Prompted by reading about the recent Munich conference on the philosophy of science, I am reminded that many people regard the idea of a multiverse as so wild and wacky that talking about it brings science into disrepute. My argument here is the reverse: that the idea of multiple Big Bangs, and thus of a multiverse, is actually more mundane and prosaic than the suggestion that there has only ever been one Big Bang. I’m calling this a “philosophical” argument since I’m going to argue on very general grounds rather than get into the details of particular cosmological models.
First, let me clarify that several different ideas can be called a “multiverse”, and here I am concerned with only one. That “cosmological multiverse” is the idea that our Big Bang was not unique, but rather is one of many, and that the different “universes” created by each Big Bang are simply separated by vast amounts of space.
Should we regard our Big Bang as a normal, physical event, being the result of physical processes, or was it a one-off event unlike anything else, perhaps the origin of all things? It is tempting to regard it as the latter, but there is no evidence for that idea. The Big Bang might be the furthest back thing we have evidence of, but there will always be a furthest-back thing we have evidence of. That doesn’t mean its occurrence was anything other than a normal physical process.
If you want to regard it as a one-off special event, unlike any other physical event, then ok. But that seems to me a rather outlandish idea. When physics encounters a phenomenon, the normal reaction is to try to understand it in terms of physical processes.
So what does the evidence say? We know that our “observable” universe is a region of roughly 13.8 billion light years in radius, that being the distance light can have traveled since our Big Bang. (Actually, that’s how we see it, but it is now bigger than that, at about 90 billion light years across, since the distant parts have moved away since they emitted the light we now see.) We also know that over that time our observable universe has been steadily expanding.
At about 1 second after the Big Bang, what is now our observable universe was only a few light years across, and so would have fitted into (what is now) the space between us and the nearest star beyond our Sun. Before that it would have been yet smaller.
We can have good confidence in our models back to the first seconds and minutes, since the physics at that time led to consequences that are directly observable in the universe today, such as the abundance of helium-4 relative to hydrogen, and of trace elements such as helium-3, deuterium, and lithium-7. Before that time, though, our knowledge gets increasingly uncertain and speculative the further back we push.
That size at a mere 1 second old was still very large compared to its (presumed) origins. We need to think logarithmically, in terms of factors-of-ten reductions in size, and there are still a lot of those to go before we’d get to the “Big Bang” itself. In terms of orders of magnitude, in going back to 1 second we’re less than halfway there.
One could, if one likes, try to extrapolate backwards to a “time = zero” event at which all scales go to zero and everything is thus in the same place. But trying to consider that is not very sensible since we have no evidence that such an event occurred (from any finite time or length scale, extrapolating back to exactly zero is an infinite extrapolation in logarithmic space, and making an infinite extrapolation guided by zero data is not sensible). Further, we have no physics that would be remotely workable or reliable if applied to such a scenario.
So what is it sensible to consider? Well, as the length scale decreases, quantum mechanics becomes increasingly important. And quantum mechanics is all about quantum fluctuations which occur with given probabilities. In particular, we can predict that at about the Planck scale of 10−35 metres, quantum-gravity effects would have dominated. We don’t yet have a working theory of quantum gravity, but our best guess would be that our Big Bang originated as a quantum-gravity fluctuation at about that Planck-length scale.
So, we can either regard our Big Bang as an un-natural and un-physical one-off event that perhaps originated absolutely everything (un-natural and un-physical because it would not have been a natural and physical process arising from a pre-existing state), or we can suppose that our Big Bang started as something like a quantum-gravity fluctuation in pre-existing stuff. Any physicist is surely going to explore the latter option (and only be forced to the former if there is no way of making the latter work).
At times in our human past we regarded our Solar System as unique, with our Earth, Sun and Moon being unique objects, perhaps uniquely created. But the scientific approach was to look for a physical process that creates stars and planets. And, given a physical process that creates stars, it creates not just one star, but oodles of them strewn across the galaxy. Similarly, given a physical process that creates Earth-like planets, we get not just one planet, but planets around nearly every star.
It was quite wrong to regard the Sun and Earth as unique; they are simply mundane examples of common physical objects created by normal physical processes that occur all over the galaxy and indeed the universe.
But humans have a bias to a highly anthropocentric view, and so we tend to regard ourselves and what we see around us as special, and generally we need to be dragged kicking and screaming to the realisation that we’re normal and natural products of a universe that is much the same everywhere — and thus is strewn with stars like our Sun, with most of them being orbited by planets much like ours.
