Relational reasoning (Rovelli's RQM, Smolin etc)

1. Aug 1, 2009

Fra

There has been a few recent threads relating to the interpretation of QM, in particular the interpretation of probabilities and the RQM interpretations of Rovelli. I'd just like to highlight some from my point of view general constructing principles and ways of reasoning characterizing the relational interpretations but also (this is My main point) some differences in the various versions of this, and what status to assign to the SET of relations and their internal transformations (symmetries). And I will briefly argue that Rovelli's relational views, are only partly relational, and that this is related to a realist view of symmetry.

(Personally I would have chosen to place this thread in the beyond the standard model section, but since most of the related discussions has taken place here I'll put it here, and let the moderators move it if they wish.)

The general arguments here are also partly applicable to the discussion of background independence, where some argue for BI, but where there are different definitions or degrees of BI. It also applies by analogy to symmetry arguments. Symmetry, BI and relationality are closely related from the point of view of reasoning.

One of Rovelli's suggestions, in analogy with the classical theory of relativity is that information states are not objective, they are relative to the observer in direct analogy to the classical measurements are dependent upon the relative motion of the observer.

So far, it seems highly plausible.

But, one major difference is that at the time of classical theories there was alot of realist thinking. In particular is there realist view of the transformations groups that RELATE the diversity of observations. This is poincare transformations in SR and also diffeomorphism transformationsin GR.

These transformations, seen as properties of nature, are not subject to measurement in the direct sense. This is fine in realist classical physics, which is somehow realistic from construction.

I have however serious doubts about this way of reasoning in the higher standard of modern physics. Rovelli partially carries over this realist view of physical symmetries into QM and QG. This is to me a major speculation, and I don't think it's right.

Rovelli considers QM to be relational, but IMHO he is treating the symmetries connecting the observers in a realist way. This breaks IMO the idea behind a measurement theory, because some majors elements are put in by hand, without critical analysis.

I am fully in agreement with Rovelli's relational ideas, if it wasn't for this realist view of symmetry, which is also very clearly expressed in this papers of what's observable in QM and QG, and his papers of partial observables.

So what is the alternative route within a relational interpretation? One alternative is where inference of symmetries are subject to the same constraints as all other physical interations. This leads almsot to a circularity, which can instead be interpreted as an evolutionary model. This is why I think this alernative reasoning points towards the ideas of Lee Smolin, which is also relational, but in a very different way than Rovelli. Smolins critique to eternal timeless laws and symmetries are also rgumetns in this.

I think Rovelli is aiming for something TOO complete, but fixing the structure in the hierarchy of relations of relations at some level, and consider them in a realist way to represent eternal strucutre of reality.

So I just wanted to highlight that there are several different relational interpretations, and where I personally consider rovelli to not be "relational enough".

Rovelli says that all measurements are relative to the observer. I wish he would also say that the set of relations for the set of observers are ALSO relative to the observer.

But this would lead to circularity. I think he wants to avoid this complication. But this circularity could evolution of laws Smolin is sniffing on. Where hte LAWS are the ultimate form of STATE, and even these are relative!

In this sense, I like rovelli's starting point, but he doesn't go all the way. And the point where he flips to realism seems ambigous to me.

/Fredrik

2. Aug 1, 2009

ueit

I think that you should state clearly what the basic principles behind your proposed interpretation are. You cannot just state that everything evolves, because you cannot build a theory from there. What are the building blocks of this interpretation? Do you accept the objective existence of spacetime? If not, how is an observer defined? Etc.

I would also add that there is plenty experimental evidence for the existence of "timeless laws". For example the speed of light is always c regardless of what/where/when the observer is. The theory of relativity (which Einstein preffered to call "invariance") states that the laws of physics are the same in every reference frame. Is there any reason to deny this?

3. Aug 1, 2009

Fra

sidetrack on the relational interpretation + evolving laws

First I should say I didn't mean to focus on arguing for why my own ideas are right, I just wanted to add to the recent discussion here about relational views that there are different way of thinking about relational views an in particular the status of the relations, as an alternative way of reasoning.

