Monday, August 20, 2007

Cosmology under attack, too

It has recently become a fashionable sport for people who have never contributed anything significant to science - and most likely, they never will - to use media and populism to sling mud at the key results of science as we know it in 2007 in order to replace serious science by an irrational media-driven hysteria, at least in the eyes of the public, and to erase the difference between the scientists who are doing or have done something serious on one side and themselves on the other side.

Paper can withstand anything and ignorant enough people are ready to believe anything, too, and it is in fact easier for them to swallow an emotionally loaded untrue cliché than a difficult technical argument. So this is quite a good business for these folks. Modern cosmology is slowly joining the list of fields that are under attack. The only difference from high-energy theoretical physics is that I expect that the hysteria won't catch up in the case of cosmology because cosmologists kind of enjoy activities related to P.R.

Michael J. Disney, a retired astrophysicist, wrote an article for American Scientist,
Modern cosmology: science or folktale?
He starts with a sequence of hostile comments about the money flowing to cosmology. The subtitle expresses one of the main assertions of the article: current cosmological theory rests on a disturbingly small number of independent observations.

The similarity with the proclamations by people like Woit can't be more obvious. Let us first look at some of his opinions and terminology.


In the text, cosmic inflation is called a "vague conceptual solution", the Lambda CDM theory is "the currently fashionable concordance model of cosmology". Big Bang cosmology "is not a single theory" while dark matter and dark energy are "insubstantial notions". Also, "modern cosmology has at best very flimsy observational support". He even says that "cosmology has always had such a negative significance, in the sense that it has always had fewer observations than free parameters, though cosmologists are strangely reluctant to admit it."

An expanding Einstein model was elegant but "it has since run into serious difficulties". He ends up with the thesis that "Acceptance of the current myth, if myth it is, could likewise hold up progress in cosmology for generations to come".

Are these statements based on actual science? Maybe the kind of science used in Disneyland but certainly not the science used by serious researchers.


You may read it yourself: you will see long kilobytes of vitriolic and untrue statements that are clearly not addressed to people who have any idea about the field. Disney says very openly that "his approach should be appealing to nonspecialists, who otherwise would have little option but to believe experts who may be far too committed to supply objective advice." Why doesn't he try to convince experts? Because he must know very well that every expert knows why his text is complete junk. But what should the laymen do? Indeed, I think that if people can't learn the technology themselves, they should better believe experts rather than loud non-experts with a personal agenda. I wonder when "supply of objective advice" exactly became an insult.

What he wants to do is to brainwash uninformed people who are unable to figure out why his conclusions are completely absurd, without any need to offer any technical evidence for these statements whatsoever. What a convenient strategy.

Progress in cosmology

Cosmology, the science about the gross structure and history of the whole Universe, has been exciting and mysterious for millenia. A long time ago, the boundaries between science, myths, and religion were somewhat fuzzy. The situation started to change in the 20th century, especially because of Einstein's discovery of general relativity. Because of many observational breakthroughs in the last decade, cosmology has recently become a precision science analogous to particle physics. Many quantities, including the age of the Universe, are measured at one-percent level. Huge amounts of data agree with a simple theory.

How much data do we have?

Disney's statement that the amount of data we have to check the cosmological statement is lower than the number of parameters in the theory is absurd. But how absurd is it? For example, look at this picture from WMAP. It has more than one million pixels of data that can be processed in many ways to obtain hundreds of overall, qualitative numbers such as the spherical harmonics. Each of these numbers must agree with the right cosmological model and the error can't be too high. They do agree.

WMAP is not the only source of the data and constraints that the cosmologists use. They observe billions of stars and other celestial bodies to check the model and deduce that something like dark matter must exist. They observe billions of other galaxies and their relative motion to deduce that something like dark energy must exist, too. Theorists tell them that the cosmological constant is the only known plausible model for dark energy with the right equation of state and most cosmologists simply accept it: it doesn't influence their work too much anyway.

Some of their observations are made during collisions or explosions that offer a huge amount of additional information that constrains the theory. The same conclusion may be deduced from a very large number of these stars and galaxies. The Universe is huge and rich and it offers a diverse spectrum of objects and events to study. Many of them can be measured and have been measured.

