pbsspacetime
patreon
Why The Multiverse Could Be Real (script)
🕑 Added 2025-08-14 15:07:43 +0000 UTCThe multiverse pops out of quite a few theories in physics, and has been proposed as a solution to certain vexing problems. But it’s also been argued that the very idea of a multiverse is just bad science. That it’s unfalsifiable and a dead-end to inquiry and as bad a violation of Occam’s razor as you could imagine. But the multiverse might also exist. Can something that exists be bad science?
—-
In recent episodes we’ve talked about how the fundamental parameters of our universe are, apparently, finely tuned for the production of entities that can speculate about the fine-tuning of the universe. One possible explanation that we’ve teased several times now is that there exists a preposterously large ensemble of universes spanning all possible parameters, both habitable and uninhabitable—the latter being the extreme majority. We are of course in an inhabitable one, by necessity. Some would say that this anthropic principle—this argument of an observer bias—actually predicts that a multiverse exists. Others say that the multiverse is just bad science.
Before we decide one way or another, let’s review the types of multiverse and where they come from. And before we review multiverses, let’s remind ourselves what we mean by uni-verse. The original meaning of the word was the totality of all existence, which makes the idea of a multiverse nonsensical. If it exists, it’s in the universe. But in modern usage, “universe” tends to mean this particular continuous spacetime that traces back to a particular big bang. For that reason we have more concrete terms. For example, the observable universe refers to everything we can see in that spacetime given the amount of time light has had to travel to us, whose outer boundary we call the particle horizon. But if you could teleport just beyond the particle horizon, presumably things would be pretty similar—more stars, more galaxies, etc. whose light hasn’t reached us yet---and we typically think of those regions beyond as being part of our universe. So maybe the universe is everything connected to the spacetime that we live in.
But there are conditions that might justify calling some distant but connected region a different universe. For example, if the laws of physics change over extremely long distances, there could be regions of a vast, connected spacetime that operate very differently to here. This is often called a .. or quilt multiverse. A related multiverse comes from the eternal inflation hypothesis, where the “main” greater spacetime is expanding vastly more quickly than our universe. Bubbles of more gently expanding space appear and become universes like our own. In this case, these bubbles may also have different laws of physics to each other. Other ways to generate universes with different physics include: birthing them inside black holes as in Lee Smolin’s cosmological natural selection; or various cyclic universe options in which the laws shift between cycles.
There’s also the multiverse of the Many Worlds interpretation of quantum mechanics. Some of what I get to today is relevant to that, but I’m going to focus on the multiverses that allow varying laws of physics.
There are concrete reasons that the laws of physics might change between universes. For example, changing the quantum vacuum state changes the physics. In our universe, the mass of the Higgs boson, and hence of all other particles, depends on the slightly non-zero energy of the Higgs field ground state. But the ground state of a given field may be a local minimum, and other ground states are possible. Change the Higgs ground state changes the Higgs mass and our universe looks very different. For almost all random choices of a Higgs mass the universe is uninhabitable.
Another example where the laws of physics change depending on the configuration of the underlying fields is in string theory. There, the machinery generating all particles and forces is in the form of compactified extra dimensions, coiled into something called a Calabi-Yau manifold. The geometry and topology of this manifold and the corresponding string vibrational modes determine the properties of particles and strengths of interactions. This leads to an enormous number of possible configurations for the ground or vacuum state—at least 10^500 of them in fact—each producing a universe with different physics. This is the string landscape—a colossal landscape of possible universes that may or may not actually exist “out there”.
String theory can’t tell us why our universe’s Calabi-Yau manifold landed on the particular vacuum state that generates the particles and forces of the standard model. Nor can the standard model of particle physics by itself tell us why the underlying quantum fields have their particular vacuum states. It would be nice to find such a mechanism, but what if there just isn’t one? What if all possible vacuum states can happen, and we just happen to be in one of the versions capable of producing life?
That’s the anthropic principle. In its most reasonable form, it argues that we shouldn’t be surprised to find ourselves in an unusual part of the uni-slash-multiverse because we ourselves are unusual. At some levels that’s not in the least controversial. For example, we find ourselves in a very rare environment in our universe: a habitable biosphere. Even though habitable biospheres occupy an infinitesimal fraction of the volume of the universe, we’re not surprised to find ourselves in one. We have to have emerged somewhere that we can have emerged. Extended to the multiverse if there is one, it should be equally unsurprising that we emerged in a universe that can produce habitable biospheres, which in turn requires a very specific configuration of the laws of physics. That seems to be the case—change the strength of dark energy or the mass of the Higgs boson or even the number of dimensions of spacetime and our universe likely would not have produced life.
