Martin Rees [5.7.96]

Alan Guth:Martin Rees is my favorite theoretical astrophysicist. Whatever subject in astrophysics you ask him about, he's incredibly knowledgeable and incredibly helpful as well. If you ask him a question, he'll go on and on explaining in detail what is known about that subject. He's just marvelous.


MARTIN REES is an astrophysicist and cosmologist; Royal Society Research Professor at King's College, Cambridge; author ofBefore the Beginning: Our Universe and Others, forthcoming, 1997, and, with John Gribbin,Cosmic Coincidences: Dark Matter, Mankind, and Anthropic Cosmology (1989).

Martin Rees's Edge Bio Page

[Martin Rees:] The public is always interested in fundamental questions of origins. Just as they like dinosaurs, they're interested in cosmology. It's rather remarkable that the subjects which interest the public most consistently are sometimes so remote from everyday concerns. People who say that we have to make our work "relevant" to attract public interest are clearly on the wrong lines, because nothing could be less relevant than dinosaurs and cosmology.

Cosmology is exciting to the public because it's clearly fundamental, and this is a rather special time in the subject. For the first time, it's become a part of mainstream science, and we can address questions about the origin of the universe. We can talk about the details of what the universe was like when it was one second old. We can talk about even earlier stages, and ask basic questions; it's a very special and exciting era in the subject.

I would describe myself as an astrophysicist and cosmologist, in that order. An astrophysicist tries to understand individual objects, like galaxies, quasars, stars, and their evolution, whereas a cosmologist is concerned with the entire universe, not the contents of it. I try to span those two disciplines, which after all are very closely linked. I'm not a particularly good mathematician. My work tends not to be deductive system-building, as it were, but attempts to explain the phenomena.

When I started out, cosmology was primarily a theoretical subject, because there were essentially no data at all. It's only since the 1960s that we've known much beyond the fact that the universe is expanding. There have been exciting developments, to the extent that now we can talk in a quantitative way about the early stages of the universe, and there's been a tremendous extension in the range of cosmological issues we can discuss in a serious scientific way. Those issues used to be purely speculative, but now they're real science.

I haven't focused on any single fundamental question; I've tried to keep the big picture in mind, and I've been fortunate, because the subject is one in which a synthetic approach does often bear fruit. Data come in from optical telescopes, radiotelescopes, and spacecraft, and my colleagues and I try to put these together and make sense of them. It's like telling an engineer he's got to make something work meeting certain specifications. Nature gives us specifications, and we've got to use the laws of physics to see if we can "make something work," and make sense of these phenomena.

But there's always the nagging possibility that perhaps the laws of physics as we understand them now are inadequate. That's an extra motivation. The first reason for studying astronomy and cosmology is simply exploration, to discover what's out there. The second reason, which is what motivates astrophysicists, is to try to interpret what's out there and understand how the universe evolved, how the complexity of the present universe has emerged from the primordial simplicity. The third reason is that the cosmos is a laboratory that allows us to probe the laws of nature under conditions far more extreme than we could ever simulate in a terrestrial laboratory, and thereby to extend our knowledge of the fundamental laws of nature.

Another thing which interests me is the psychology of practitioners of the subject. Many people become strongly emotionally committed to their theories and defend them, almost like advocates, against contrary evidence. It's a real trauma for them to have to give their theories up. I've never been like that myself. I've always been quite happy to work almost simultaneously on two contradictory hypotheses, simply because if we don't really know what the explanation for something is, and we want to understand it, then exploring the consequence of different ideas is a good methodology. One's research may lead to a new test, or reveal a new contradiction. The scientific community collectively works like that, but not all individuals seem as content as I am to work simultaneously on two different theories.

One of the themes of my work is trying to understand extreme objects in the universe, objects that involve black holes, energetic outbursts, and so on. I'm associated with several ideas on quasars and the centers of galaxies. This subject is called high-energy astrophysics. In the last ten years, I've increasingly moved towards what you might call cosmogony. It's now feasible to learn not just about the present structure of the universe by surveying nearby galaxies, but about the early universe by looking at its distant parts, so that we're probing what the universe was like when galaxies were just forming, and even the pregalactic universe.

The most active areas in which I'm involved are how galaxies and galactic clusters formed, what the dark matter is, and whether the universe has enough material in it to cause it eventually to collapse or whether it will go on expanding forever. We don't yet have the answers, but I would expect that within the next decade we'll have a consensus view on some of those questions. I believe we'll understand more about how galaxies form, just as we now understand how stars form, and I hope we'll discover what the dark matter is. One of the embarrassing features of our current perception of the universe is that 90 percent of what it's made of is unaccounted for. This so-called missing material could be anything from very faint stars to exotic particles or black holes. Obviously, we can't understand the galaxies until we understand what makes up 90 percent of their mass.

