Universe, past, probable inference, Arch's argument

[Martin]

In another thread, Arch made this argument. there are challenging issues here (or so it seems to me). I want to take it out of context and focus on it in this separate thread. I think the word "create" here is a bad choice. the present derives from the past. conditions and events of 2008 derive from events of 2007. it gives the wrong idea to say that 2007 created 2008. but I don't want to worry about wording.

 

there is something more substantial. It is about what we can know.

 

Since we say that the Big Bang event created “the universe”, then any references to the universe would already have a meaning. Something not created by the Big Bang, would then be “outside this universe”. That said, can we prove that there is anything outside this universe? If we cannot interact in any way with what’s outside, then we will never have a scientific definition, because the statement isn't falsifiable. As I have already mentioned, WMAP may give an answer to this question. However, this also presents a challenge if it doesn’t.

Let’s say that we discover something controversial. To prove it requires an accelerator the diameter of the solar system. Assuming that Congress won’t fund our Totally Freaking Large Hadron Collider right away, what do we do in the mean time? Should we just ignore the discovery, or refuse to discuss it or even think about it? Maybe it has interesting applications to other branches of science. Should we only concern ourselves with the “easy science” and ignore anything else? Everything that we know today was once speculation. What should we do with current speculation until it gets to be falsifiable?

 

In some sense we cannot know the past. The dinosaur bones might have been buried in the ground by aliens with a sense of humor. We can only make probable inference about the past.

 

In some sense we do not interact with the past. the flow of information is one-way. there is no give-and-take. it acts on us but we can't act on it. does the past exist?

 

for a cosmologist, or any kind of astronomer really, the past consists of probable inference based on mathematical models. we don't see the bigbang event, but we see microwaves that we think date back to 380,000 years after the bigbang event.

 

a model that seems to fit the data rather well takes us back to year 380k, the earliest we can actually see, and then relying more on inference, back still further to year 0.

 

and now, with some new models, back still further to before year zero. Year zero is an arbitrary designation, it happens to be where the timescale is set.

 

In any case where you are using a dynamical model to make inferences about the past you naturally want to test the model by comparing what it says you should observe in the present with actual data. The situation is qualitatively the same with the models people are working on now, that go back before year zero, as it was with the old cosmological model that broke down exactly at that point.

 

In any case all you have is probable inference and comparison with observational data.

 

Ultimately there is a subjective and social element here---what is considered probable depends somewhat on expert consensus. Who is considered expert depends on things like custom and intitutions, and the consensus of other experts. There is no automatic machine that spits out the truth. Science is partly a human social process. We have to be practical and make do with our limitations. In a certain sense we are all part of this---it is a collective process based to some degree on consent, even though hierarchical. It all comes down to humans hopefully doing the best they can.

 

So now we are seeing a bunch of new cosmology models coming on, where they don't break down at year zero. the keep running back further in time and typically (with what is getting the most research attention) they find that before the BB event there is ordinary spacetime probably with ordinary crud in it, the main difference being that the space is contracting instead of expanding.

 

The language issue, then, is do you say that this pre-BB part of the past is "outside the universe", or do you not say that?

 

My inclination is not to exclude. Partly I am influenced by seeing that the people who are constructing and running these models are getting a lot of recognition honors and respect from their peers. Partly it is because I've studied the models---the new ones that don't break down at the BB-event---and they make better sense to me than the old one that breaks down at that point.

 

But I can hear Arch's argument. A key point is the idea that you might need a big expensive Collider to verify something. I think this is mistaken and that Colliders are unlikely to be important. The kind of instruments needed are much cheaper than accelerators and they are of new types.

 

Modern astrophysics instruments are both orbital and groundbased. There are groundbase detectors and imaging Cherenkov telescopes of sorts which did not exist when the first accelerators and colliders were built. Particularly interesting are the neutrino, gammaray, and cosmic ray instruments. Also the microwave background mapping satellites. But optical and infrared telescopes (both groundbase and orbital) are still pulling in new information. None of this stuff seems to be prohibitively costly, or reaching some kind of practical limit.

 

So I don't accept Arch's argument that simply because for the time being one is unlikely to build much bigger Colliders one should regard new cosmology models as untestable----and therefore regard portions of the inferred past as "outside the universe". The collider people don't have a monopoly on generating new physics info (they just talk as if they do sometimes )

 

On the other hand I can see the sense, with a little care one could try to define limits on what we can know, and argue that we cannot know beyond some definite point, and therefore we should consider anything past that point as "outside". If one could only make sure what that point is! I don't see any evidence that such a point is in sight, but if it were, then that kind of epistemological limit to the universe would probably make sense. Can't settle that one right now

Any comment?