Whoever really thinks that the amount of data we can extract from the Universe is small or even smaller than the number of parameters of the state-of-the-art cosmological model (roughly 17 parameters) is senile beyond imagination.

Expanding Universe vs global warming

Let me offer you a comparison with a subject that is heavily covered by this blog: global warming. According to this theory, the world is getting warmer and everyone is even supposed to accept an explanation based on one particular effect whose importance is a purely speculative issue. According to cosmology, the Universe is expanding and cooling down: Disney questions even this statement.

How does the evidence supporting the models compare in these two cases? In the case of global warming, you can't figure out that global warming exists if you look at individual stations, countries, or continents. For example, there has been no noticable trend in the U.S. temperature record. The local data look pretty much random. The only way how you can deduce a trend in the last 100 years is to take the global average. What you obtain is one number as a function of time. Global warming is not a feature of individual observations at different stations; it is a feature of one number that changes quite chaotically with time but if you wish, you may interpret this chaos as being superposed with a warming trend.

On the other hand, we can observe tens of billions galaxies and other celestial bodies and for each of them, the link between the red shift (relative velocity) and their distance from us is exactly as predicted from the expansion. It's like if you had 10 billion weather stations on the Earth and each single one of them would imply that the warming trend is 0.62 Celsius degrees per century.

In particle physics, we take the uniformity of the Universe and the universality of its laws and constants for granted. In the high-energy context, the value of the cosmological constant is one number, one parameter entering our theories, regardless of the place where you measure it. But when one actually works on cosmology and measures the cosmological constant via different kinds of galaxies or supernovae, these objects simply can't be counted as one measurement. They are many measurements and because many of them agree with the nonzero value of the cosmological constant or something that is effectively indistinguishable, it means that we have strong evidence for our conclusions about the cosmological equations and parameters.

Our certainty about the fact that the Universe expands exceeds any certainty about the very *existence* of global warming by ten orders of magnitude, to say the least, as measured by the small probability that the basic theory could be wrong. Every rational person knows that it is ludicrous to even compare these two cases because the basic facts about the Universe such as its expansion are directly and pretty accurately seen almost everywhere while global warming is a speculation based on one lonely and wiggly function of time. But with the help of media, it is not hard to create the atmosphere in the society in which global warming is as certain as the expansion of the Universe if not more so. In fact, it is politically correct to say that global warming is a fact just like it may become politically correct to say that the Universe is perhaps not expanding and the "deniers" of the cosmic expansion should be celebrated.

This combination of facts is no small unfairness: it's an example of a gigantic madness. A comparison of these situations shows that the public discourse about scientifically loaded questions is entirely irrational.

Elegance of the theory

Is the current cosmological model elegant? Well, it depends how you measure it and what you compare it with. It is pretty cool but it is not infinitely elegant. Surely, many of its aspects came as a surprise. But if something is surprising, it is not necessarily inelegant.

Various terms in the equations - such as the cosmological constant or the density of dark matter - are phenomenological in character, indeed. Some of us would always prefer to understand what is the deepest possible way to explain the origin of these entities. And of course, people are making progress in this direction, too. But I would like to stress that this is not really a job for cosmologists; it is a job for high-energy theorists.

Cosmologists are satisfied with a low-energy effective description of the Universe at the cosmological distances and their approach is fully legitimate. From their viewpoint, the agreement of the state-of-the-art model with observations is excellent. Their measurements and calculations imply that what they observe must be explained by the theories we have or something that gives virtually indistinguishable predictions.

The microscopic and conceptual details of a deeper theory could become very different if high-energy theoretical physics makes some progress in this question. But it won't be too great a deal for cosmologists because the approximate theory at the level they're interested in has already been established. Even if high-energy theorists suddenly use a very different set of concepts to describe what's going on in cosmology, it is likely that cosmologists will continue to use the current concepts and equations - simply because they're simple enough and they demonstrably work.

