John Peacock and Alasdair Richmond
Our night skies contain stars in different stages of development - newly formed bright stars, swollen red giants and dense stellar remnants like neutron stars. Why such diversity? We observe this range of stars because we are chemically complex beings who require several different elements in order to survive - from simple, abundant elements like hydrogen through to heavier, rarer elements like iron and sodium. Nucleosynthesis (the fusion of nucleons to form atoms) almost certainly began within 3 minutes of the Big Bang and produced the hydrogen and helium that formed the first stars. Heavier elements usually form only through stellar nucleosynthesis, i.e. nuclear fusion inside stars, and have to be distributed before they can form complex structures and (thereby) life. Heavier elements only get scattered across space when stars grow old and explode as supernovae. We are literally stardust. Before we could evolve, supernovae had to occur. Hence our night sky contains the remnants of old stars - times containing only young stars boasted few observers. It's not surprising complex creatures like us don't live three minutes after the Big Bang, when only hydrogen and helium existed in quantity. Looking at the kinds of observers we are and the elements that compose us supports deductions about the stars we will observe and when in the universe's history we will live. These are anthropic effects.
Anthropic reasoning reflects the important philosophical idea that good explanations should make whatever you're trying to explain more probable or typical. You should expect, all else being equal, to be in a probable or typical location for creatures like yourself. We usually favour explanations that make our location unexceptional rather than exceptional - it is not an explanatory virtue to make something less probable.
The anthropic principle is an umbrella name for a range of different observations about how the sort of observers we are reflects our place in the physical world. The term was coined by physicist Brandon Carter and first appeared in print in Carter (1974). Despite subsequent widespread use by scientists and philosophers, the anthropic principle has been defined independently in different, often incompatible ways. Carter's key idea reflects the complex and subtle interrelations between the kinds of observers we are and the physical conditions we observe. Many fortuitous contingencies (e.g. the presence of carbon, the three-dimensionality of macroscopic space, electromagnetic forces being 1040 times more powerful than gravitational forces, etc.) appear necessary for observers like us to evolve.
For many centuries, science assumed our place in the universe was uniquely privileged. The Ptolemaic astronomy that prevailed until the sixteenth century placed the Earth (and hence humanity) at the centre of the universe, and the planets and stars in concentric orbits around us. However, the heliocentric (sun-centred) astronomy associated primarily with Nicolaus Copernicus dethroned us from a privileged centre and, coupled with discoveries in geology, suggested the universe extends far further in space and time than had hitherto been thought. Just as the Earth is a tiny fraction of the universe's spatial extent, so humanity's recorded history is a tiny fraction of the universe's history. Charles Darwin's discovery of evolution by natural selection suggested human beings could fruitfully be considered as one more product of undirected, non-intentional processes. Natural selection lets us explain how complex living things (even life) can arise, without invoking external oversight or Design.
After this apparent dethroning of humanity, one might go to the opposite extreme and conclude there is nothing remarkable about our location in space and time at all. However, excessive insistence on our typicality is also misleading, since human beings are not scattered haphazardly throughout space and time. Only the Earth in the solar system offers just the right temperature conditions for carbon-based life - orbiting in the so-called 'Goldilocks zone': neither too hot nor too cold but just right. Because carbon is tetravalent (i.e. has four valency opportunities in its outermost electron-shell), it can form long-chain molecules. Carbon is great for forming complicated chemical structures using many different elements and hence is the best element known for building life. Life needs carbon chemistry; carbon chemistry needs a narrow range of temperatures, and that range is found only on Earth. Barring one, the solar system's many bodies are very hostile to life, being (e.g.) too hot, too cold, too volcanic, riddled with radiation, having unbreathable atmospheres, etc. So unsurprisingly, the solar system is mostly dead. Of all the bodies in the solar system, only one bears the right conditions for life, and (unsurprisingly) that one is Earth.
