The cosmic landscape
Peter Woit has blogged (and no doubt many others will also blog) about Leonard Susskind's new book The Cosmic Landscape: String Theory and the Illusion of Intelligent Design. Before I go any further I must confess that I have not yet read Susskind's book, so I will confine my comments to standard background material, and I will write later on about the specific details that Susskind includes in his book.
To summarise, Susskind says that string theory does not make a useful number of testable predictions, but it does predict a mind-bogglingly large number of vacua (each of which has its own laws of physics) so by random chance it can give rise to a universe in which the laws of physics are indeed as observed.
This last property is called the "cosmic landscape", because there is a landscape of alternative vacua, which allows the weak form of the anthropic principle to bear fruit. What this means is that we observe the universe to be exquisitely fine-tuned for our existence, because it is one of the mind-bogglingly large number of alternative vacua that happens to allow the development of observers like us. So there should be no big surprise that this fine-tuning is observed by us, because otherwise we would not be around to make the observations. The only universes that can be observed by their occupants are precisely the ones that are tuned in such a way that observer-occupants can exist in them, and for which the universe must therefore appear to be fine-tuned from their own point of view.
Note that the anthropic principle is not circular reasoning, because any system that can make measurements of its own behaviour (i.e. which has dynamics which allows its various parts to interact) must necessarily have certain correlations between its measuring parts (A) and its measured parts (B), so A will express "surprise" that B (i.e. what A observes) has certain properties that are (seemingly unexpectedly) related to the properties of A. This is a generalisation of what the phrase "fine tuning" means.
Another way of looking at this is using Bayes theorem, which provides a uniquely rigorous and consistent way of calculating what correlated entities "know" about each other. If the properties of the observer-occupants (A) are correlated with the laws of physics (B), then it follows using Bayes theorem that each can make predictions about the other. However, B => A is uncontroversial, whereas A => B is the weak anthropic principle. People who like B => A but not A => B are implicitly denying Bayes theorem, which means they are happy to reason inconsistently. Enough said?
Despite the logical consistency of the weak anthropic principle many physicists hate it, and denounce it as unscientific. Susskind is not one of these people; he accepts the "landscape" as a fact, and follows up its logical consequences where it is a physical framework for invoking the anthropic principle. However, he attracts the venom of those who hate the anthropic principle, because they perceive it as pulling the rug out from underneath science. This is because if the particular point which you occupy on the "landscape" has to be determined by experimental observation, rather than by theoretical prediction, then they claim that science has no predictive power, because the laws of physics are fixed by experiment rather than predicted by theory.
Just how arrogant can they get? Of course the laws of physics fixed by experiment rather than predicted by theory! That is what science is about, after all. The role of theory is to interrelate the results of experiments where these results are correlated with each other. That is still possible in a universe whose laws of physics are determined by experiment, which is how scientists have been working all along. The problem is that the goal (dream?) of string theorists is to assign a much larger role to theory than to experiment, so only a few basic properties (or perhaps no properties at all!) need to be determined by experiment, and then string theory is used to fill in the rest of the details. It is tough luck if the properties of the universe are such that that the predictive power of theory is smaller than string theorists would like it to be.
If the properties of the universe are correctly described by string theory (this may indeed be the case), and string theory predicts a whole "landscape" of alternative laws of physics, then the string theorists have to work with that, and they will have to measure what they cannot calculate. There is nothing wrong with that. It is called doing science, rather than doing philosophy. It seems to me that the string theory zeolots who do not like the "landscape", and who pour scorn on the weak anthropic principle, are trying to start a new type of philosophy that prefers only (or mainly) to calculate rather than to observe and calculate.
String theorists should get out more, and they should do science rather than philosophy. The weak version of the anthropic principle is science, but asserting that the laws of physics have to be derivable from first principles is philosophy.
posted by Stephen Luttrell @ 7:04 pm
8 Comments:
So there should be no big surprise that this fine-tuning is observed by us, because otherwise we would not be around to make the observations. The only universes that can be observed by their occupants are precisely the ones that are tuned in such a way that observer-occupants can exist in them, and for which the universe must therefore appear to be fine-tuned from their own point of view.
I don't think that's right, because it should be surprising to us that the multitude of anthropic coincidences are balanced between diametrically opposing runaway tendencies, where any sustained deviation would send conditions so far away from your wildest dreams for life that "life as we know it" is the only form of life that's possible.
I also think that it will eventually be proven that these are "ecobalances".
My argument was incomplete. I gave only the first step in the anthropic chain of reasoning.
Fine-tuning manifests itself in many ways, two of which are:
1. It can be anthropically selected fine-tuning, where you have an ensemble of slightly differently tuned systems, and only a very small subset of these has the desired properties.