We have good reason to believe that there's a lot of stuff in the universe which exerts a gravitational force but which we don't see. The simplest line of evidence comes from a disk galaxy, like our Milky Way, which is spinning. If you look at the outer parts of disk galaxies, you find that gas, way out, is orbiting surprisingly fast. It's orbiting faster than it would be if it was just feeling the gravitational pull of the stars you see. That's one line of evidence indicating that there must be a lot of dark matter holding these galaxies together. Other evidence comes from gravitational lensing and from the internal motions of clusters of galaxies. We believe that the dark matter is ten times as important gravitationally as what we see, and its nature is completely uncertain. But obviously the cosmogonic process — the origin of structure — is dominated by gravity, and therefore unless we know the nature of the stuff exerting most of the gravity, we're not going to have a definite answer to how the galaxies formed. The nature of the dark matter is one of the key uncertainties now.

If I was to say in one sentence what I'm trying to do, and what I suppose all cosmologists and cosmogonists are trying to do, it's simply to understand how the universe has evolved — over its fifteen-billion-year history — from a hot, compressed, amorphous fireball to its present state, in which we see galaxies and clusters of galaxies, and stars and planets, all displaying an enormous range of complexity of which we're a part. We want to understand the various stages in the emergence of structure: how the expanding universe developed condensations that turned into galaxies and galactic clusters, how stars formed in those, how the stars evolved, how the chemical elements were made, and how, on at least one planet, around at least one star, complex creatures evolved able to wonder about it all.

What's impressive is that we can address these questions at all. One reason we can is that in some respects the universe displays more simplicity than we had any right to expect. It displays simplicity in two senses. First, the large-scale structure of the universe is quite uniform and symmetric. There are all kinds of inhomogeneities on the scale of galaxies and clusters, but on the very large scale the universe is fairly uniform. Every bit has evolved and has the same history as every other bit, provided that by a "bit" we mean a "box" a few hundred million light-years across. In a broad-brush sense, the universe is smooth and homogeneous. When we look at a distant part of the universe, we're confident that we're seeing conditions as they were in our vicinity a long time ago. We could not assume that, if different parts of the universe had quite different histories.

The other remarkable feature is that the laws of physics are the same in all observed parts of the universe. When we take spectra of the light from distant quasars, the spectra indicate atoms just the same as those around us, and we believe that the laws established in the lab are adequate to explain everything in the observable universe, right back to when it was only a microsecond old. When we get earlier than a microsecond, the densities, energies, and pressures were so high that we have an uncertainty about the basic physical laws. After the first microsecond, the universe had expanded to where the densities were no higher than those we can achieve in the lab, and therefore we're likely to know the relevant physics.

There's also an added interest, because it's through the inferences we might be able to draw about the ultra-early universe — the first microsecond — that we can perhaps learn things about fundamental physics which we can't learn directly in the lab. Even in our biggest accelerators, we can't achieve the energies that particles possessed in the ultra-early universe. Also, many of the key properties of the universe — such as why it's expanding the way it is, why it has the simplicity and symmetry without which cosmology would be quite intractably difficult, and why it contains the observed ratio of matter to radiation — can't be understood without better knowledge of the first microsecond.

The microwave background radiation, discovered in 1965 by Arno Penzias and Robert Wilson, was the most important advance in cosmology since the late 1920s, when Edwin Hubble discovered that the universe was expanding. Hubble's discovery suggested that the universe had emerged from a compressed phase in the past, but there was then no evidence for that phase. Indeed, the steady- state theory, developed by some rather vocal Englishmen, held that such a compressed phase had never existed and that the universe had always been the same. It was the discovery of the background radiation that clinched the case for there having been a dense, hot, early stage of the universe, and almost all cosmologists became convinced fairly quickly. The resultant shift in cosmological opinion was almost as sharp as the concurrent shift in geophysical opinion in favor of continental drift — which was another formerly wildly speculative idea shown to be true. After the mid-sixties, almost everyone believed in the hot big- bang theory, of which this background radiation, now cooled to 2.7 degrees above absolute zero, was a fossil.