Anthropic arguments seek to balance excessive anthropocentrism and excessive insistence on human typicality: the conditions we observe may be typical for observers but atypical of the universe as a whole. We are physically based observers and can only survive (and evolve) in very specific physical circumstances - we require an environment where carbon can form complex chain molecules, where water is readily available in liquid form and where oxygen can be breathed. We require a planetary environment in which to live, and specifically one which is not subject to extremes of temperature, pressure and radiation. Hence we need to live on a planet with a breathable atmosphere and a stable orbit about its parent star. We require different elements, ranging from the lightest (i.e. hydrogen) through to heavier elements like iron and selenium. Furthermore, it seems observers like us can only survive in universes where space is three-dimensional (at least at the scale of everyday objects). Our world has three (macroscopic) mutually perpendicular, spatial axes - running left-right, up-down and forward-back. You don't live in a two-dimensional plane or in a space with four or more dimensions. Why? In less than three dimensions, nerves can't run through cells without crossing one another and nervous systems like ours cannot exist. However, in four (or more) dimensions, it's very hard to enclose volumes, hence cells and complex organs can't function.
Many apparently incidental features of the universe have something in common: they cannot differ from what we observe and still have observers around to notice them. The physical conditions we observe appear 'fine-tuned' into the narrow band that permits life. Does fine-tuning show the universe was designed or generated so that life had to evolve in it? While some anthropically inspired arguments claim life is the reason (or goal) which the universe exists to serve, such arguments may go rather beyond what anthropic effects actually suggest.
Anthropic arguments begin by noting that the conditions necessary for context-sensitive observers set restrictions on the sorts of conditions such observers will probably find themselves observing. 'Anthropic' comes from the Greek anthropos, which literally means 'Man', but anthropic reasoning makes no special reference to humans in general, or male humans in particular. Any physical observers can use anthropic reasoning: 'The same self-selection principle would be applicable by any extra-terrestrial civilisation that may exist' (Carter 1983, p. 348). Different kinds of observers would find the conditions necessary for their evolution constraining their observations too. Carbon-based observers like us are more likely to find ourselves in conditions suitable for carbon-based life. If you were a sulphur-based observer, expect to have evolved in conditions amenable to the evolution of sulphur-based observers. (Likewise, if you're made of high-energy plasma, expect to find yourself in plasma-friendly conditions - e.g. inside a star's hot interior.) Likewise, anthropic reasoning need not imply any (Design or other) intentional explanation for our existence. Both anthropic and natural selection explanations explain features that suggest Design but in non-Design terms. Anthropic reasoning assumes life is highly context-sensitive and requires a narrow range of physical conditions. Postulating an intelligent Designer outside the physical universe assumes intelligence can transcend our physical environment. So arguing from fine-tuning to Design risks arguing from fine-tuning to the conclusion that consciousness does not require fine-tuning at all.
Closest to Carter's original proposal is the weak anthropic principle (WAP): 'What we can expect to observe must be restricted by the conditions necessary for our presence as observers' (Carter 1974, p. 291). Any observers who require a delicate range of conditions for their sustenance will almost certainly only evolve where those conditions are met. Typical observers should inhabit amenable surroundings. If you're thinking 'But I'm a unique snowflake of a person', and bridling at being described as typical, here are some tests:
Presumably not. While you are a unique and irreplaceable human being, it does not follow that you are unique in absolutely every way. We're all unique in various ways, but we're also fairly typical in various ways. Context-sensitive observers will more likely find themselves evolving in, and hence observing, areas of spacetime which are hospitable to them but possibly atypical of the universe at large. Hence, WAP suggests we should beware of extrapolating too hastily from the conditions we observe in our spatiotemporal neighbourhood to conditions in the universe as a whole. Thus, anthropic reasoning seems to counterbalance the Copernican principle, which counsels us to view ourselves and our environment as being as typical as possible. However, any tension between anthropic and Copernican thinking is more apparent than real. Another test: observers will probably live in (comparatively rare) planetary environments and not in (very common) interstellar or intergalactic space. You aren't reading this floating unprotected in interstellar space, for all that interstellar space is vastly bigger than the Earth. Interstellar space is much more 'typical' of conditions in the universe as a whole - but interstellar conditions are deeply inimical to life and it's not surprising you neither evolved there nor reside there.
