excuse means that the proponents aren't doing science and can't therefore reliably find any important insight about Nature.

Hossenfelder cried a few tears when she read a Stephen Wolfram's answer to John End-of-science Horgan:
“[Maybe] down at the Planck scale we’d find a whole civilization that’s setting things up so our universe works the way it does.”
LOL. By the way, I doubt that John Horgan could have been happy about the new civilization we will find at the Planck scale because it would mean that science hasn't ended yet, after all! ;-) As I've discussed many times, this idea of fractal substructures and civilizations is very old. Before Stephen Wolfram, it was promoted by Vladimir Lenin:
The electron is as inexhaustible as the atom, nature is infinite, but it infinitely exists. And it is this sole categorical, this sole unconditional recognition of nature’s existence outside the mind and perception of man that distinguishes dialectical materialism from relativist agnosticism and idealism.
If you want to be a true Marxist-Leninist, you simply have to believe in a civilization of little green men inside each electron and a civilization of little green submen inside the little green men's subelectrons.

The only problem with the self-similar Universe is that it is demonstrably rubbish. At the Planck scale, geometry breaks down so there can be no things with structures much shorter than the Planck scale. But even without this observation that the "march towards ever shorter distances" cannot continue indefinitely, we know that a Planck cube or an electron cannot be whole Universes that carry a huge amount of information. If electrons carried something inside, two electrons couldn't be identical, and wave functions of their positions couldn't be exactly antisymmetric etc. The permutation operator wouldn't exactly commute with the evolution operators in time so the permutation symmetry (antisymmetry) couldn't be imposed.

There's surely no known, usable theory in which there would be a substructure with objects inside an electron or every Planck-sized cube. Hossenfelder adds something else I totally agree with:
Again, however, the details don’t matter all that much – just take my word for it: It’s not easy to find a consistent theory for universes within atoms. What matters is the stunning display of ignorance – for not to mention arrogance –, demonstrated by the belief that for physics at the Planck scale anything goes. Hey, maybe there’s civilizations down there. Let’s make a TED talk about it next. For someone who, like me, actually works on Planck scale physics, this is pretty painful.
Right. Underlying is the assumption that anything goes at the Planck scale. If they – including Stephen – still had some respect towards the actual learning, they would know that it's extremely hard to design theories at the Planck scale that aren't immediately falsified – and theories of his type are. One can make talks about the paradigm postulating some non-existent little green men theories at the Planck scale but it's foolish and arrogant to pretend that they differ from any other crackpottery that is out there.

The only problem with Hossenfelder's assertion is that she has written a dozen of paper with "anything goes at the Planck scale" herself. In particular, all of her double special relativities and similar things are perhaps a little bit less obviously childish than Lenin's or Wolfram's proposals but they're still ludicrously wrong examples of "anything goes". Whether you like it or not, phenomena at the Planck scale must follow the laws of string/M-theory – or at most something that is conceptually equally abstract and advanced and morally equivalent.

Well, except that she hasn't made it beyond the late 1960s and still knows nothing about string theory even though she claims to study the very same questions, I am going to almost fully subscribe to her final two paragraphs:
In summary, it isn’t easy to develop theories that explain the universe as we see it. Our presently best theories are the standard model and general relativity, and whatever other explanation you have for our observations must first be able to reproduce these theories’ achievements. “The programmer did it” isn’t science. It’s not even pseudoscience. It’s just words.

All this talk about how we might be living in a computer simulation pisses me off not because I’m afraid people will actually believe it. No, I think most people are much smarter than many self-declared intellectuals like to admit. Most readers will instead correctly conclude that today’s intelligencia is full of shit. And I can’t even blame them for it.
Amen to that.

No comments:

Post a Comment