Similarly, when astronomers first realised that we are in a galaxy, they anthropocentrically assumed that there was only one galaxy. Again, it took a beating over the head with evidence to convince us that our galaxy is just one of many.
So, if we have a physical process that produces a Big Bang then likely we don’t get just one Big Bang, we get oodles of them. No physical process that we’re aware of happens once and only once, and any restriction to one occurrence only would be weird and unnatural. In the same way, any physical process that creates sand grains tends to create lots of them, not just one; and any physical process that creates snowflakes tends to create lots of them, not just one.
So, we have three choices: (1) regard the Big Bang as an unnatural, unphysical and unexplained event that had no cause or precursor; (2) regard the Big Bang as a natural and physical process, but add the rider that it happened only once, with absolutely no good reason for adding that rider other than human parochial insularity; or (3) regard the Big Bang as a natural and physical event, and conclude that, most likely, such events have occurred oodles of times.
Thus Big Bangs would be strewn across space just as galaxies, stars and planets are — the only difference being that the separation between Big Bangs is much greater, such that we can see only one of them within our observable horizon.
Well, I don’t know about you, but it seems to me that those opting for (3) are the ones being sensible and scientifically minded, and those going for (1) or (2) are not, and need to re-tune their intuition to make it less parochial.
So, let’s assume we have a Big Bang originating as a quantum-gravity fluctuation in a pre-existing “stuff”. That gives it a specific length scale and time scale, and presumably it would have, as all quantum fluctuations do, a particular probability of occurring. Lacking a theory of quantum gravity we can’t calculate that probability, but we can presume (on the evidence of our own Big Bang) that it is not zero.
Thus the number of Big Bangs would simply be a product of that probability times the number of opportunities to occur. The likelihood is that the pre-existing “stuff” was large compared to the quantum-gravity fluctuation, and thus, if there was one fluctuation, then there would have been multiple fluctuations across that space. Hence it would likely lead to multiple Big Bangs.
The only way that would not be the case is if the size of the pre-existing “stuff” had been small enough (in both space and time) that only one quantum fluctuation could have ever occurred. Boy, talk about fine tuning! There really is no good reason to suppose that.
Any such quantum fluctuation would start as a localised event at the Planck scale, and thus have a finite — and quite small — spatial extent. Its influence on other regions would spread outwards, but that rate of spreading would be limited by the finite speed of light. Given a finite amount of time, any product of such a fluctuation must then be finite in spatial extent.
Thus our expectation would be of a pre-existing space, in which there have occurred multiple Big Bangs, separated in space and time, and with each of these leading to a spatially finite (though perhaps very large) universe.
The pre-existing space might be supposed to be infinite (since we have no evidence or reason for there being any “edge” to it), but my argument depends only on it being significantly larger than the scale of the original quantum fluctuation.
One could, of course, counter that since the initial quantum fluctuation was a quantum-gravity event, and thus involved both space and time, then space and time themselves might have originated in that fluctuation, which might then be self-contained, and not originate out of any pre-existing “stuff”. Then there might not have been any pre-existing “stuff” to argue about. But if quantum-gravity fluctuations are a process that can do that, then why would it happen only once? The natural supposition would be, again, that if that can happen once, then — given the probabilistic nature of physics — it would happen many times producing multiple different universes (though these might be self-contained and entirely causally disconnected from each other).
In order to explain various aspects of our observed universe, current cosmological models suggest that the initial quantum fluctuation led — early in the first second of its existence — to an inflationary episode. As a result the “bubble” of space that arose from the original quantum-fluctuation would have grown hugely, by a factor of perhaps 1030. Indeed, one can envisage some quantum-gravity fluctuations leading to inflationary episodes, but others not doing so.
The inflationary scenario also more or less requires a multiverse, and for a similar reason to that given above. One needs the region that will become our universe to drop out of the inflationary state into the “normal” state, doing so again by a quantum fluctuation. Such a quantum fluctuation will again be localised, and so can only have a spatially finite influence in a finite time.
Yet, the inflationary-state bubble continues to expand so rapidly, much more rapidly than the pocket of normal-state stuff within it, that its extent does not decrease, but only increases further. Therefore whatever process caused our universe to drop out of the inflationary state will cause other regions of that bubble to do the same, leading to multiple different “pocket universes” within the inflationary-state bubble.
Cosmologists are finding it difficult to construct any model that successfully transitions from the inflationary state to the normal state, that does not automatically produce multiple pocket universes. Again, this follows from basic principles: the probabilistic nature of quantum mechanics, the spatial localisation of quantum fluctuations, and the finite speed at which influence can travel from one region to another.
