In a previous blog, I had analyzed Sir Roger Penrose's new cosmology (CCC) as a reworking of Sir Fred Hoyle's "Steady State Theory". And of course, Sir Fred's is a reworking of the Greek model of Democritus, and one of the three pillars of materialism. Also in the news is Hawking's attempt to make the Creator unnecessary with a different kludge, invoking an infinite Monte Carlo casino of universes. Since Sir Roger is not a theist, it may be that like Sir Fred, he finds a certain attraction to recovering an infinite universe that makes a Creator unnecessary. Unfortunately, it looks as if he's proven the opposite. But first let's look at his theory.

After taking some withering abuse for his novel theory, Sir Roger has found some data. This, as you know, is the gold-standard for theory, that it predict some data. This blog takes a hard look at the prediction, and attempts a critique, intended more for philosophers than physicists, but a critique nonetheless.

The Data

The data come from a satellite that has been measuring the temperature of the night sky (otherwise known as the Cosmic Microwave Background (CMB) radiation) with five digit precision for over 7 years. There's two ways to do this. One either measures the temperature absolutely, say, with a liquid helium cooled thermometer, or one measures the temperature relatively, by comparing two spots on the sky. The Cosmic Background Explorer (COBE) used the first method, and its successor, the Wilkinson Microwave Anisotropy Probe (WMAP) mission utilized the second. Comparison of the data sets show the huge advances in instrumentation between COBE and WMAP, with WMAP being both more accurate and more resolved.

Now if there were images in the WMAP data, if there were lines, circles, or random doodles, you can be sure it would have been discussed by now. Sir Roger doesn't look for these, but rather to look for changes in the noise, in the scatter of the data. Specifically, he picks a point (one of 10,885) and then maps the sky-noise into a bin labelled with the distance from that point. This turns a 2-D noise map of the sky into a 1-D line of bins, which he quickly scans for anything interesting. Then he picks another spot on the sky and does it again, looking for is a 1-D line that has big dips or peaks in it, which corresponds to a circle in 2-D space around that center.

Just to be sure he's found something real, he does this for another sky-temperature experiment, BOOMERang, which was lofted up on a balloon in the Antarctic. He claims they are two independent measures, which is a strong argument for a real measurement. The balloon experiment only measured a small patch of the sky visible from the southern hemisphere, and it may have been cross-calibrated with WMAP, so I am not sure it is quite as independent as Sir Roger claims.

Even if independent instruments see the same thing, it may not be real, since it may be an artifact of the technique. Both WMAP and BOOMERanG used differential or relative temperature differences to find a signal that is the difference between two points. Now imagine that one of the points is unusual, say, an active galactic nuclei (AGN) that is beaming microwaves toward the earth. The AGN signal can be modelled and removed from the data set, which is trying to take the temperature of the empty sky rather than stars and galaxies, but the variability of the AGN signal is much harder to model, and perhaps can never be removed from the data set. Now when large numbers are subtracted, the first half-dozen or so significant digits are eliminated, and the insignificant digits retained and moved into the "significant digit" column. In other words, taking differences amplifies the noise, and circles are just what you'd expect if one of the two points had lots of noise.

In most astrophysics papers, which are limited to a handful of photons, there is a big section on "systematic errors", which are errors that aren't "noisy" but "directed", pushing the result in the same direction. In these data papers, such non-trivial errors are carefully tested on the data set. Sir Roger does not have this section in his paper.  Instead he does two, relatively easy things to replace it: he claims BOOMERanG support, and he claims that random noise does not reproduce this effect. As we said earlier, the independent support he claims may not really be independent (nor does he show side-by-side images for us to compare), and the random noise test is useless because it hasn't been generated with the same data-processing algorithms used in WMAP. That is, if he could get hold of the original WMAP raw data, replace it with noise, and process it to see the effect, then I'd be impressed.

The Predictions

Lots of people have been combing through the WMAP data over the past 7 years. Since the coverage was complete after 3 years, the subsequent years have been used to refine the measurements, bring down the error bars and reduce the noise. One of the first confirmations was that there were bright-dim regions in the data corresponding to "acoustic waves" from the Big Bang.  Furthermore, these acoustic modes were predicted by inflationary cosmology, confirming inflationary models.

Sir Roger's model has no inflation in it, and in fact, inflation should erase all circles that are wider than 13.7 billion light-years. Therefore he argues that the confirmation of his prediction and the mere presence of these structures makes inflation wrong.

Now I have other reasons for disliking inflation, but it seems curious that two contrary predictions can both be valid, while invalidating each other. Surely, "prediction" is the gold-standard for a theory, for when Einstein's 1915 theory predicted curvature of light, and Eddington's 1919 eclipse expedition found it, Einstein became an overnight sensation. Why is this not true for "inflation" or Sir Roger's "CCC" model?

We've looked at the ambiguity of data analysis for this paper, but similar problems beset the inflationary models, which have been modified at least 3 times already. Even then, there are multiple "dials" or free parameters that can bring agreement between the data and the models, so not all predictions are created equal. Einstein's prediction had no dials, and did not have to be reformulated when the data came in. Both inflation and Sir Roger's do. Thus the importance or significance of a prediction depends a lot on how constrained it is. This is generally known as Occam's razor, and mathematically expressed in Bayesian model fitting as the "Ockham factor".

Neither Sir Roger nor the conventional inflationary models have made solid predictions. At best, they can be tweaked to some, but not all aspects of the data. They possess far too many adjustable parameters drawn from some sort of metaphysical repository of models that is itself questionable. Sir Roger attempts to address the metaphysical problem with Creation, and inflationary models address the metaphysical problem of contingency. Neither is particularly convincing. (And a closer look at CCC theory raise more questions about the infinite pressure of dark energy, the rescaling time, the supposed decay of protons and electrons, etc.)

Which may mean that it is time to assess the situation. Some of the best minds of two generations have addressed cosmology since the discovery of the 3-degree blackbody CMB radiation and not found solutions to the twin problems of a necessary Creator and a necessary Design. Given their failure, and I do mean abject failure to find anything beyond a metaphysical black box with lots of dials, I would suggest that we claim support for the contrary. All of them have put the cart before the horse, trying to create a no-beginning-no-design theory by massaging the data, post-dicting their data to support their theory, and still have not succeeded, thereby proving the very thing they did not want to prove.

That is, if the CMB, discovered by Robert Wilson and Arno Penzias, implies a creation theory (T) that was designed, CMB --> T, while Penrose along with conventional inflationary modelers prefer a non-created, non-design theory (!T where "!" means "not") that they attempt to defend with other data, d. Then the failure of their inference ! ( !T --> d ) gives the logical result, !d --> T. 

Sir Roger is presenting evidence for an intelligently designed creation.