Coupled with the anthropic principle, the multiverse gets us out of having to explain “why these specific laws”, and in a way also gives an explanation for why the physics of our universe seem finely tuned to allow life. It’s a selection bias, just like the bias that leads to us observing the universe from an unusually warm, wet, biosphere. Some would take this much further, and say that this apparent fine-tuning of our universe’s parameters in favor of habitability is actual evidence for the multiverse.
So now that we have the multiverse and the anthropic principle very loosely defined, let’s decide if this is bad science. We’re going to break this into two broad categories of criticism. One is that the multiverse is unparsimonious—it violates Occam’s razor. Two is that the multiverse is scientifically useless in various ways—it’s unfalsifiable and/or an explanatory dead-end.
Let’s start with one. The principle of parsimony, more often called Occam’s razor, can be stated in several ways, but let’s start with one of the phrasings used by William of Occam himself: “Never posit pluralities without necessity.” Hm. A multiverse is about as plural as you can get, so maybe we’ve already annoyed Brother William.
Occam’s edict has been refined over the years. Another common phrasing is “Entities must not be multiplied beyond necessity” Again, sounds damning, but we can start to pick this apart. What’s an entity in this context? Isaac Newton offers some guidance: “We are to admit no more causes of natural things than such as are both true and sufficient to explain their appearances.”
This is a critical refinement: we’re told to limit the plurality of causes in an explanation, which is very distinct from limiting, say, the plurality of predictions. Way before William of Occam we have some enlightening phrasing by Aristotle "We may assume the superiority, other things being equal, of the demonstration which derives from fewer postulates or hypotheses.” A more modern version is: “All else being equal, simpler explanations should be preferred over more complex ones.”
So the thing we’re advised against multiplying is not the outcome or predictions of a hypothesis, but rather the number of “entities” in the explanation, the “causes” not the predictions, according to Newton. And what are these entities?
They’re the number of postulates or assumptions. Basically, the number of new things required to make the hypothesis work. These could be principles like a law of physics or an initial condition of the universe, or they could be physical stuff, like a new quantum field or a multiverse. In general, Occam’s razor wants us to be reductionists. To either minimise the complexity of the underlying theoretical framework or to minimize the number of fundamental building blocks—to seek theoretical or ontological parsimony. In fact, these two may ultimately be the same thing.
Occam’s razor basically warns against overfitting. If you want to find a mathematical description of some arbitrary curve, you can always do it if you have enough degrees of freedom in your function. For example, a a Fourier series with an arbitrarily large number of terms can fit any continuous curve. But the resulting mathematical form has no explanatory power—it’s not a general law of nature that can make predictions beyond its overfitted curve. We could equally describe the universe as a vast map of points indicating the locations of all particles, but that’s complex and useless compared to the relative parsimony of the standard model plus general relativity and some initial conditions.
So, with that refinement of Occam’s razor, where does the multiverse stand?
In cases where a theory predicts the existence of a multiverse, it seems that Occam’s razor doesn’t apply. Especially if the theory was concocted to do a different job and the multiverse just popped out. In that case, the multiverse is not a prior assumption for the theory—it extends from it. So things like the string landscape, the eternal inflation multiverse, and honestly even the splitting timelines of Many Worlds shouldn’t be considered a problematic plurality when it comes to assessing these ideas.
When we bring in anthropic arguments the situation is muddier. We could say that the multiverse is predicted by anthropic arguments—-in that fine-tuning plus selection bias demands many untuned universes. But we could also frame the multiverse as a cause or an assumption in our argument: we observe fine tuning because there’s a multiverse.
In the latter case, we need to pick apart our principle of parsimony a bit more carefully. If we propose or assume or postulate a multiverse as an explanation, have we multiplied entities beyond necessity? Have we overfit?
Some would say yes. I don’t agree. It’s tempting to want to assume the least extravagant universe possible, but the size or number of universes doesn’t map to the number of new, prior assumptions that we would use to weigh our hypothesis against Occam’s razor. And the impulse to argue for the least extravagant universe has failed us in the past.
