Hacker News new | ask | show | jobs
by Certhas 698 days ago
I used to work close to people who were actually somewhat close to the stuff contained in the papers. The consensus was that there is nothing of substance to engage with.

Edit: When the technical papers appeared in 2020, I personally went through them in some detail. Tl;dr there are almost no novel ideas of substance in there.

Specifically I looked at the "launch documents" provided here:

https://wolframphysics.org/technical-documents/

which, to my knowledge, still are the closest we have to a coherent description of what the grand vision is. Unfortunately I didn't keep my detailed notes, but looking specifically at the relativistic paper, it might appear substantial, but that is because large parts of it review well-known basic results in discrete geometry and causal sets. The actual content is described in a hand wavy way, with little in calculations or rigor (and some elementary mistakes, too).

The issue is that everything that goes beyond standard results is essentially wishful thinking or circular. "If my update rules are such that they produce a causal structure that corresponds to that of a 4-dimensional spacetime, then the wolfram model produces a 4-dimensional spacetime!". This would be interesting if there was any way to characterize the update rules that do so. However, there is not. There is simply the implication that since update rules are very general it must surely be possible to find one that does. Actually doing so is left as an exercise to the reader.

A prime example is in Section 3.3:

In all that follows, we shall assume one further condition on the hypergraph update rules, beyond mere causal invariance: namely, “asymptotic dimensionality preservation”. Loosely speaking, this means that the dimensionality of the causal graph show converge to some fixed, finite value as the number of updating events grows arbitrarily large.

However, abstractly defining ensembles of causal graphs that actually produce (at least with high likelihood) the causal graphs of low dimensional manifolds is exactly the core of the issue. If you are able to do that, then the standard results of causal set theory get you the rest of the way. This central difficulty is simply "assumed" to be solved. No further discussion is given on what type of update rules would actually be dimensionality preserving, nor is this identified as a key research question, nor is any evidence or heuristic provided that WOlframs approach has anything new to say on this problem.

As far as I recall the quantum mechanics paper was even worse.

4 comments

(Disclaimer: I have a physics degree but I’m not a practicing physicist.)

I think the above comment perfectly summarises the situation.

There has been a lot of fanfare but no action coming from Wolfram’s research.

It’s even more disconnected from physical reality than the more abstract mathematical corners of string theory.

The hard part of a TOE is showing how it maps to reality, how the theory constrains what we can measure in future experiments, etc… This is the part that Wolfram keeps skipping over.

I’ve had an interest in theories of everything (TOEs) and I’ve read through hundreds of papers from serious publications to gibberish put out by mentally ill cranks. I’ve developed a checklist to filter out the noise. Wolfram’s theories don’t tick any boxes! Even crazy rants on some personal blog written in random fonts with ten text colors do better.

I remember looking at the hypergraph stuff a while back and thinking "this can't possibly handle the double slit experiment". I looked through the different pages and found no references to it. I shrugged and moved on. A few years later another blog post by one of the members went up saying "of course we can explain the double slit experiment, from the beginning. We left it out of the early posts on purpose so someone else could do it and get some credit, but nobody took us up on it, what a sad world". It went on with some further hand waving that I didn't understand. It all just felt like a sure sign of brokenness.
My key threshold is the three particle generations, e.g.: electrons, muons, and tau particles.

If your TOE doesn’t even mention them, then that’s a bad sign. Any mention at all will have me sitting up straight and putting my reading glasses on.

Wolfram’s theories are so, so far away from this threshold that it’s hard to even explain to non-physicists.

I saw a claim that they had produced GR from their model. Does that count for anything?
No. That's surprisingly easy[1], and the last time I checked they hadn't reproduced anywhere near the full theory of general relativity. They've just found a thing that suggests that it's possible to map things to a curved spacetime manifold. If I remember correctly, they didn't actually show that this was in any way natural or the only possibility.[2]

[1] There is more than one way to model GR or large subsets of it either mathematically or physically. It sounds like a complex theory but in many ways it is actually "minimal" and highly constrained, making it pop out of unrelated things surprisingly easily. For example, crystal defects moving under thermal vibrations of the lattice can model GR! Similarly, variable index of refraction has very GR-like mathematics and can model everything except torsion (I believe, I'm not an expert).