Since 1965, there's been a succession of more and more accurate measurements of the spectrum of this radiation, and of its angular distribution over the sky, because it's clearly a key cosmological probe. Two crucial discoveries were made. Nearly twenty years ago, the astrophysicist George Smoot measured our motion relative to the universe by finding that the background radiation, instead of being exactly the same temperature in every direction around us, was slightly hotter in one direction than in the opposite direction. This is because we and our entire galaxy are moving relative to the frame of reference defined by the large-scale universe, at a few hundred kilometers per second. Smoot made this discovery by flying his equipment on a U-2 spy plane and measuring the background radiation to a precision of more than one part in a thousand.

Smoot then went on to become one of the key people involved in the COBE satellite, which was launched in 1989 to investigate the radiation further. He was the PI, or principal investigator, for an instrument that looked for variations in the radiation temperature in different parts of the sky to a precision of one part in a hundred thousand. He found that the temperature was not completely uniform: some regions were slightly colder than others. The interpretation of this is that the early universe was not completely smooth. It's smooth in the sense that the surface of the ocean is smooth — a mean curvature, but with ripples superimposed on it. The "ripples" had been predicted to exist, as the seeds from which galaxies and clusters formed. Smoot's instrument onboard the COBE satellite was the first one sensitive enough to have found these fluctuations.

Had they not been found at that level of sensitivity, persons like myself would have been deeply disconcerted, because we all believed that the galaxies, clusters, and superclusters had formed by gravitational instability — a process whereby any part of the early universe that was slightly denser than average would lag behind as the universe expanded, and would eventually condense out. Galactic clusters and superclusters could not have condensed out by the present time unless inhomogeneities already existed in the early universe, with an amplitude that would imprint one part-in-a-hundred-thousand fluctuation in the microwave background. That was the level theorists knew one had to shoot for in doing this experiment, and that was the level achieved by Smoot's instrument on the COBE satellite.

I'm interested in what general properties our universe had to have in order to develop complexity. One of the obvious requirements is a force like gravity, which allows structures to condense, via instabilities, in an initially featureless universe. But, ironically, the weaker gravity is, the better the chances of a complex universe developing, because if gravity were so strong that it crushed things the size of complex organisms, there would be bleak prospects for evolution. Indeed, if gravity were much stronger, the lifetimes of stars would also be very short, and this would allow less time for complexity to emerge via any evolutionary process. If the force of gravity didn't exist, no cosmic structures would ever have condensed, but the weaker it is the grander its manifestations are. It's because gravity is so weak that stars and galaxies are so huge, on the scale of ordinary phenomena. It's interesting to try to quantify this, and see if we can understand why gravity should be so weak.

The general idea of the emergence of complexity is very relevant here, because gravity has the unusual property of allowing an initially featureless universe to develop structure. Gravity leads to instabilities, and pulls material together to form galaxies and stars. As stars lose energy, they get even hotter in their centers and more compact; eventually, nuclear- fusion reactions ignite inside them, allowing the temperature contrasts between stars, planets, and the dark night sky that are essential — as Prigogine and others have taught us — for the "nonequilibrium thermodynamic" processes that built up complex molecules and life. So gravity drives things ever further from equilibrium and allows the disequilibrium, which is the prerequisite of any kind of complexity, to develop from an amorphous early universe. This is the kind of process we're trying to understand quantitatively. Another development in the last few years is the possibility of doing realistic simulations of gravitational clustering, gas dynamics, and so forth, to explore how a structureless universe can evolve.

I've also ventured into more speculative topics, like whether physicists might by accident destroy the universe by doing a particular sort of experiment. This issue arose because of ideas stemming from Alan Guth's inflationary-universe theory. The whole idea of the inflationary universe requires that even empty space (what physicists call "the vacuum") had unusual properties in very early times and underwent what's called a phase transition — something like what happens when water freezes. Some people — the physicist Sidney Coleman was one of the first to make this point — suggested that our present vacuum may not be in the lowest possible energy state. Space might therefore undergo a further phase transition to a different kind of vacuum state, in which the laws of physics would be changed. All particles as we know them, and everything we see around us, would be destroyed. Our present vacuum may be, so to speak, supercooled, as very pure water can be supercooled without undergoing the phase transition to ice; and, just as the insertion of a speck of dust makes supercooled water freeze suddenly, maybe some trigger could transform the whole of space into some other quite different state. Could physicists, by an experiment done in an accelerator, trigger this effect by inadvertently producing a bubble of the new vacuum, which would then expand at the speed of light and engulf the universe?

This might appear absurd, but it's easy to think of ways in which we've produced conditions that have never existed naturally anywhere. For instance, there was never anything in the universe colder than 2.7 degrees above absolute zero — the present temperature of the microwave background — until we made refrigerators (unless, that is, there's intelligent life elsewhere). The kind of thing that might create "dangerous" conditions would be a collision between very-high energy particles in a big accelerator; such a collision might create a big local energy density of just the kind that might trigger a phase transition.