Another thing that Disney criticizes is the fact that the existing cosmological model had to be edited and re-adjusted to agree with new observations. Indeed, it is the case. And indeed, it is plausible that more editing will be done and will have to be done in the future. Well, this is how science works. People are usually not ingenious enough to guess the complete theory of everything at the very beginning. But this trivial sociological observation has nothing to do with the question whether their theory is safely verified by the data. Sociological observations can never be used to make arguments about natural sciences.

What's important is that today, we understand that the cosmological constant could always have been nonzero. In the recipe how the effective field theories are constructed today, cosmological constant is one of the allowed terms. In fact, what is the real mystery is why it is so small. In the same way, it is clear - and should have always been clear - that not all matter in the Universe must be visible. These facts are not contrived in any way if they are looked at scientifically instead of sociologically. The only thing that they show is that the Universe is not a trivial, naive system whose properties can be correctly predicted by people with no experience. Such an unsurprising conclusion is very different from the conclusion that there is something wrong about the theory used to explain the observations.

How much data do we need?

Finally, I want to say that we have far more data in cosmology than we need to verify models such as the Lambda CDM model. Let me put this counting into a historical perspective.

Albert Einstein may be credited with the discovery of the most important theoretical insight that has changed our understanding of the Universe at the longest distance scales and time scales: general relativity. How much data did he need to obtain his conclusions? He only needed special relativity, a theory that was itself found by pure theoretical reasoning, and the equivalence principle i.e. the equivalence of gravitational and inertial masses, a footnote observation that was well-known to Newton who had no deeper explanation of it.

Everything else followed by pure thought even though Einstein needed 10 years to complete his masterpiece. Surely, once we already know the final result, it is easy to reproduce his reasoning in one lecture on general relativity and erase the "unnecessary" mistakes in the reasoning that Einstein had to make before he saw the light.

Could he be sure that his theory was correct? Well, he was proposing temporary theories that were wrong and could be easily falsified - even though Einstein as the pioneer often needed years to falsify them. But the final form of general relativity was a settled thing once it was carefully checked that it reproduces all verified previous theories and once a single correct prediction that goes beyond them - more concretely, the perihelion precession of Mercury - was found to lead to the same result. I am certain that Einstein was certain about his theory once this one highly non-trivial number worked, and he was right. It was very rational for him to feel certain.

Many other people need much more evidence than that because they haven't really spent time with analyzing possible theories. Consequently, they are not able to realistically appreciate how strong constraints the known observations, the known principles, and additional corrections that go beyond the known theories place upon the new theory. These constraints are huge. Cosmology is not a theory about a complete chaos where one can decide to "see" an additional "signal" in the chaos. When you measure the right quantities, virtually everything is a nearly pure signal - sometimes a very accurate signal - and we measure millions of such signals.

That's a very different situation from global warming where virtually all data may be treated as "noise" and one is free to cherry-pick aspects of this noise that can be interpreted as an additional "signal" and to propose somewhat arbitrary explanations for such a "signal". The higher signal-to-noise ratio you have, the more seriously you must treat experimental data and the more certain you may become that a theory is right or wrong.

I am sure that in 2050, when people have a deeper understanding of the equations of cosmology and their origin, they will explain why this picture is correct while using less than 10% of the observations that we find "crucial" to support these theories today. There's nothing wrong with redundant data especially if they don't cost too much money but there's nothing essential about them either. The future presentations will rely on detailed experiments much less simply because their theoretical framework will be more robust and interconnected than ours, just like our theoretical framework is more robust than the framework used 50 or 100 years ago. And there will also be people who won't appreciate how much information is actually stored in different kinds of data and who will be writing similarly insane texts as likes of Peter Woit and Michael Disney write today.

And that's the memo.

1 comment:

  1. Just this morning I read Disney's article at breakfast and shared it with my partner, cosmologist Ken Olum (Tufts U.). He told me Disney was all wet, and referred me to this blog post. I'm very grateful to the blogger for a reasoned, comprehensible, authoritative rebuttal to what HAD been (to a lay person) a convincing and disturbing article. HOWEVER (you could hear the "but" coming, couldn't you?) I would urge Mr. Motl to revise it somewhat. His rebuttal loses some credibility when he takes a "cheap shot" at Prof. Disney by playing on his name, and when he uses terms like "senile".