According to one popular misreading, WAP says (e.g.) 'Observers must have evolved where they have evolved', as though WAP was a mere tautology like All swans are swans'. As a result, it's often thought WAP must be devoid of any practical or scientific implications. However, WAP does not say 'You evolved where you evolved', which would be tautologous, but: 'You are far more likely to have evolved in conditions suitable for your evolution'. While it is overwhelmingly more probable that you evolved in conditions suitable for your evolution, it is conceivable that it was not so. In a big enough universe (or universe-ensemble), some very odd observers may exist, who arose in conditions where their evolution would be very unlikely. However, you should not expect to find yourself among such unusual observers. An example: if we assume consciousness is purely material or generated by material structures, then it's conceivable (albeit astronomically unlikely) that minds like ours could spring into existence fully formed through pure happenstance. Maybe lightning struck some chemicals in a swamp and triggered spontaneous cellular activity that issued in me, complete with false memories of a life I never had. This non-evolutionary 'swamp person' story 'explains' my physical and mental make-up. (In a sufficiently big universe, or universe-ensemble, with sufficient variation in conditions, the generation of swamp-people somewhere in space and time will be a certainty.) However, it's a poor explanation that makes me so bizarrely atypical. I won't accept the 'swamp man' story unless I find exceptionally powerful evidence in its favour: if I think most conscious observers arise through evolutionary processes and I've no reason to regard myself as an exception to this generalization, then I should believe I too evolved.
Some scientists and philosophers (not least Brandon Carter) took this 'tautology' criticism as a challenge and proposed ways to derive empirical predictions from WAP. Carter (1983) considers the number of crucial steps that are necessary for advanced life forms to evolve, noting that the time it took intelligent life to evolve on Earth (c. 3.5 billion years) is of the same order of magnitude as the age of the Earth (c. 4.5 billion years) and the likely time remaining before the sun burns out (again c. 4.5 billion years). Why does the time the Earth has been in existence resemble the time it took observers like us to evolve? In Carter's (1983) terms, life had to take a number of 'crucial steps' while evolving intelligent observers, e.g. generating multicellular from unicellular life. Many of these steps might have taken millions of years. Carter says the apparent similarity between the Earth's age and the history of Earthly life is explicable in anthropic terms if intelligent life typically takes longer to evolve than the life-sustaining period of its parent star. Thus, most life forms will take only a few crucial steps towards intelligence before their parent stars die, and so make further evolution impossible nearby. Most potentially life-supporting stars do not support life long enough to let intelligence evolve. So, Carter (1983) predicts, even if extraterrestrial life is common, extraterrestrial intelligence is rare. Carter's formula explains the so-called 'Great Silence' from alien life forms (as in Fermi's paradox): if the Earth is a typical abode of life, and many extraterrestrials exist, where are they? Perhaps Earth is a specially maintained 'zoo' surrounded by aliens who observe us undetectably. A 'cosmic zoo' hypothesis is hard to falsify but more importantly, it makes us very atypical life forms, and is thus counter-anthropic. On this hypothesis, typical observers don't live in zoos.