Edited August 26, 2008 by Martin

[fredrik]

If I may participate, I couldn't help reflect on this...

 

On the other hand I can see the sense, with a little care one could try to define limits on what we can know, and argue that we cannot know beyond some definite point, and therefore we should consider anything past that point as "outside". If one could only make sure what that point is! I don't see any evidence that such a point is in sight, but if it were, then that kind of epistemological limit to the universe would probably make sense. Can't settle that one right now

Any comment?

 

These reflections reminds me of the relational nature of information, and perhaps we can draw a parallell here between some epistemological limit and a complexity bound regarding relations - could the complexity of an observer somehow defined the "resolution of the image" of the world around him, that lives in him? I think such a thought is plausible but as far as I know there is not yet a completely satisfactory model for this, and it's something I'm still thinking of myself.

 

One complication though is that even though it's plausible to picture an epistemological limit (due to the complexity of the reference (ie "the observer")), the observer is also subject to change and in particular the complexity of the reference can grow! (like a memory system can grow). So part of the equations here we must not only describe the world relative to a reference of bounded complexity, it must also describe the state of changes of both the image of the world(universe) and the reference.

 

I have always made an association here to gravity. Gravity could perhaps be interpreted as the attraction of complexity, and the logic of gravity would be the logic of this strengthening of images. Clearly a more complex/massive reference, plausibly provides a more confident "image of the world" than does a volatile one.

 

I think these question on epistemological limits will be understood better when we understand the evolution of informationsystems better. Information systems as in, where "matter" is the manifestation of information in general, and the state of matter makes up the state of information.

 

There is a lot of philosophical stuff here though, mixed in with foundational questions of mathematics and it's relation to physics, which is why I think so little progress has been made on this. It's difficult to analyze without falling into the trash bin, so it might be safer to leave this for others.

 

I think these are extremely interesting things, but that are too rarely discussed.

 

/Fredrik

[Arch2008]

On a more general level, what do we do with research into theories that cannot be proved…yet? I am told that it took more than twenty years to conceive, design, fund and construct the LHC. What if it takes a century for technology to come up with the hardware to earn M theory its surname? Should we just shelve discoveries until we can prove them? Envy is not scientific, but scientists with projects that are not funded are envious of scientists whose projects are. Ridicule for new theories is nothing new in science, but should a lack of immediate progress be the deciding factor in funding further research? Scientists are still human and ideas that develop via a network fraught with human frailty could be as wrong as anything else. What if Einstein had been persuaded by a well meaning peer to forget about relativity? Who should decide which discoveries become fact and why? In World War One, a politician remarked that”War is too important to be left to the generals.” Perhaps, theoretical research is too important to be left to the scientists.

[Martin]

What if Einstein had been persuaded by a well meaning peer to forget about relativity?...Perhaps, theoretical research is too important to be left to the scientists.

 

Einstein was concerned with testability and made sure the theory he formulated had testable predictions. Gen Rel was published in 1915 and tested by astronomical observation in 1919.

 

David Hilbert was also searching for Gen Rel. Basically they were in a race. Hilbert got to some part or parts of the theory neck-and-neck, maybe slightly ahead. The action of Gen Rel is called the Einstein-Hilbert action.

 

I have a hard time imagining your hypothetical case. What if....? I can't imagine how or why any peer would have tried to persuade E. to stop working on Gen Rel. He was obsessed with it. But if he had been run over by a truck then almost certainly someone else would have gotten it----very likely Hilbert (the top math boffin of his day.)

=====================

 

To respond overall, I'd say that theoretical work is cheap. A smart strategy is just to make sure they are well trained and give the good ones their head. Like hunting dogs.

 

There is an issue nowadays that in the USA the theory postdocs don't have enough freedom to work on what their instincts tell them to work on. The old guys control the money and jobs---it you don't keep in line your career is dead.

 

In Europe (and curiously enough Canada) the young researchers have more freedom. So much of the progress in quantum gravity is being made outside the USA. It's strange. The US used to be a leader in physics theory.

 

Anyway, with theory it's cheap so you just train them to be self-motivated independent creative researchers and let them go free to find what they want to work on. A young theorist will naturally want to construct theory which can be tested with available means, so he stands a chance of making an historical contribution.

 

What can cost big money is experiment. But fortunately the next big developments in theory will probably be in quantum gravity, which does not require expensive accelerators/colliders to test it. Instead it needs high energy astrophysics observation. Those instruments are comparatively cheap---they aren't peanuts but they are cheap compared with a collider.

