Tuesday, June 02, 2009

Hartle and Srednicki on xerographic distribution

Hartle and Srednicki wrote another interesting, short article explaining the correct reasoning that should be used instead of the flawed reasoning based on Boltzmann Brains and similar ghosts:
Science in a very large Universe (PDF)
Their main point is that what we're testing are not just physical theories (with their initial conditions for the whole system etc.) but physical theories (including the initial conditions) plus the answer to the questions:
Who are we? Where are we? Where have we come from?
Answers to these questions are unlikely to ever become a part of scientific theories because scientific theories objectively deal with the "whole systems" and they don't care "who are we".

In slightly more technical terms, we must always make assumptions about our location within the larger system (usually a very large universe). This information is usually not known "absolutely", so we must only determine it in terms of some "xerographic distribution".

What we're testing by experiments and the scientific process are therefore not just theories: we're testing the pairs
(theory T, xerographic distribution xi).
The main problem of the anthropic and the Boltzmann brain people is that they don't understand that "xi" itself is subject to experimental tests and refinements. Instead, they think that the anthropic anti-God has brought them a revelation (a dogma) what "xi" must forever be. And they think that the right distribution is the uniform measure, "xi=1/N" for each copy of the observed data: there are "N" of them.

Once again, Hartle and Srednicki explain that it's not the only possible assumption - in fact, in normal science, we almost always make other assumptions, especially if we're looking for predictive theories (that are likely to create "sharp peaks" for our future expectations of quantities). And the "typical" assumptions are not even well-defined because "N" is infinite, and additional assumptions about the measures at these infinite sets have to be made. That's what all the weird discussions about the details of distributive angels on the tip of a needle in eternal inflation are all about.

I essentially agree with everything they write but there is a point I would say differently. We surely intuitively want to look for predictive theories. But the predictivity of a theory is independent of its validity, and whenever it's possible to focus on arguments about the validity (or probability of validity) of a theory, it should be favored over arguments based on predictivity.

And theories that apparently lack predictivity - e.g. the Boltzmann Brain hypothesis - usually have a problem with the validity, too. They predict that it is very unlikely that the future observations will exhibit some order. Every time we observe something that makes sense, these theories become less likely relatively to theories involving order. I believe that the conclusions that can be made after we make observations - like the arguments from the previous sentence - should be enough to do science.

In principle, we don't need the a priori preference for theories that are likely to generate peaked predictions for the future, before we actually make these tests. Such peaked predictions are still likely to be proven wrong which is worse than to make unpeaked predictions. ;-)

The Boltzmann Brain hypotheses should already be expo-exponentially suppressed relatively to sane hypotheses. Since the people began to think about the world, they have made so many observations of the ordered real world that had to look like miracles from the Boltzmann Brain viewpoint that whatever the Boltzmann Brain prior were, they were already suppressed essentially to zero.
Needless to say, I agree that we are looking for predictive, testable, simple, beautiful, precisely formulable mathematically, economical in their assumptions, comprehensive, unifying, explanatory theories.
(I actually mildly disagree that we should be looking for "theories accessible to current intuition" because it sounds like a discrimination against conceptual breakthroughs and paradigm shifts such as quantum mechanics.) But many of these favored adjectives can actually be demystified and converted to the universal, key adjective - that these theories are more likely to be valid, by a certain rational argument based on inference.

So while I agree with all of their main messages, I think that the situation is somewhat less mysterious than they paint it.

And that's the memo.

Physics World cracks the pot

Physics World that used to be one of the last semi-popular journals that haven't been overrun by crackpots has joined the new "mainstream". Crackpot Lee Smolin wrote a rant called The unique universe that tries to fight not only against the analyses of the most general backgrounds that follow from the same known principles of physics, but also against the emergent character of spacetime.

The emergent character of space in quantum gravity has been established and the emergent nature of time, while supported by a much weaker set of well-defined equations, must morally be true as well, by the Lorentz symmetry that relates space and time. At any rate, the emergent time is one of the likely future revolutions. Smolin's desire to kill this future research program before it can actually begin shows that he is a very narrow-minded, destructive pseudo-intellectual with an extremely lousy physics intuition.

Once again, he also tries to resell his scientifically thoroughly falsified childish theory of fucking and mutating universes. These anti-selection mechanisms promoting pseudoscience and pseudo-scientists despite all the evidence spread from the media to realms that are ever closer to professional science. Physics World is already too serious a point to be hijacked by these morons.

Smolin also repeats his meme that the laws of physics can't hold forever. In his new version of solipsism, he insists that the laws of physics shouldn't hold before we're here - they should be as character-less and flexible as jellyfish kibitzers like himself. They should "only hold once". ;-)

So he not only wants to add ill-defined laws to science; he wants to eliminate all well-defined, objective, fixed laws. Given the fact that this can't possibly mean anything rational, I can only conclude that he's a nutcase.

Yellow cab drivers love physics; minority physicists hate it

"A Public Display of Attention" wrote a very revealing story about the discussion between the author, a physics blogger, and a yellow cab driver. He was absolutely thrilled by physics and string theory: she just hates it. This is not the first time I learned about it: she sent me quite a lot of mail explaining how she hated theoretical physics.

This counterintuitive arrangement of opinions of a paid professional and a taxi driver largely follows from some new and counterproductive "priorities" in the hiring process of recent years. Things just shouldn't work like that in the ideal world, and they didn't word like that years ago.

No comments:

Post a Comment