Which path that will first lead us to solve some of the open problem noone knows, each one are entitle to make their own speculations.
It is not a priori obvious that you evaluate any reasons of mine I communicate to you "clear". What is plausible for me, need not be plausible for you. The fact that we all often disagree is not really unexpected. My speculations are my gain or my loss, which of course applies to everyone. But discussing different ideas are often interesting as it may give new idea.

But loosely speaking here are some of my personal ideas fwiw just from the top of my head. I'm working out some paper, that wil lbe published when time is right, but that's along way left. So this is information argumentation only.

a. Physical interactions in nature are more or less isomorphic to reasoning upon incomplete information.
- This is not a new idea, see for example the ideas of Ariel Caticha and ET Jaynes.

b. The scientific process is also sort of isomorphic to reasoning upong incompelte information.
- This is not a new idea either, see ET Jaynes in "probability thery - the logic of science" for a starter

c. Neither the scientific process nor physical interactions, nor reasoning upon incomplete information are deductive processes. It more closely resembles a game. This is the game of life, and we are all players.

d. Observers are players in the mentioned game. Players come and go, and also the rules of the game come and go.

e. The predictable part of the action of each player is based on available information, but there is also an undecidable part.

f. The translates into physics so that the action of any system is depending (normally probabilistically) upon the information encoded in the system itself. An implication here is that the behaviour or any system gets simpler and simpler, the lower the information capacity of the system.
In my view theory building is not a deductive proces. I think the correct conclusion is that there is no deductive scheme to build a correct theory.

But then that's not what I'm aiming for. The theory itself is encoded in observers in my view. Theories lives and dies with the observers.
Objective existence of what spacetime? 4D spacetime? Then no, I don't.
I think anything raising a question, or inquiry defines an observer.

I envision an observer to be an information structure, who is living a game. Observer-observer interactions are matter-matter interactions.

I think your question is about as hard as to define matter without spacetime - or howoto define spacetime w/o matter for - and interactions. They only make sense together. The emerge together.
There are plenty of evidence for very stable laws. That's a major difference.

I like a comment Smolins comment on this in a talk on "reality of time". He asked what is the meaning of eternal law, when it seems the universe is only say 14 billion years? So please define how it's possible to acquire timeless evidence by any process involving time? :)

I take the inference process seriously. And there is no conclusive evidence for eternal law. There is alot of support for stable laws. It's exactly this inference process I'm analysing. It will related to the problem of origin and reality of physical symmetries in a similar way.
In the view of emergent symmetries, the effective stability in the observable environment is no coincidence. They are a result of evolution.

Ariel Caticha's idea is that GR is a result of seeing physics as a reasoning on incomplete information, and that the logic of reason will imply the right equations.

Olaf Dreyer also thinks that GR is impled once you find the correct inside-view.

I don't buy 100% what they do, but both of them are examples of attempts in a related spirit.

So your examples aren't contradicted by what I have in mind. Instead they might help explain why the laws look like they do, and the origin of natures (current observed) symmetries.

/Fredrik

Last edited: Aug 1, 2009
4. Aug 1, 2009

ueit

Re: sidetrack on the relational interpretation + evolving laws

This is not a matter of plausibility. I just want to understand your ideas. While, at some intuitive level I think I get your point I cannot put those ideas in a logical construction so that the foundation of a physical theory may emerge. I will detail this below.

You speak about "incomplete" information. This presumes that a more complete information exists. What is the reason this information is not transmitted? Do you have in mind some type of locality that precludes instantaneous information transfer? In this case you should also define what do you mean by "distance" between observers. Is the unit of information, a bit, the basic building block of your theory?

I think the "scientific process" is a high level, emergent phenomenon that should not be approached in a discussion about a fundamental theory.

I understand this.

This I don't understand. There is no such thing as a game with ever changing rules. This is against the definition of a game. The reason communication is possible at all is because some basic rules never change. Can you give a simple example of a game whose rules change?

OK

This seems tautological.

OK, but then you have to start from somewhere. Is your universe like an infinite RAM memory or what?

If you agree with my proposal of a game running on an infinite RAM memory the next step is to somehow localize the observer. An entity requires some sort of time persistence, that means you also have to "attach" an objective existence to the time dimension.