 

In the present era, astro instruments (IceCube, GLAST, Auger...) are tending to find out more new physics per dollar. So the trend is to go slow on investing in another big collider beyond LHC and put more funds into new orbital and groundbased astro observation. New astro instruments typically deal with much higher energy processes than old-concept machines like LHC. An imaging atmospheric Cherenkov telescope (IACT) sees TeV photons. Auger sees protons with much higher energy than anything humans can make in a collider. The sky is the high-energy lab for future physics, nowadays at least.

 

Smart theorists will always be able to develop theory which is testable with the means at hand. It may take them awhile, but just give the good ones freedom to work on what grabs their interest and they will eventually (as they always have) produce theory you can test----can't say in advance with what type of instrument however.

 

You mentioned M-theory. Whatever happened to M-theory? You don't hear much about it these days. Apparently it was never formulated. So one doesn't know what M-theory is. It doesn't make predictions because so-far it isn't a theory, no equations. If you would like to see what is currently of interest to string theorists, they just had the big annual conference and the list of talks is online. I will get the link.

http://ph-dep-th.web.cern.ch/ph-dep-th/content2/workshops/strings2008/?site=content/talks.html

 

I am not as worried about the present course of physics theory and experiment as you seem to be. The US theory establishment is pretty screwed up (locked into overinvestment in string/M and run by an old agenda) but that is just one country and natural processes will eventually cure that (old guys die off ). Meanwhile exciting non-string theory progress in UK Europe Canada and some other countries. Competitive, evolutionary processes ultimately control what happens. A little bit like a "Free Market".

Edited August 27, 2008 by Martin

[Arch2008]

(Thanks Martin for the link, which I have added to my "favorites")

 

As a casual observer, it seems that prior to the scientific method being rigidly enforced to prove something, the methods used for research seem more appropriate to a Las Vegas racetrack (and they’re off!). For the Manhattan Project, the General in charge didn’t let the scientists critique each other. Each group had a specific mission and worked alone directly toward that goal. I know that this is not a good general (no pun intended) plan for research, but a Free Market gives us what we want, not necessarily what we need. If a hundred great minds get funding to work on a super spoon theory, is anyone working on forks? Again, whatever method is used, someone chooses what we get to know. I hope that they choose well.

[Martin]

... If a hundred great minds get funding to work on a super spoon theory, is anyone working on forks? Again, whatever method is used, someone chooses what we get to know. I hope that they choose well.

 

I like the depth you get to. Serious concerns here. In answer I cant emphasize too much the cheapness of theory research. You just pay the guys (and dames) modest salaries. Make sure they have office space, paper, pencils. And get to conferences now and then. The internet makes the whole theory research process a cinch.

 

From grad school days (mathematics) I remember being deeply impressed by the CULTURE of independent leading-edge theory research. It is learned partly from one's thesis advisor. It involves a mysterious intuition. There are blood-lines. Some people can trace their PhD thesis advisor back to Gauss, or back to Sommerfeld, by way of Heisenberg.

 

We don't have to think about this. the tradition is a living tradition. All we have to do is NOT INTERFERE. the guys are cheap. they love what they do. you just make sure they are well trained (in the tradition of chosing your own problem and getting into it) and then let them go.

 

The idea is, if you see that somebody is good, inventive, selfmotivated, driven on his own, able to recognize an important problem. If you have a good one you pay him whether or not he studies spoon theory.

 

then you will automatically get some that write spoon papers, and some that write fork papers, and some that write papers of a kind you didn't even think of!

 

this is one of the better things about human nature and the culture of Western Civilization since 1600 (Bacon, Kepler, Galileo ....) We have a lot to feel bad about or dissatisfied with but this is one of the good things. It works.

Well trained people find their own problems.

===============

 

I think the trouble with the US system now is that the older tradition has degraded down to where you hand out the money to theory research programs, not to people. There is a highlevel scrimmage over the budget and you have so and so much for string, so and so much for condensed, for dark energy, for phenomenology etc. Nothing is earmarked for independent bright postdocs chosing their own path to explore. (spoons, forks, or something else entirely that they maybe think up!)

 

My ideal is when Gauss recognizes something in Riemann and encourages him to give a lecture at Goetingen about something nobody had ever heard of. Riemann invents differentiable manifolds (a new model continuum) and makes that his first lecture. He gets the job.

===============

 

On the experimental side, I agree you need organized programs, with thousands of people. Administrators, big bucks. A different culture.

It is there that a kind of Free Market or Darwinian thing can operate, does operate whether we like it or not. Eventually the administrators figure out where they get the most bang (new physics) per buck.

 

It will have something to do with what theoretical predictions the theorists have put on the table, in the meantime.

 

Anyway I'm thinking out loud and talking too much. I like what you are concerned with, in your last few posts.

[Arch2008]

Thanks for the deep insights into the scientific process, Martin. Your posts don't leave me much to dislike, either.