The strong anthropic principle (SAP) generalizes WAP and says the presence of observers suggests the universe is amenable to the evolution of such observers, thus: The Universe must have those properties which allow life to develop within it at some stage in its history' (Barrow and Tipler 1986, p. 21). If we assume our ancestors evolved in this universe, this universe must have boasted conditions that made such evolution possible. Thus, SAP applies WAP to the whole universe, not to any isolated subregion. SAP is often (mis)read as necessarily carrying teleological explanation, i.e. invoking life as the universe's goal or intended end-state, and hence claiming that the universe must (in some categorical sense) have been set up so as to produce life. Advocates and opponents of SAP alike have encouraged the idea that SAP is teleological or Design-orientated. In fact, teleological readings of SAP confuse a categorical inference with a conditional one. If I find frogs thriving in a pond, and I assume they grew there normally, I should assume the pond recently held frogspawn. Likewise, if I believe I breathe oxygen, I should believe it's likely that there has been oxygen in my environment - it does not necessarily follow that Nature has been rigged so this pond inevitably yielded frogs or that oxygen must have been present so my survival would be ensured. Given you exist, it follows your existence must be possible but it does not follow that you must (categorically) have existed. So the conditional WAP inference 'If we evolved hereabouts, local conditions must have been suitable for our evolution' and its generalized SAP cousin 'If we evolved in this universe, this universe must have been suitable for our evolution' seem plausibly in tune with Carter's original idea; however, less plausible is any inference that runs 'The presence of conditions suitable for life in this universe shows this universe must have been designed with such evolution in mind'. Imagine a frog in our pond reasoning: 'The presence of frog-sustaining conditions in this pond shows the laws of nature were designed for frogs'. Such seems a presumptuous frog. The Earth's diameter is roughly 12.7x l06 m, whereas the observed universe's diameter is roughly 8.8x 1026 m, so a pond 1 metre across occupies a vastly bigger fraction of the Earth's surface than Earth does of the observable universe.
Further extensions of anthropic reasoning include the participatory anthropic principle and final anthropic principle (PAP and FAP, respectively). PAP says observers somehow determine or create the physical properties they observe: 'Observers are necessary to bring the universe into being' (Barrow and Tipler 1986, p. 22). So, humans are not passive observers of the universe's physical order but active creators of it: we help make the natural order even as we observe it. However, while PAP often invokes quantum theory in its defence, many quantum theories do not accord consciousness any such special role.
FAP says life, once created, must thereafter endure forever: 'Intelligent information-processing must come into existence in the Universe, and, once it comes into existence, it will never die out' (ibid.). Amongst FAP's difficulties is trying to derive testable predictions that support a literally infinite future for life. FAP can make predictions, e.g. the universe must allow infinite extendibility of the material processes that sustain consciousness, but such predictions don't lend themselves readily to falsification (PAP and FAP command markedly fewer adherents than WAP or SAP).
In recent years, interest has grown in using the above ideas on anthropic observer selection to understand some of the ways in which cosmology seems puzzlingly unnatural. Naturalness is often invoked as a principle in order to see whether a physical theory may hide a deeper or simpler model. For example, the standard model of particle physics contains twenty-five numbers (if we include massive neutrinos). Can these numbers be related and/or do some take strange values? The fundamental masses in the standard model are generally felt to be unnatural, since quantum corrections would tend to make them larger: comparable to the largest energy scale that can be imagined, which is the Planck scale of quantum gravity. This is an energy at which the fundamental idea of particles existing in an uncurved background spacetime breaks down, degenerating into a foam of quantum black holes. The characteristic energy at which this happens translates (via E = mc2) to a mass of 2 x 10-8 kg. This is a very small mass by everyday standards, but huge in the world of particle physics: 1019 times the mass of the proton. Thus the small magnitude of the masses of elementary particles is a puzzle.
Cosmology is also unnatural to some extent. There are various coincidences of value between apparently unrelated quantities (e.g. the densities of dark and visible matter are equal to an order of magnitude; the time of matter—radiation equality is within a factor of 2 of the time of last scattering of the CMB (cosmic microwave background)). But the biggest naturalness problem is the magnitude of the dark-energy density, which is far smaller than seems at all reasonable. The measured density of dark energy is about 10-26.2 kg m-3. As discussed earlier, the electromagnetic zero-point energy of the vacuum could exceed this figure unless we cut off the contribution from high-frequency radiation. In practice, this cut-off must be performed at a wavelength of 0.5 mm, which makes no sense at all. The corresponding photons are less energetic than the most energetic photons produced in the LHC (Large Hadron Collider) by about a factor of 1015; so even if we assume that some new physics is just around the corner at CERN, the characteristic energy of the vacuum is unnaturally low. A more dramatic way of emphasizing the same point is to go to the extreme of cutting off at the energy scale of quantum gravity, the Planck scale. In that case, the vacuum density would be over-predicted by about 120 powers of 10. Either way, there is a major puzzle here.