The question may be hard but is justified. Indeed, you should only use notions that make sense in your theory. If you say that there is no time you are not allowed to speak about evolution because evolution presupposes time. If you speak about interactions you should explain what you mean by this word in the absence of space and so on.

I don't think it is such a "major difference". If the speed of light stays constant since 14 bln years and was never found to be different, under any circumstance, I think it is safe to assume that it is always the same. Of course, you cannot have a definite proof of that but the evidence for the contrary (the speed of light varies dependeng on each observer) is many orders of magnitude weaker. I can only ask why even contemplate such a possibility when you run in the face of so much evidence?

5. Aug 1, 2009

Fra

Re: sidetrack on the relational interpretation + evolving laws

Ok, that makes sense. Several of the questions you raise probably would require more discussion than what I planned for in this thread.

I'll get back later and try to comment more on your issues.

Personally I do have a "plan" on howto turn this into predictions. But clearly the ultimate "argument" is when I have some finished explicit solutions on the table. I am not at that stage yet.

/Fredrik

6. Aug 1, 2009

Fra

Reasoning upon incomplete information

I speak of incomplete information because it's a standard phrasing. The purpose is to make you associate to something existing to get on track to what I'm proposing.

However, what I am suggesting contains implicit a reconstruction of a new kind of information theory.

Reasoning incomplete information is simply a label for the situations which are very common indeed, where you are forced to make a choice of action, given some kind of goal, but you lack the information necessary to deduce which is the best choice. Thus the problem becomes howto make the best choice, given that the measure of what is best is not known.

Ie. it's the quest of the logic of guessing.

The problem of optimal inference, is what "rule of inference" is optimal? There are several ideas on this. There is bayesian reasoning, there is MAXENT reasoning etc.

I'm suggesting something that is neither, but related.

I'm suggesting that the rule of inference is evolving, and the optimatal rule is evolved. The physical manifestation of these rules are the microstructure of matter.

So there is no problem of defending what a particular rule is to be preferred, maxent rules as per some choice of entropy measure or bayeisan updates.

I think you see that there is no way I can give simple answer here, but hopefully you get the idea of direction. The purpose of the thread was just to note that there are several ideas in the relational direcition, not present a read solution.

My unit of information is distinguishability measure, which is isomorphic with a bit. More complex measure are constructed by interactions and evolution of the measures. At this point there is certainly no such thing as a continuum spacetime. Spacetime is also a part of the measure structure that emerges when the information complex (=the observer), is complex enough. A very simple observer (think below planck domain) can not _distinguish_ ordinary spacetime. What it does see, I don't know yet. But the same applies to the distinguishable interactions. The laws of physics (the laws of inference) becomes simpler as the complexity of the structures goto zero.

No. I am not advocating a hidden varible theory or any superdeterminism.

I'm not sure how to put it simple.

One reason is that information doesn't just exists, it needs a physical structure for encoding. When a given structure is saturated, further information can not be distinguished. IT would be misleading to say it exists from that point of view.

It could in a sense exists to other observers, but then we are mixing the views.

The only locality I assume is that the action of a given observer takes, instantly depends only on it's own information. Thus information existing elsewhere, doesn't affect the observers action.

Distance measures and ultimately spacetime are constructed in a information geometric way. The unit of space and time would then ultimately be related, since all measures has evolved from a common origin.

About time, it is defined from the uncertainty. There are two things, the direction of time, and the unit of time. The direction of time is always implicit in the uncertainty of the information state. There is ALWAYS a direction of time implicit, but it's not always sharp. Classical sharp time is one extreme, the other extreme is symmetric diffusion into all directions.

But to understand the direction of time, from the information structure the makeup of this structure needs to be sorted out. This is related to the thermodynamic time, but with the difference that each observer sees their own "thermodynamic type of time" since they all see different information structures. When two such observers try to communicate, it is implied in this an interaction.

Classification of interactions would be done from starting from the complexity zero, which have only somewhat trivial interactions, and then scale the entire construct as complexity goes up. At some point we will probably get a continuum like structure.

If the standard model, including spacetime dimensionality at some point comes out of this, then it's like I hope. IF not, I'm simply wrong.

Edit: one thing is sure, and that's that as the complexity/mass of the observers increase, new interactions and structures become _distinguishable_.

