Reinventing Science

August 8, 2015 1:55 PM

Science "explains things" in various ways. You can start with initial conditions, and apply laws of motion (classical kinematics). Or you can predict things via evolving probabilities (quantum mechanics). Or you can find emergent laws (thermodynamics). Or ... - There are many different modes of explanation. Recently, David Deutsch invented a new one: Constructor Theory.

You say what is, and what is not, possible. And take things from there. The motivation seems to be to unify physics and computation:

[...S]tudying the universal quantum computer is the same thing as studying every other physical object. It contains all possible motions of all possible physical objects within its own possible diversity. [...] But then I realized that that isn't quite true, and there's an important gap in that connection.

In addition to information (first link), this new approach has been applied to life itself.

[references; previously; more]

posted by andrewcooke (22 comments total)

Sufficiently advanced science is indistinguishable from crackpottery (at least if you're me).
posted by miyabo at 4:13 PM on August 8, 2015 [4 favorites]

If physics is finally catching up with biology, does this mean we ll have a new kind of bomb soon?
posted by eustatic at 4:45 PM on August 8, 2015 [1 favorite]

crackpottery
yeah, me too. i wondered whether to post it, but i think it might be important. david deutsch is "the" guy behind quantum computing - the first paper he wrote pretty much invented the field. and there's this feeling that somehow information / computing / quantum mechanics / physics are deeply linked. at the same time, the standard attempt to connect quantum mechanics to "the rest of physics" has failedstalled. so there's maybe the hope that a "new kind of science" makes sense, although really this is much more about uniting quantum mechanics with "normal intuitive physics" than with relativity, at least at this point. and i am already out of my depth and bullshitting...
posted by andrewcooke at 4:47 PM on August 8, 2015

From the Scientific American article at the very last link:

According to constructor theory, the most fundamental components of reality are entities—“constructors”—that perform particular tasks, accompanied by a set of laws that define which tasks are actually possible for a constructor to carry out. For instance, a kettle with a power supply can serve as a constructor that can perform the task of heating water. “The language of constructor theory gives a natural way to describe the most fundamental principles that must be obeyed by all subsidiary theories, like conservation of energy,” explains Chiara Marletto, a quantum physicist also at Oxford, who co-authored the new paper. “You simply say that the task of creating energy from nothing is impossible.”

Something sorta smells a little bit like type theory in there...
posted by tss at 4:52 PM on August 8, 2015 [9 favorites]

sorry for threadsitting, but i forgot to add this to the post - quantum computing since democritus is the book i've just been reading that started me looking at deutsch. anyone interested in this, may also be interested in that BUT wow there is a lot of computation complexity stuff in there - i think i understood maybe half the points he made and at times was just skipping pages until i found something more understandable. it's not a light read by any stretch of the imagination.
posted by andrewcooke at 4:58 PM on August 8, 2015

This is maybe a risky basis on which to assign trust, but I read Deutsch's "Fabric of Reality" several years ago, and on that basis I extend him a lot of credibility. That's a book that puts forward some huge ideas, ideas that will take your breath away, but it roots all of those ideas in descriptions of physical processes that are both (a) lucidly explained and (b) free of any hint of New Age style Woo. It's easily my favorite book of popular science, and I'm fairly confident that it might be one of the more important books that I've read from a philosophical standpoint as well.

The last couple of times I've seen his name come up, it's been in the context of his actual scientific work (as opposed to popularization-of-science) and I've therefore been unable to follow it with any certainty that I'm truly understanding it, but I'd love for somebody who does understand to explain it to me, because I'm eager to know anything at all about what the guy is working on.
posted by Ipsifendus at 5:14 PM on August 8, 2015 [6 favorites]

So he's a Nietzschean? That would be the laugh to end all laughs, if it turned out Nietzsche was right about the fundamental nature of the universe.
posted by howfar at 5:42 PM on August 8, 2015 [2 favorites]

paging physicsmatt

background reading: infromation theory and "it from bit", wheeler's spiel
posted by lalochezia at 5:47 PM on August 8, 2015

Irritation derail, but the medium piece keeps saying "these guys" when Deutsch's co-author is a woman. Is guys gender neutral now?
posted by DangerIsMyMiddleName at 7:29 PM on August 8, 2015 [4 favorites]

That was a peeve of mine when reading.

"Guys" *can* be gender neutral (where we grew up it was a familial/neutral collective term: "Hey youse guys!").

But in this particular instance it has a very distinct feel of meaning 2 males as it's not directly addressing them in the same manner.
posted by symbioid at 7:44 PM on August 8, 2015 [3 favorites]

Especially when misogyny in science is such a persistent problem, it would be great not to default to that, which becomes sort of invisibilizing even if unintentional.
posted by Miko at 8:20 PM on August 8, 2015 [2 favorites]

the medium piece

It's just a random blog that happens to be on Medium.com. Which makes it really hard to tell if it's reputable or not.
posted by miyabo at 8:26 PM on August 8, 2015 [1 favorite]

Here is a Scott Aaronson post related to constructor theory from last year that brings in cellular automata as well, and here is the original paper.