In considering such naturalness arguments, we are clearly taking a probabilistic view: a low dark-energy density seems 'unlikely', requiring an unusual cancellation. This is implicitly a Bayesian approach - i.e. treating the degree of belief in a theory as a probability. The Bayesian view of statistics can be controversial, and things are more clear-cut when we know that an ensemble of outcomes exists, so that probability can be defined simply via relative frequency. Remarkably, this is very much the case in modern cosmology, where the preferred models inevitably create a multiverse of causally disconnected bubbles, each of which can display different forms of the 'laws of physics'. Given such an ensemble, it is possible to go one step further and ask about how the different physics in the different bubbles can influence the probability of observers arising. In this way, issues such as the strange value of the measured dark energy density can potentially be understood as arising from observer selection out of the ensemble.
Muniverses seem to run afoul of Ockham's razor, or the principle of ontological parsimony. However, Ockham's razor only cautions against multiplying entities beyond necessity, where multiverse theories have unique explanatory potential. Secondly, Ockham's razor may only counsel qualitative economy (i.e. economy over the number of different kinds of things) and not quantitative economy (i.e. economy over the number of similar things postulated). Certainly, this mode of reasoning is increasingly seen as an attractive mode of cosmological explanation, for reasons we now articulate.
The concept of inflationary cosmology was mainly developed in the early 1980s, although the essential idea first arose in a visionary 1965 paper by the Soviet cosmologist E. B. Gliner. The desire was to remove the need to supply arbitrary initial conditions for the Big Bang model: what started the expansion at t = 0? What happened before this initial singularity? Most especially how was the early universe able to start in such a symmetric and uniform state? This last problem is especially challenging, although less obvious at first sight; it is known as the horizon problem. To see the difficulty, imagine that two observers within the current universe wish to exchange information about their local conditions (the mass density, the CMB temperature, etc.). This takes time, since light signals need to be exchanged: a billion years for points one billion light years apart. But once the signals have been exchanged, the observers will find that they experience very much the same conditions, with CMB temperatures equal to 1 part in 100,000. Clearly there must have been a previous episode of causal contact in order to set this uniform temperature, and it might be thought that this would be easy to arrange, since points now separated by a billion light years were once only (say) 1 centimetre apart. But the solution of the Friedmann equation shows that the universe expanded very rapidly near the Big Bang, so that causal contact only occurs progressively. Today, when we observe the CMB, we see radiation that has been travelling since near the start of the Big Bang, so we are looking out to a distance of order 1010 light years. We can see to this horizon distance in two opposite directions, which thus have twice the separation - and therefore cannot yet have come into causal contact with each other. Nevertheless, we can see that they have the same temperature: so causal contact must have occurred, even though it is apparently impossible in the framework of the hot Big Bang expansion.
A more detailed analysis reveals that the solution to the horizon problem requires that the expansion of the universe should be accelerating at early times, rather than the strong deceleration expected in the standard Big Bang. Gliner's initial insight was that this could happen if the equation of the state of the universe changed at early times: being dominated by vacuum energy (in effect an extremely large cosmological constant), which later decays into radiation. In this case, the early expansion of the universe would be exponential in form - i.e. without a singularity even in the infinite past. But what is the mechanism for changing the vacuum density? This was invented by Peter Higgs in 1964. In order to explain the masses of elementary particles (particularly why the W and Z particles that carry the weak nuclear force were not massless), he postulated that the universe was filled with a scalar field, ϕ, and moreover that this field generated an energy density as a function of the field value, V(ϕ). The detection of the Higgs boson in 2013 tells us not only that the field exists, but that it can change with time: the Higgs particle corresponds to small oscillations about the minimum in V(ϕ). This gives the desired mechanism: V(ϕ) in effect functions as a changeable energy density of the vacuum. In practice, we cannot explain an inflationary start to the universe using the Higgs field, but it is assumed that another scalar field (the infiaton) exists and can have the required properties.