With the Dutch astrophysicist Piet Hut, I wrote a paper addressing the question of whether accelerators could create concentrations of energies that had never existed anywhere in the universe since the big bang itself. Our conclusion was quite reassuring. We calculated the collision rate between cosmic-ray particles — which are particles that move, at very low densities, in interstellar space, at very close to the speed of light. We worked out the most energetic collisions that ever happened in our part of universe, and we discovered that these would have been substantially more energetic than any conceivable event that could occur in an accelerator. That's reassuring. It means you'd have to go a long way beyond the collision energies expected in supercolliders before there was any risk of Doomsday.

I'm also trying to bring into a scientific context the concept of an ensemble of universes, each with different properties. These ideas are associated with many people, but I'll mention only the Russian physicist Andrei Linde, who proposes chaotic and eternal inflation — that is, the idea that new universes can sprout from old ones, or can inflate into a new domain of spacetime inside black holes. He and others have argued that our universe is just one element in an infinite ensemble. Different universes in this ensemble may be governed by entirely different physical laws, numbers, and dimensions. Some may have very strong gravitational force, some may have no gravity, some may have different kinds of particles. If that's a possibility, then this concept of an ensemble, which I prefer to call a meta- universe, gives a scientific basis to anthropic reasoning — the idea that it's not a coincidence that we find ourselves in a universe where conditions are somehow attuned for the development of complexity. If all possible universes governed by all possible laws exist, then obviously it occasions no surprise that some of them will have laws of nature that allow complexity, and then it's no coincidence — and, indeed, inevitable — that a universe like ours exists, and, of course, that's the one we're in. This suggests the idea of "observational selection," as it were, of universes. I take this seriously. There's an ensemble of universes. Insofar as one can put a "measure" — in the mathematical sense — on relative numbers of universes, most will be stillborn, in the sense that there would be no complexity evolving within them. Some, contrariwise, may have vastly greater potentialities than our own, but these are obviously beyond our imaginings.

I have substantial confidence in talking about the universe back to when it was a microsecond old; I have as much confidence in the relevant theories as I have in inferences about the early history of the earth from geophysics or paleontology. The level of evidence and the nature of the argument are similar — indeed, the cosmological evidence is rather more quantitative. But when we get back into the first microsecond, we confront important ideas, like inflation and phase transitions, which in some form will be part of the eventual correct world picture. The trouble is that we don't yet know enough about the extreme physics to be able to predict anything very quantitatively. But the new concepts certainly expand our perspective, by admitting the possibility of an entire ensemble of other universes with different properties. We have to then distinguish different definitions of "universe." You can mean by "universe" what we observe — a region some fifteen billion light years across; you could define it as larger than that — as the domain from which light will eventually be able to reach us; or you could define it as the grand ensemble, which contains all possible universes, governed by all possible physical laws. It's the last concept — the meta universe — which I find the most fascinating, and which I believe is just coming within the scope of serious scientific discourse.

Rather than use the phrase "anthropic principle," I would prefer to talk about "anthropic reasoning." This is the general line of argument that some features of the universe are a prerequisite for the existence of observers, and so we shouldn't seek a basic explanation of those features: they're just a function of the fact that we're here. In one sense, anthropic reasoning is obvious and quite banal; we don't bother wondering why we're in a special place in the universe, near a star like the sun, and not in a random place in intergalactic space. Nor do we wonder about why we're living in the universe when it's fifteen billion years old rather than in the first few seconds, because for us to exist the universe had to cool down, and a long chain of prior evolution plainly must have occurred.

Some people have tried to take anthropic reasoning further, by claiming that it's somehow mandatory that any basic laws of nature must permit conscious observers. I find this view hard to take seriously. The status of anthropic reasoning depends very much on the nature of the basic laws. If these laws — that is, the relative strengths of gravity and the other fundamental forces, the masses, spins, and charges of the elementary particles, and so forth — are, in a sense, accidents of the way our universe cooled down, then you can perfectly well imagine universes where the laws are different and which are not propitious for life. All these universes might exist, and we happen to be in the one that has the "right" conditions. There's nothing remarkable about that.

On the other hand (and here I would sympathize more with the line taken by the late physicist Heinz Pagels), if the fundamental laws of nature are unique — if it turns out that the laws of physics could not have any other form anywhere, and some unique equation tells us the strength of the forces and the masses of the particles — then it would seem just a brute fact, or luck or providence, according to your perspective, that those unique and simple laws allow complexity to evolve. I'd be astonished at this outcome, but my reaction would be rather like- -to make an analogy — my amazement at the fact that you can write down a simple algorithm for something as complex as the Mandelbrot set, with its infinite depths of structure. It is indeed amazing, but that's just mathematics; and, similarly, there may indeed be unique fundamental physical laws that just happen to have such incredibly rich consequences.