Personally, all of this is way out of my depth, but I love trying to understand the deep and mysterious relationship between computation, information, and physical reality -- and this seems like a more reasonable attempt to do so than Wolfram's.

(Also, that blog is perfectly reputable, it belongs to the people who run arxiv.org and is well-regarded in the physics community AFAICT.)
posted by goingonit at 8:30 PM on August 8, 2015 [4 favorites]

You say what is, and what is not, possible. And take things from there.

Oh hell. Theoretical physicists have always been doing that.
posted by Twang at 12:32 AM on August 9, 2015

Physics is not the theory of everything, and there are certain things you can't make it explain. Information, in the non-Shannon sense that involves meaning, and is fully substrate-independent, is an example. It doesn't matter how niftily you can handle physical situations because the assignment between physical states and meanings is arbitrary. Anything can in principle mean anything.

I speak with due hesitation, because I've found that the more you read of Deutsch's own words the more sense the 'crackpot' stuff makes. But here I really think he's catastrophically missed the point.
posted by Segundus at 4:05 AM on August 9, 2015 [1 favorite]

But Deutsch isn't talking about any "theory of everything", which would be logically redundant. That's not what his constructor theory is about. Second, based on the article I think he would disagree with your counterexample because an assertion "meaning is arbitrary" precisely entails those issues about solipsism and evil which he also raises in his article.
posted by polymodus at 4:14 AM on August 9, 2015

Perhaps I'm missing something, but this doesn't sound so very different to the "if it's not forbidden, it's possible" which drives a lot of physics these days anyway, certainly in quantum theory. Or "Here are things, here are laws, combine the two to find out what can happen".

Also, I don't particularly understand if there's a difference between information and state. If you take a thing, it necessarily has a state. That state may appear differently depending on how, when and what it interacts with, but the combination of what the thing is and how it is is all that there is about the thing, surely? More information about it - say, where it's been or what caused it to be - has nothing to do with its instantaneous state, but with other knowledge from elsewhere in space-time, derived from other observations. but that other knowledge has to be kept by the states of other things. A photon doesn't know (ie, it contains no information in its state) what sort of atom performed the electronic event that brought it into existence, nor does it know how long ago that was (a photon has no time or distance in its frame, despite EM waves having phase. I'm never sure whether that's incomparably weird or totally banal).

We may know that a photon of such-and-such an energy is produced by an excited sodium atom, even that under the circumstances of observation that's exactly where that photon originated, and that it's vastly unlikely that this photon was produced by a cold atom of something else. But by itself, encountered whizzing merrily along, you can't say that the photon wasn't produced by a number of other events, even though many of them are unlikely. Does thinking about information as being different from state help?

So you have constructors operating on things (and presumably states) and this defines physics in a useful way... how does this change the way we think about these things anyway? Has it moved on from Newton's introduction to the third volume of Principia - "It remains that, from the same principles, I now demonstrate the frame of the system of the world."?
posted by Devonian at 4:47 AM on August 9, 2015

But Deutsch isn't talking about any "theory of everything"

Not in quite the sense some physicists use that phrase, but if anything his ambitions are even wider. At the end he seems to want it to go beyond the empirical and gobble up philosophy and presumably maths. If you don't think physics explains everything, I don't know why you'd think it explains meaning.

I think he would disagree with your counterexample

Of course he would (did I somehow come over as trying to agree with him?), but that's because he's wrong. If you doubt that the assignment of meanings is arbitrary: by 'VBNM' I mean Deutsch.

VBNM is mistaken.

The charge of solipsism is rhetorical, too poorly grounded to even address.
posted by Segundus at 5:22 AM on August 9, 2015 [1 favorite]

it from bit - the paper.

what does VBNF stand for?
posted by andrewcooke at 9:35 AM on August 9, 2015

Yeah, like tss I’m getting a strong category theory / homotopy type theory vibe from this work. Which means it’s either full of deep insights or a large pile of mathematical onanism & you’ll never be quite sure which one :)
posted by pharm at 10:01 AM on August 9, 2015 [1 favorite]

If you doubt that the assignment of meanings is arbitrary: by 'VBNM' I mean Deutsch.

See, but I have no idea what arbitrary even means there. Does that make my point clear?
posted by polymodus at 1:49 AM on August 10, 2015

Segundus: Physics is not the theory of everything, and there are certain things you can't make it explain. Information, in the non-Shannon sense that involves meaning, and is fully substrate-independent, is an example.

He explicitly, very clearly states that this is not the definition of information that he is using, so your arguments are irrelevant to his point.
posted by IAmBroom at 3:02 PM on September 1, 2015