David Deutsch’s Constructor Theory | Hacker News Reader

David Deutsch’s Constructor Theory

constructortheory.org

108 pointsqualudeheart44 comments

> Instead of describing the world in terms of trajectories, initial conditions and dynamical laws, in constructor theory laws are about which physical transformations are possible and which are impossible, and why.

When I was a researcher, I was very interested in a conjecture that the existence of a class of reversible dynamics was enough to derive all of quantum theory. Colleagues and I published a result or two along these lines e.g. https://iopscience.iop.org/article/10.1088/1367-2630/17/12/1...

This sounds similar although much broader perhaps.

I don't understand anything about this, or even basic physics, but I clicked on one of the videos on the page and the speaker used the terms "reversible dynamics" and "quantum" so I think you might be right?

Timestamp is 00:00-02:15 in this video:

https://www.youtube.com/watch?v=Go2EO9l_L3o

From one of the talks linked at the bottom, titled:

  "Interview with Maria Violaris. A short introduction to Constructor Theory’s application to irreversibility in homogeneisation machines."

Sorry for poor punctuation, I copy-pasted from YouTube's transcript:

    [INTERVIEWER] what does it mean that the dynamical laws of physics are time reversible and
    specifically what does this mean in the context of quantum physics?

    [RESEARCHER] yeah so the the dynamical laws of physics being reversible is something that we
    see not just with quantum physics but also with normal classical physics with
    classical mechanics and electromagnetism. there's this property that when we
    look at the dynamical laws they don't distinguish between a process happening
    forward in time and the process happening backwards in time. so if we look at
    any system around us and we zoom into what's happening to the particles, and
    then if you played a video of them interacting forwards in time and you reversed
    it, the dynamical laws let us allow these interactions to happen both ways and
    times. 

    ...

    [RESEARCHER] the way that happens in quantum theory specifically is that, any interaction
    that's allowed within quantum theory, the reverse interaction is also allowed
    within quantum theory. so any process that happens, that we use quantum
    mechanics to describe, then we can also use those equations to describe the
    reverse process happening. so that's how we see this -- this reversibility is a
    property inherent in the laws of quantum mechanics

I still have absolutely zero idea on what is supposed to be new about this. Last time I read any of it, it seemed like they needed to go read some modern maths books because their ideas didn't really seem new. The papers just read like a poor approximation of something. I want some precise statements. Yeah re framing existing problems in a new language can be useful, but it doesn't even feel like it can do that yet.

The next question is why does it need a fancy website? If its that groundbreaking why has no one else said anything else about it since ~2016, which is when some of the papers on there are released eg [0-1]. They don't even seem to say much. Is there supposed to be a scam?

[0] https://arxiv.org/pdf/1507.03287.pdf

[1] https://arxiv.org/pdf/1608.02625.pdf

...since everybody (Stephen Wolfram, Eric Weinstein), seems to be investing in promoting their theory since the peer-review process doesn't seem very good fit to wade through this massive works of questionable value, and it seems David Deutsch has joined the game.

Which is a very good thing since his work looks like it might actually bring value, and it might otherwise get drowned in the noise made by the other fouls!

Good luck & keep it real!

They publish a number of papers in peer-reviewed journals. It seems like this is more about how you raise money for theoretical work and less about peer-review.

But the money needed is exceptionally low. Enough to pay for salaries and chalk... It's not like they're building a Particle Accelerator.

Salaries and chalk... and travel, and equipment (even if it's just computers), this web-site, the person who designed it... It easily could get into hundreds of thousands per year, or even millions. It's not like this is a revenue-generating venture, so there's no return on that investment. It's pure donations.

Most likely the money they raised came with strings attached. This website feels like that. Notice all the names of sponsors.

"Sure, we'll give you a few million dollars, but we want to see that you are getting the idea out there and building education. Also, put our name all over it."

Money doesn't grow on trees and getting it out of donors isn't easy.

Promoting your work gets you more citations as well. Promotional ability is one of the distinguishing attributes of the top academics.

How does constructor theory relate to Wolfram's Ruliad concept[1] and the physicalization[2] of metamathematics? It seems both theories would have major implications for the foundations of physics and mathematics.

[1]: https://writings.stephenwolfram.com/2021/11/the-concept-of-t...

[2]: https://writings.stephenwolfram.com/2022/03/the-physicalizat...

It's amazing work. Building from his work around The Beginning of Infinity, and The Fabric of Reality (1997), which Naval has been doing a podcast on: https://nav.al/agi

What makes it amazing to you?

Anyone + thoughts on what predictions it makes that can potentially tell this theory apart from the prevalent ones? Would we need to transcend falsifiability as a requirement to get there? (Ref philosopher Dawid's model)

I haven't read TFA but David Deutsch is usually pretty keen on falsifiability. I learned most of what I know about Popper from reading "The Fabric of Reality".

Potentially:

"This argument provides the indispensable theoretical basis for recently proposed tests of non-classicality in gravity, based on witnessing gravitationally-induced entanglement in quantum probes."

Thanks for this ref! (arxiv paper link - https://arxiv.org/abs/2003.07974)

Does it make predictions ? It looks like a logical fundation for physics. In computer science, it's the equivalent of inventing a new programming language instead of solving actual problems.