In this way, Guth, Linde and other pioneers explained how the universe could begin without a Big Bang (or, at least, without the Big Bang being where we thought - this time is now simply the point at which inflation ended, where the density from the inflaton ceased to dominate the total). The theory also explains how the universe can be so large and uniform: a tiny patch of subnuclear scale is inflated during the inflationary period so that it becomes a bubble of scale very much greater than the presently visible universe. One should picture a watermelon as the inflated bubble, with one of the black seeds representing the sphere of radius roughly 1010 light years, which is all that we can observe. Inflation therefore explicitly predicts that there is much more to the universe than we can observe. In space, it continues exponentially far beyond the maximum scales we can observe. In time, it is unclear how long the universe has existed. A Big Bang singularity can predate the start of inflation, but once there has been sufficient expansion for vacuum-dominated expansion to set in, things settle down to very nearly exponential expansion - which is hard to distinguish from a state that has persisted forever. Thus inflation explains the puzzles about the apparent Big Bang singularity (which is an illusion), but by making it very hard to learn anything about any true ultimate Big Bang.
Unexpectedly, it became clear by 1982 that inflation also explained the origin of structure in the universe, via quantum fluctuations in the early stages of inflation when our current universe was of subnuclear scale. In retrospect, this period in the early 1980s therefore stands out as one of the most stunning eras of progress in physics - matched only by the revolution of quantum mechanics in the early 1920s. At a stroke, a new view of cosmology sprang into being, which in principle offered answers to all the puzzles about how the initial conditions of our current universe were set. This is an astonishing and ambitious vision, but how can it be tested? The key point is that quantum fluctuations should exist in all fields, and therefore inflation predicted not merely fluctuations in density, but a corresponding set of fluctuating primordial gravitational waves. These would fill the universe in the same way as the photons of the microwave background radiation: arriving almost uniformly from all directions, and traceable to a 'last scattering surface' but since gravitational radiation is much more weakly interacting than electromagnetic radiation, the gravitational waves would have interacted much earlier in time: back during the era of inflation in which they were generated.
For decades, detecting these ultimate messengers of the early universe seemed a rather ambitious goal. But astonishingly, on 17 March 2014, the BICEP2 experiment claimed that they had been seen. These relic gravitational waves cannot be seen directly, but via the CMB: they originate 'behind' the CMB's surface of last scattering, and perturb the primordial plasma that exists at this time as they pass through it. The result is to add a characteristic pattern of polarization into the CMB, and it is this polarization signal that was detected, using a telescope observing from the USA's base at the South Pole. So in what sense have the gravitational waves been 'detected'? Experiments on Earth have sought these waves for decades, and most are in the form of an interferometer, where lasers measure the separation between two freely hanging mirrors, looking for a change in separation as a gravitational wave passes through. So is there any real difference between a detection in a local experiment and a detection via the CMB, even though in the latter case we cannot touch the matter that the gravitational waves cause to move? A line does tend to be drawn between 'experimental' science such as laboratory physics and 'observational' science, of which astronomy is the archetype. The distinction is that astronomy has to function with the universe as its laboratory, and experiments designed by nature, rather than under the control of the experimenter. Astronomers tend to be less concerned by the distinction: if an experiment works, who cares who designed it? Part of the attraction of the subject for researchers is indeed the challenge of overcoming natural experiments that yield less information than would be desirable.
This purely observational nature of astronomy becomes strongly apparent when considering a discovery as potentially important as the BICEP2 experiment. Science requires replication and cross-checking, and that is especially true with a discovery of such potential importance. But the result cannot be checked with an independent technique, so other experimenters will seek to measure the same CMB polarization signature. Initially, a focus will probably even be given to observing the same part of the sky, to verify that different telescopes see the same thing. In addition, it will be necessary to understand in more detail the issue of foregrounds: are we sure that the claimed signal is truly cosmological, and not produced by emission from within our own Milky Way that is larger than expected? At the time of writing, evidence has recently emerged that the level of such foregrounds is expected to be larger than estimated by the BICEP2 team, and potentially consistent with their signal. At present, it is therefore not possible to make any definitive claims that primordial gravitational waves have been detected - but current precision of foreground estimates would permit a substantial part of the BICEP2 measurement to be truly primordial: things are fascinatingly poised.