This is like an imagined inside view. Suppose you were a physical observer on subatomic scale, or subplanck scale. What COULD you see? what could you distinguish? And how does this affect your action?

/Fredrik

7. Aug 3, 2009

Fra

Re: sidetrack on the relational interpretation + evolving laws

I disagree. I see plenty of connections.

Your objection is I think correct from a FAPP point of view, but if you see that there are analogies between the laws of physics and the rules of inductive inference, you might see it differently.

The logic and philosophy of science, and the scientific method are IMO inseparable from the fundamental productions of it (ie physical laws).

Ok it depends if you take "game" litterly. There are definitions of games in game theory where there are set of rules that defined the game. But I'm thinking a little bit more general. The game theoretic analogies are often nice.

I'd say in most real games (not chess or poker!) the games are always subject to negotiation. Take society, see life as a game. The laws of society are certainly not fixed. The evolution of the rules are in fact part of the game. Each player acts as per their view of what the rules are. Sometimes, the players simply have different "opinons" on what the rules are - this leads to conflicts that "violates the game" as seen by them both.

It's not necessarily so that one of them is the bad guy. This is seen frequently in human conflicts on earth. Due to many reasons, traditions, religion etc, people have different opinon on what is right, and this leads to conflit, the outcome of the conflicts can be several. Sometimes some opinons of the rules are wiped out, sometimes a new negitoation takes place and all players revise their opinon.

It's this I mean with real games. Chess are static toy games, it's not what i'm talking about. Nevertheless say poker games are sometimes great analogies to QM, where each player (subsystem) acts upong not what is really out there, but to what information they have (ie. what cards they THINK the others have, and what strategies he THINKS the others have, and what he THINKS are the rules).

In most games, too much violation of the effective rules aren't possible as you will be destabilised by the environment. This leads to effective inertia and stability of rules/laws. So it is no conicidence that many laws are effectively SET. But this doesn't mean that are really set in stone, they might in fact be sloowly evolving, just like a stone is changing.

So I'm thinking of evolutionary games where the rules are changing.

/Fredrik

8. Aug 3, 2009

Fra

Re: sidetrack on the relational interpretation + evolving laws

I don't agree on that idea. I'm not sure how you mean.

In particular there is no objective view of the game. There are only inside views. An effective objectivity is emergent from selforganisation.

A chess game you can described from the birds view. You have a well defined state space etc and two players. This is not the case with reality I think.

I didn't say there is no time, I only say there is no objective time. But I mean this in a quite different way that rovelli or does. Rovelli thinks there are only subjective time, and that the subjective views are related by a god-given symmetry.

I think that there are only subjective time, and that the relation of subjective views can only be evaluated from the point of view of an observer, this means that there are only subjective symmetries, taking it one step further than rovelli.

To me the origin of time, stems from the uncertainty of defining the state of information. The state is unstable, and the expected "arrow of diffusion" is the arrow of time, and if you pick an arbitrary physical clock under the observers control, it's also follows that there is a maximum expected rate of change as measured by a physical clock. Ultimately I will relate this to the maximum signal propagation speed. This would be a information theoretic way to characterise and understand why there is a limit of communication, that doesn't need a priori a theory of electromagnetism, or even the normal space.

The symmetry transformations between subsystems/observers are emergent as they interact. Thus, without interaction there is no space. If you only have inside-views, as opposed to outside views, you do not need spacetime first and interaction later, the spacetime emerges as a picture from the inside view as the interactions proceed.

Because there is a difference to fundamental and effective. Most theoretical consturctions are extrapolating principles known to be true in a special case, to the general case. This isn't always wise.

Another reason for contemlpating this is that consistency of reasoning, requires me to question physical law, by the same logic that I question the position of an electron, or the moon.

A theorist interacting with nature, and building a theory, is conceptually analogous to an atom interacting with another atom, in order to find the most stable internal configuration.

There are too many coincidences here to think that there is no connection with the the physics of science and science of physics.

/Fredrik

9. Jun 15, 2010

John86

I don't know if this is the right section to post this !..

Spitting the ArXiv archives i found some interesting papers by olaf Dreyer, i'ts been a while since he last presented an new paper, maybe he is working on something fascinating..

Has anybody of you got some thoughts ?.