The dropping out of the inflationary state is what produces all of the energy and matter that we now have in our universe, and so effectively that dropping-out event is what we “see” as our Big Bang. This process therefore produces what is effectively a multiverse of Big Bangs strewn across that inflationary bubble. Thus we have a multiverse of multiverses! Each of the (very large number of?) quantum-gravity fluctuations (that undergo an inflationary state) then itself produces a whole multiverse of pocket universes.
The point I am trying to emphasize is that any process that is at all along the lines of current known physics involves the probabilistic nature of quantum mechanics, and that means that more or less any conceivable process for creating one Big Bang is going to produce not just a single event, but almost inevitably a vast number of such events. You’d really have to try hard to fine-tune and rig the model to get only one Big Bang.
As with any other physical process, producing multiple Big Bangs is far more natural and in-line with known physics than trying to find a model that produces only one. Trying to find such a model — while totally lacking any good reason to do so — would be akin to looking for a process that could create one snowflake or one sand grain or one star or galaxy, but not more than one.
But, the critics exclaim, these other Big Bangs and other universes are so far away that we can never — even in principle — gain any information from them. We can never have empirical evidence of their existence, and therefore it is utterly unscientific to speculate about them. Doing so brings science into disrepute and opens the door to pseudo-sciences.
My reply is several-fold. First, it may well be true that we can never gain empirical evidence about other hyper-distant universes. But, given that we know we have one Big Bang — our own — should we then believe that it is the only one? Or should we consider that it is likely one of many? Or should we just not ask the question, as being unanswerable?
I’m suggesting that the first of those is the least sensible, noting that a claim that there was only ever one Big Bang just as much exceeds the evidence as the claim that we live in a multiverse. The last option is sensible and cautious, though what we know about physics seems to make the middle option more likely than not.
Further, in science, we should not interpret the demand for empirical evidence too narrowly. Science is about doing our best to understand the world around us using whatever we have available. The strong focus on empirical evidence is then a product of scientific enquiry, since that focus has been found to give the best results. Yet science is inevitably pragmatic: it’s up to science to adjust itself to how the world is, following evidence and reason to that end, rather than make preconditions about what we will accept.
Science is about constructing models to explain the empirical evidence, and we can have some justification in extrapolating successful models to scenarios where we have no direct empirical evidence. For example, you are presumably in no doubt that you had a great-great-500-greats grandmother. And yet you’d be pretty hard pushed to find any empirical evidence at all about that specific person. Your belief would thus be an extrapolation from a model of how things work, a model that has been so well verified by evidence that you are justified in making that extrapolation.
The existence of the 500-greats grandmother is not itself the hypothesis that we are trying to test. Rather, that ancestor is a consequence of a much wider and more general theory about how the world is. We can test that wider theory sufficiently well that we can then have confidence in its implications. In the same way the idea of a cosmological multiverse is not itself a standalone hypothesis, rather it is a consequence of our understanding of cosmology and of physics more generally, and we can test those cosmological models against our own observable universe, sufficiently so that we can then have confidence in their implications.
Thus there is nothing in-principle unscientific about making statements that we have no hope of ever verifying by direct empirical evidence. We need to distinguish between what it takes for a concept to be “scientific” and what it takes for that concept to be proven true. The multiverse idea can still be scientifically valid, even if we have no possibility of directly proving it.
The evidence for a multiverse is, of course, weak. But the evidence that there was only ever one Big Bang is weaker! If you think we should simply declare that, since we’ve seen only one Big Bang, then likely there is only one, then realise that such a declaration would have been wrong about suns, moons, solar systems and galaxies. On every occasion in the past, it is those arguing that the universe is bigger than we previously thought who turned out right.
Everything we know about how physics works tells us that having multiple Big Bangs is more natural and more likely than having only one. Thus, it is entirely scientific to adopt the tentative conclusion that it is more likely than not that we live in a multiverse.
 The abundances of the light elements are one of the strong pieces of evidence for the Big Bang model.
 Matt Strassler argues against extrapolating to a singularity here.
 This was the “Shapely–Curtis” or “Great Debate” among astronomers in the early 1920s.
 This has been argued by, for example, Lawrence Krauss in his book A Universe from Nothing.
 For a discussion of the inflationary model of cosmology and the argument that it inevitably produces a multiverse, see for example this article by Alan Guth.