If the BICEP2 detection is eventually confirmed, we would be able to make a number of profound claims, most notably that inflation has now shifted from being just a neat idea, to something that is considered proved. The universe would have gone through an early stage of exponential expansion, where it doubled its size about every 10-38 seconds; the gravitational waves would have been generated about 10-36 seconds before the end of inflation, when the universe was 1055 times smaller than at present.
Now, it is never the case that a theory can be proved true with perfect certainty. Even when many years have passed, with complete agreement between the theory and all available observations, a flaw can arise - as shown by the example of Newtonian physics, which was eventually supplanted by the relativistic approach. Thus all physical theories can only be provisional working hypotheses. In a sense, we can be happier using a framework like Newtonian physics, since we know it is an approximation that is valid in a certain regime. It is therefore a valid truth about the universe, but we are spared having to wonder whether it is the ultimate truth. In a sense, what physicists do is similar to the operation of a courtroom, where what is sought is 'proof beyond reasonable doubt': a consistent version of events that accounts for all that is known, and to which no alternative exists. In this sense, inflation will be considered proved if the BICEP2 results are verified.
A deeply significant characteristic of inflation is that it is eternal: quantum fluctuations continuously perturb the trajectory of ϕ(t) as it 'rolls' down the potential V(ϕ). Most of the time, these fluctuations cause a 'random walk' in which the scalar field meanders further from the minimum in V(ϕ). But every so often, it arrives by chance near the minimum and inflation comes to an end, seeding a bubble universe. This is separated by continuously inflating space from the adjacent bubble, so these members of the multiverse are causally disconnected.
Furthermore, it is easy to find ways in which the vacuum density in each bubble could be different. The modern incarnation of such physics is provided by the landscape of string theory. Without saying too much about what string theory is, it suffices to note that a major feature of the theory is the existence of more than 3 + 1 dimensions (an idea that goes back to Kaluza and Klein in the 1920s, and which could outlive string theory). We are creatures that do not experience the higher dimensions, but we can be influenced by them indirectly: if the universe expands in one of the higher dimensions, the radius of curvature in that dimension can affect properties such as the mass of particles. In short, the higher dimensions have to be stabilized, and the means for achieving this is the introduction of many additional scalar fields like the Higgs or inflaton, each with their own potential. The result is that there are many different points of local stability - i.e. minima in the potential function. Each of these corresponds to a different vacuum density, and the question is whether there are sufficiently many of them, given that the density must be fine-tuned to 1 part in 10120, as we have seen. It was therefore an interesting step when the number of states in the string-theory vacuum was estimated as 10500. Thus in principle the physics basis exists to say that different members of the inflationary multiverse could have different values of the vacuum density: very large in most cases, but occasionally very much smaller.
High values of the vacuum density suppress the gravitationally driven growth that transforms the initial small density fluctuations from inflation into astronomical structures capable of hosting life. Therefore, one can predict that the bubbles with high vacuum densities are sterile, and that the non-sterile bubbles (such as the one we inhabit) will experience a vacuum density that is high enough to be a marginally suppressing structure formation (since large values are preferred). This argument was used in 1987 by Steven Weinberg to predict that we would observe a non-zero vacuum density (at a time when there was no observational evidence for it).
This approach is a violent departure from the Copernican principle, which attempts to reach cosmological inferences by the assumption that humans are not privileged observers of the universe. Although obviously false on a local scale (a point selected at random would be most unlikely to lie so close to a star), there is no reason why our existence should depend on special features of the large-scale appearance of the universe. It is therefore a reasonable supposition that, if the universe appears isotropic (the same in all directions) about our position, it would also appear isotropic to observers in other galaxies. This is quite a powerful argument, as a universe that is isotropic about all points must also be homogeneous - i.e. with a density that is constant at a given time. So this Copernican approach is fruitful within our universe, but we now explicitly break with it when it comes to the multiverse.