If the laws of nature are unique, then there's no room for anthropic selection, because the laws are just "given." Either you accept the laws, and their remarkable consequences, just as a brute fact, or you go all the way with the "strong anthropic principle." But if there's an ensemble of universes that cool down differently, then some have conditions propitious for life, and others are short lived; they will be too cold, too empty, and so forth. Then there is room for straightforward anthropic selection. And we are perforce in one that's hospitable enough to allow the requisite complexity.

A reason for downplaying anthropic arguments is that physicists would do well not to believe them too strongly. Obviously, many features of the universe we can't yet account for will be explained by straightforward physical arguments. If people believed that some features of the universe were not fundamental but just accidents, resulting from the particular way our domain in the meta universe cooled down, then they'd be less motivated to try to explain them. When I interviewed Steven Weinberg for a radio documentary about ten years ago, he made this point — that it would be best that physicists not believe in the anthropic principle, because otherwise they wouldn't be so motivated in seeking a unified theory, and if they didn't seek it they certainly wouldn't find it. These new concepts of a meta- universe (or ensemble of universes) bring anthropic selection closer to the mainstream of scientific discourse.

Lee Smolin: I met Martin Rees only recently, during a visit to Cambridge University. Of course, I'd heard about him for many years, as he's admired by a great many people. He's certainly one of the most influential people working in astrophysical and cosmological theory, and after some discussions with him it was obvious to me why: he is simultaneously open to new ideas and suggestions and careful and rigorous in his response and criticisms. Also, it's difficult to suggest an idea about the evolution of structure in the universe or the formation of the galaxies that he hasn't thought of or played with or perhaps even written about at some time. He's also great fun to talk with, and as far as I could tell completely without pretention. It's not fun to hear some people criticize one's ideas, because they turn such discussions into something competitive, but I can say that I really enjoyed hearing his criticisms of some ideas of mine. He didn't believe them, but he'd thought about them carefully, and he told me exactly where he thought they were most likely to go wrong.

Much of the credit for what I like to think of as the discovery that the laws of nature are special in ways that allow the universe to be very structured is due to him. This idea was suspected initially by an earlier generation — particularly by P.A.M. Dirac, Fred Hoyle, and Robert Dicke. But it's my understanding that it was really Martin, together with a younger colleague, Bernard Carr, who assembled all the evidence for the specialness of the laws of nature. The result was a paper they published in Nature which has had an enormous influence on all those who think about the anthropic principle. More than one book has been written by expanding their article. But what I think is most important is that they've made the case for the specialness of the laws of nature strong enough so that those of us, like myself, who aren't attracted to the anthropic principle have to take it seriously. Then the question is, If we don't accept our own existence as the explanation for why the universe is so special, can we find another explanation?

If I can use him to say something more generally, there's something truly wonderful about the English tradition in astronomy and physics that we in America could learn a lot from. There's no country in the world that has had such a collection of inspired originators of cosmological and astronomical ideas. In this century there has been Arthur Eddington, Fred Hoyle, Dennis Sciama, Roger Penrose, Stephen Hawking, and Martin Rees himself, and there are others not as well known. There's a way in which these people have been educated to work with the highest standards of rigor and honesty, and then allowed to develop their ideas in an atmosphere much freer and more tolerant of individuality, and even eccentricity, than the American scene. The American scene is larger, and in terms of money we're better supported, but there's something unhealthy about the way in which we're so often worrying about how the National Science Foundation and the community will respond to our grant proposals. Perhaps I'm naive, but I have the impression that the British seem, at least up until now, to have avoided this overbureaucratization of science.

It's also only England that could have produced scientists like Jim Lovelock or the physicist and philosopher Julian Barbour, who stay at home, unconnected to any university, but do original and important work that wins the respect of their less courageous colleagues in the universities. Perhaps the point is that the English have never forgotten that in the end the advances of science are made by creative personalities, so that the best way to advance science is to give people the best possible education intellectually and morally — I say "morally" because I think science works because scientists practice an ethics of honesty and tolerance — and then give as much freedom as possible to those who show themselves to be creative. This is something I think we need to think about more in the United States.


Back to Contents

Excerpted from The Third Culture: Beyond the Scientific Revolution by John Brockman (Simon & Schuster, 1995) . Copyright © 1995 by John Brockman. All rights reserved.