The ancients rejected the idea that the Earth is moving around the Sun because, if that were so, then the distant stars should appear to change position relative to each other due to parallax. They didn’t consider that it was even reasonable for the stars to be so far away that parallax was too small to observe. Their bias towards smallness led them astray. In the great debate in 1920, Heber Curtis argued that the redshift of the spiral nebulae meant they had to be entire galaxies far beyond the Milky Way, while Harlow Shapely represented the status quo feeling that such distances would mean an uncomfortably large universe. Now we’re completely comfortable with an observable universe 93-billion light years in diameter and a greater universe hundreds of times larger, or even infinitely large. So as our measured boundary of the universe expands, so does our acceptance of new, increasingly preposterous sizes.
For this reason, it’s no longer compelling to argue against the multiverse based only on a sense of excessive bigness. If that bigness encompasses regions where the constants of nature shift or stretch to regions where inflation never stopped … that’s not necessarily unparsimonious due to the sheer bigness. Occam’s razor asks that we seek the fewest moving parts, but it doesn’t say anything about the sizes of those parts.
So, the parsimony of a given multiverse theory is equivalent to the parsimony of the theory that predicts the multiverse, which has little to do with the multiverse that it predicts. If eternal inflation or string theory or cosmological natural selection or whatever are not too far fetched as solutions to fine tuning then neither is the multiverse.
OK, on to the next point. Is the multiverse bad science because it’s unfalsifiable or an explanatory dead end. I’m only going to touch on this one, because there are details that warrant spending a lot of time on this, which we don’t have.
So the badness of multiverse science depends very much on your definition of science. If you adhere to a strict Popperian view that science is only that which can be falsified—disproved by doable experiments—then you could argue that the multiverse is not science. But by the same logic you would argue that hypothesizing that galaxies exist beyond the particle horizon is bad science because that too can’t be tested. A more relaxed view would say that science applies logical reasoning coupled with a whole suite of tools to collectively find a consistent picture of what reality is. If something is potentially a part of reality, then it’s in principle approachable by scientific methods. So, if the multiverse might exist then it’s very much in the realm of science.
And, in fact, it’s not true that we can’t test multiverse theories. There are a few proposals for direct tests for other universes that I won’t get into here although we’ve talked about a couple of them before. But there are also tests that involve predicting what we should see in this universe if its unusual properties really are a result of anthropic selection. For example, Steven Weinberg showed that anthropic arguments gave a pretty clear prediction for the value of dark energy under the assumption that our universe should have the most likely value for dark energy that could permit our existence. The broader topic of anthropic predictions is deep and fascinating, and I’ll also save it for another time.
The fact that various multiverse ideas are testable given sufficient creativity means they also aren’t scientific dead ends. It’s not true to say that, by allowing anything to be possible, we no longer have a path to explaining our universe because all universes exist. In fact, the foundational theory that allows a multiverse to exist will make pretty specific predictions about the distribution of properties of the universes in that multiverse. And that can place strong constraints on the likely properties of our universe under anthropic selection.
So, the multiverse doesn’t have to be bad science, but it also can be. If the idea is proposed as a blanket answer to fine-tuning without a concrete mechanism that can be used to make other, testable predictions then your proposition is indeed unparsimonious, unfalsifiable, and an explanatory dead end. Bad science. But that’s not how serious proponents of various multiverse-generating theories do it—treated with due care, we can indeed science the prospect of a plurality of spacetime.
Comments
Tanj
There is another way to be a multiverse. Special relativity. Every frame is its own 'verse. And SR is possibly the most tested theory ever, so it is not controversial, right? The puzzles I know are not surprising from the frame of reference of the prime actors. If a photon is emitted by an atom in the middle of a sphere, and a single atom on the sphere is the end point, those two ends have no uncertainty over why none of the other zillion atoms did not participate or what caused their reality to "collapse". The origin and destination were at zero interval from each other, sharing their frames, and the energy shifted. At some later time observers may notice the event, but the event itself was inherently collapsed. It just becomes an uncertainty and a puzzle for other frames in their own 'verse to observe. So you can explain the puzzles with a single universe emergent from the myriad multiverse of infinitesimal reference frames. Which have been the consequence of relativity for well over a century.