[2] A key thing with such fundamental theories is not just to show that it can do something in physics but that it cannot do anything else.[2b] Without that constraint, any general computational theory like Turing machines "contains physics". So does the space of all computer programs, lambda calculus, etc... Those aren't useful statements, but that's pretty much what Wolfram's theories boil down to. He keeps finding very simple systems that can compute arbitrary things, pointing at them and exclaiming that "Physics is in there... somewhere!"

[2b] E.g.: Four dimensions of spacetime with a +++- or equivalently a ---+ signature, but not anything else such as ++++ or ++-- or five dimensions. Similarly, three generations of particles, not two or more than three. Etc...

> The consensus was that there is nothing of substance to engage with.

The safety boat of scientific consensus is pulled out a lot in today's environment. One should remember that many of our great scientific discoveries had a scientific consensus that it replaced. That boat won't always steer you in the right direction, sometimes you have to read the paper and come to your own conclusion.

What reason do you have to believe that your own conclusion will be better?

The causal set folks are already outsiders that go against the broad consensus of HEP-Th. I grew up scientifically in an environment that was actively challenging mainstream consensus. Sometimes correctly (e.g. low energy susy), sometimes not. If those folk can't find something redeeming in what you do you should stop and listen.

Also, yes scientific consensus at the cutting edge changes over time. That's the nature of science. But I challenge you to find a single example where the consensus was "there is nothing of substance here" and it turned out to be wrong. Not all forms of challenging consensus are equivalent.

At the end of the day though, this is HEP. Itt doesn't matter. Worst that can happen is that you waste your time learning some neat math.

Scientific consensus rarely has a large body of people who truly dived into the research. Most are very busy with their own research and lives. At worst they just repeat views they hear from colleagues they respect, and at best they will do a minor survey. Therefore it's questionable if more than a handful of people truly have any valuable input to that consensus. What you end up with is truth by authority which is antithetical to the scientific method and ethos of scientific inquiry.
https://calteches.library.caltech.edu/51/2/CargoCult.htm

It just popped up on HN. I suggest giving it a read.

I have a question for you. How many scientists does it take for consensus, that we should determine science is settled? Or is it always open for inquiry?
I’ve seen it claimed that the project has led to to a way of doing a numerical simulation of GR which is, in some cases more efficient?

If true, that still seems like something of merit, even if the project doesn’t give any progress in fundamental understanding of physics?

Maybe, rather than as they hope, being a path towards a theory of everything, it could instead be a path towards a framework for understanding good ways to do numerical simulations that respect causality, while not necessarily doing all of one time coordinate everywhere (in some reference frame) before computing later times anywhere?

If you have a citation or more information on that, I'd be curious. Numerical relativity is outside my area of expertise. It doesn't seem likely to me, but I couldn't rule it out...
I believe this is what I was thinking of : https://arxiv.org/abs/2308.07508

The author is Jonathan Gorard, who is one of the people associated with “The Wolfram Physics Project”, and checking the references, it does cite at least one document/paper that is part of the “wolfram physics project”,

But I don’t know for sure whether it exactly uses the central framework of the project.

Well the problem is that's then essentially people involved in the project making claims about the project.

That's obviously not satisfactory as a positive independent assessment, which is what this thread is calling for.

Sure, this was just me saying “I think this was the thing based-on/using it, that was said to work well”, not “here is someone saying it works well”.

The “gravitas” program is open source, so perhaps some external people may have evaluated how useful it is? (Unless it is hard to get running, idk.)

Thank you!
Is it just because it hasn't gotten as far as making physical predictions yet? To put it charitably it's very abstract, but I wonder exactly where the holes are that real physicists see.
Physicists won’t take a look at a new theory unless the person pushing it can demonstrate very good reasons for physicists to do so. Generally those have to be quite concrete reasons: for example explaining a known phenomenon in a much clearer or more intuitive way, or allowing the explanation of systems that weren’t easy to conceptualise of before, etc.

But ultimately it’s up to Wolfram to come up with those things. I don’t think most physicists feel he has done that, especially since the standards increase as the idea becomes more different to existing physics

> Physicists won’t take a look at a new theory unless the person pushing it can demonstrate very good reasons for physicists to do so

So why did all the string theories get popular?

FWIW, by my reading string theory is, incredibly, a lot closer to being testable than this thing, which is more like a proposed formalism in which to expess a theory than a theory itself.
Because it did do those things
A "theory" without predictions is just a bunch of words and numbers hanging out together.
> A "theory" without predictions is just a bunch of words and numbers hanging out together.

A famous example in Physics is String Theory. It has been around since at least 1980s and still no definitive way to prove or disprove it.