My understanding is that this is more of an approach than a single rigorous theory. Modern theories typically define operators that are applied to a current state to derive a future state; constructor theory skips all of that and says that some start -> finish pairs are allowed and some forbidden. Conservation of energy is very much of this model: it doesn't say why or how it's conserved, but from it you can figure out things like the velocity an object gains as it moves through a gravitational potential.

Looks more promising than everything-is-the-game-of-life theory.

I think in the long run they will appear complimentary. Particularly since the Ruliad is meant to represent the space of all possible entanglements.

This is also my feeling. Isn't cellular automata also about the possible and impossible transformations? I find both constructor theory and Wolfram project fascinating, refreshing and fun

It seems like a good sign if these new theories by these big brains (Wolfram/Deutsch) seem to have the same foundational concept of "sequence"?

Weinstein's doesn't seem to have the same basis

How come? If constructor theory is promising, then Wolfram's theory should be just as promising. Both are unproven and the ideas are not extremely different.

Please resist giving Steven Wolfram solitary credit for anything and everything related to cellular automata. He already does that just fine for himself. ;)

A few other important but less self-aggrandizing cellular automatists:

https://en.wikipedia.org/wiki/Von_Neumann_cellular_automaton

https://en.wikipedia.org/wiki/Edward_Fredkin

https://en.wikipedia.org/wiki/Norman_Margolus

https://en.wikipedia.org/wiki/Tommaso_Toffoli

https://en.wikipedia.org/wiki/John_Horton_Conway

https://en.wikipedia.org/wiki/Richard_Feynman

https://en.wikipedia.org/wiki/Alvy_Ray_Smith

https://en.wikipedia.org/wiki/Christopher_Langton

https://en.wikipedia.org/wiki/Bill_Gosper

https://en.wikipedia.org/wiki/Rudy_Rucker

https://en.wikipedia.org/wiki/John_Walker_(programmer)

https://en.wikipedia.org/wiki/Category:Cellular_automatists

I still don't understand how come Deutsch's theory is supposedly more promising than Wolfram's.

Maybe because it acknowledges the work of other people in the field. ;)

That's nice (no sarcasm intended), but - wrt. physics itself - absolutely irrelevant.

And I'm writing this because I hoped for an explanation, btw...

Not to forget this bit, trying to link physics and cellular automata in 1969!

https://en.wikipedia.org/wiki/Calculating_Space

Not to be all depressing like Marvin the Paranoid Android, but I happen to think "life" is overrated. ;) There are so many billions of other extremely interesting cellular automata rules besides "life" too, so don't stop once you get bored with life! ;)

https://www.youtube.com/watch?v=CAA67a2-Klk

The really interesting and efficient and deeply physics-related cellular automata rules are reversible (which the Game of Life isn't).

https://en.wikipedia.org/wiki/Reversible_computing

Reversible computing is any model of computation where the computational process, to some extent, is time-reversible. In a model of computation that uses deterministic transitions from one state of the abstract machine to another, a necessary condition for reversibility is that the relation of the mapping from states to their successors must be one-to-one. Reversible computing is a form of unconventional computing.

Due to the unitarity of quantum mechanics, quantum circuits are reversible, as long as they do not "collapse" the quantum states they operate on.

Good title for a C++ book ;)

Teach Yourself David Deutsch’s Constructor Theory in 21 Days

It looks like a physicist is re-inventing the axiomatic method, with possible/impossible as a dichotomy, instead of true/false.

That's programming language work for physical theory, not physics.

> That's programming language work for physical theory

Well, that would be insanely valuable since it would open up theoretical physics to machine-approaches (AI/ML). I know many are disgusted by this, but I'd love to see a PhysicalTheory-fold equivalent of Alpha-fold chew throw all modern theoretical physics and rank all theories on a possiblity-probability score or sh like that, since the amount of stuff a human would have to go through is already too much...

On paper, current AI/ML can allready do this with math and proofs. But it's still a grind, and doesn't produce pretty pictures like a GAN.

Actually, I don't see what's preventing Deutsch from formalizing his work in an allready existing proof assistant, like Lean or Coq.

wikipedia summary: https://en.wikipedia.org/wiki/Constructor_theory

This sounds a bit like rebuilding popular theories on a term-rewriting basis in order to open up more possibilities that are rubbing up against the limitations of the current modelling paradigm. Sounds interesting - I think term rewriting as a framework is quite conducive to creating high complexity models.

As a rank amateur in the field but a pro s/w developer, this sounds like describing the world as FSMs rather than fundamental algorithms. How off-course am I?

If I were to give a software analogy, I would say Deutsch's proposal is like trying to describe physics by specifying it in terms of the data types that the "code" of the physical equations satisfy. Instead of saying the "underlying implementation code" is what is fundamental in physics, he is suggesting that the types are fundamental. And the way to figure out what the code is doing is to figure out the types because the types are what ultimately constrain the code. His view would be something like every possible set of implementation code that satisfies the types is in fact instantiated in the multiverse. Therefore figuring out the fundamental layer of physics means figuring out the types because they are what define the multiverse.

Is this by the famous quantum computing researcher?

Yes

Isn't that a bit similar to Lagrangian mechanics ?

I loved “The Fabric of Reality”, but I fear (granted, from his Twitter, which can be less representative of his Good Ideas) Deutsch is at risk of going Dawkins.

Becoming one of the most famous public intellectuals of all time?

Becoming a selfish meme? ;)

Reply on news.ycombinator.com