2. It can also be feedback controlled fine-tuning, where a single system acquires dynamical properties that hold it stably at a sharply defined fixed point. Your ecobalances are an example of this.
If we believe the "landscape" argument, then the laws of physics are as in (1) above. As you go down to lower energies, with longer scale effective degrees of freedom, you typically get strongly non-linear dynamics such as in (2) above.
The full anthropic argument leading to us observers needs both (1) and (2) above to work simultaneously. So you need both good laws of physics and good ecobalances for the anthropic argument to work.
The full anthropic argument leading to us observers needs both (1) and (2) above to work simultaneously. So you need both good laws of physics and good ecobalances for the anthropic argument to work.
Only if we believe the "landscape" argument
I would suggest that the AP is only a tautology because it's incomplete because Dirac's Large Numbers Hypothesis was flawed because his cosmology was wrong.
I said in my original posting:
If the properties of the universe are correctly described by string theory (this may indeed be the case), and string theory predicts a whole "landscape" of alternative laws of physics, then the string theorists have to work with that, and they will have to measure what they cannot calculate.
so everything I said is conditioned on that "if ... then ...".
Yes, there is the logical possibility that the "landscape" is an illusion, and that string theory (or some other theory) may be able to derive the laws of physics from first principles.
However, if the "landscape" is real, then the weak anthropic principle has a significant effect on your reasoning.
I said in my original posting:
Another way of looking at this is using Bayes theorem, which provides a uniquely rigorous and consistent way of calculating what correlated entities "know" about each other. If the properties of the observer-occupants (A) are correlated with the laws of physics (B), then it follows using Bayes theorem that each can make predictions about the other.
This mutual relationship between A=>B and B=>A is indeed a "tautology" as you say, but that is good. In fact, it is necessary if you want to reason consistently. Being able to draw a logical inference B=>A about A (e.g. laws of physics) from B (e.g. our existence as observers) is unavoidably related to the inverse logical inference A=>B by Bayes theorem.
There are two extreme cases:
1. Weak Bayesian prior probability. The range of alternatives that A can have is large (e.g. a "landscape") then the B=>A inference has a strong influence on results, so the the anthropic principle has a significant effect in converting the prior probability to a posterior probability.
2. Strong Bayesian prior probability. There is no "landscape", so there is only a small number of alternatives, so the anthropic principle has little or no effect in converting the prior probability to a posterior probability.
The weak anthropic principle is a consequence of Bayes theorem, which expresses the desire to reason consistently.
Apologies. When compared to my original posting, I swapped round the definition of A and B in my comment above. However, this affects none of my arguments, because Bayes theorem is symmetrical with respect to interchanging A=>B and B=>A.
I think therefore I am a boundary condition - physics or philosophy?
Okay Steve, I didn't mean to disagree with what you said in context. I simply didn't quote your statement very well, but my only point was that your The full anthropic argument leading to us observers needs both (1) and (2) above to work simultaneously IF you believe that the landscape argument.
My suggestion about Dirac's Large Numbers Hypothesis was intended to imply that fixing his cosmological model would complete and clarify the principle without the need for number 1.
For example, if particle creation causes vacuum expansion via vacuum rarefaction, then gravity doesn't fall off as the universe expands, it actually increases in proportion to negative pressure, so the universe is held flat and stable as tension between the vacuum and ordinary matter increases instead.
This process will eventually compromise the integrity of the forces that bind the universe... and boom goes the causality problem… the horizon problem, the flatness problem, Matter/Anti-matter asymmetry problem, the cosmological constant problem... etc... but without need of an inflationary band-aid scenario.
I know that Supernovae, Black Holes, and us wee-little humans are the only known or expected sources for vacuum expansion in a universe that expands via the described form of matter generation, except that we are the MOST energy-efficient at it out of all of them.
So this defined need for these highly specialized dissipative structures fully justifies fine-tuning, while extending the principle to its biocentric form, where life will necessarily be as common as the universal scale need for it demands.
"I think therefore I am a boundary condition - physics or philosophy?"
You can use Bayes theorem to draw logical inferences about B from the state of A, even where the only causal influence is B=>A. Bayes theorem allows you to reason backwards. Not only that, but if you want to reason consistently, then you have to include such backwards reasoning in your framework.
Bayes theorem is the basic reason why the weak anthropic principle has to be so. The fact that you think does have an effect on the probability distribution of the laws of physics.
The thing that Bayes theorem manipulates is information about things. This is neither physics nor philosophy, but information processing. You could imagine implementing this processing using an ancillary physical device, but that is not really part of this discussion.
You don't even need to think to be a boundary condition. Any part of any system effectively acts as a boundary condition for the other parts of the system that it interacts with. Bayes theorem then tells you how you can invert this causal chain of events to make logical deductions about the boundary conditions that could have given rise to the observed response of the system.
Post a Comment
<< Home