Is this the end of science? We can only explain features we see in the current universe around us by postulating an infinity of other universes, mostly almost devoid of life, which we can never experience directly. This is rather like the winner of the National Lottery, who appreciates that their good fortune requires the existence of millions of losers - but whose names and addresses the winner can never know. And yet there are many cases where science makes a prediction that cannot be verified directly: we know beyond reasonable doubt that the temperature at the centre of the sun is about 15 million degrees (Celsius), even though no one will ever go there with a thermometer. We believe this to be true because it is a deduction made from a physical theory that is very thoroughly tested in regimes where we can make direct observations. If detection of primordial gravitational waves should prove inflation to be true beyond reasonable doubt, then arguably we would have to accept the multiverse as a reality in the same way.
Bostrom, N. (2002) Anthropic Bias: Observation Selection Effects in Science and Philosophy, London: Routledge. http://www.anthropic-principle.com/?q=book/table_of_contents (The best philosophical treatment of anthropic issues available. Chapter 3 distinguishes different kinds of anthropic reasoning clearly and in detail.)
Carter, B. (1974) 'Large number coincidences and the anthropic principle in cosmology', in M. S. Longair (ed.) Confrontation of Cosmological Theories with Observational Data, Dordrecht: Reidel, pp. 291-8. (The paper in which the anthropic principle was baptized. Covers a very wide range of physical effects that lend themselves to anthropic explanation.)
Leslie, J. (1989) Universes, London: Routledge. (Wide-ranging discussion of world-making and other anthropically relevant topics.)
Richmond, A. (2008) 'Apocalypse Now Does The Matrix: anthropic adventures from doomsday to simulation', THINK: Philosophy for Everyone 17-18: 35-46. (Popular introduction to anthropic reasoning and its applications.)
Susskind, L. (2006) The Cosmic Landscape: String Theory and the Illusion of Intelligent Design, New York: Back Bay Books. (An introduction to the multiverse, emphasizing the possible role of string theory in allowing genuine physically distinct universes.)
Barrow, J. D. and Tipler, F. J. (1986) The Amhropic Cosntological Principle, Oxford: Oxford University Press. (The best single survey volume on anthropic physical effects.)
Carr, B. (2007) Universe or Multiverse?, Cambridge: Cambridge University Press. (A set of review articles on aspects of anthropic reasoning. A more up-to-date alternative to Barrow and Tipler 1986, shorn of some of the more extreme views found in that book.)
Carter, B. (1983) 'The anthropic principle and its implications for biological evolution', Philosophical Transactions of the Royal Society of London A: Mathematical, Physical and Engineering Sciences, 310: 347-63. (Applies anthropic reasoning to our evolutionary history. Derives 'crucial steps' formula for innovations in evolution, and argues intelligent extraterrestrials are rare.)
Earman, J. (1987) 'The SAP also rises: a critical examination of the anthropic principle', American Philosophical Quarterly 24: 307 -17. (Takes a sceptical view of the utility and scientific status of anthropic arguments, and offers particular criticisms of PAP and FAR)
Leslie, J. (ed.) (1999) Modern Cosmology and Philosophy, Amherst: Prometheus Books. (Extensive collection of key articles on cosmological and physical topics, many dealing directly with anthropic issues. Reprints Carter 1974.)
McMullin, E. (1993) 'Indifference principle and anthropic principle in cosmology', Studies in History and Philosophy of Science 24: 359-89. (Detailed survey contrasting 'Copernican' indifference or (mediocrity) conceptions and anthropic conceptions of our place in nature.)
Bostrom, N. (2004) Anthropic Principle Bibliography [book website], www.anthropicprinciple.com/?q=book/bibliography (accessed 30 April 2014). (Very comprehensive bibliography to Bostrom's Anthropic Bias.)
Richmond, A. (2010) 'The anthropic principle', Oxford Bibliographies [online bibliographical resource], www.oxfordbibliograpliies.com/view/document/obo-9780195396577/obo-9780195396577-0135.xml (accessed 30 April 2014). (Survey of anthropic works from a mainly philosophical perspective.)