The multiverse has a mathematical difficulty that its own architects have spent years working on, and almost nobody outside cosmology has heard of it.
It is not an objection from critics. It was not raised by people who dislike the idea. It emerged from taking the theory seriously and trying to extract a prediction from it, which is what physicists do to a framework they consider promising.
This article explains it properly, because the popular versions either omit it or misrepresent it as a refutation, and it is neither omissible nor a refutation.
Where the multiverse comes from
First, why serious people hold this view, because a difficulty means nothing if the position is a straw man.
Inflationary cosmology proposes that the very early universe underwent a period of extraordinarily rapid expansion. The theory was developed to solve specific problems, including why the universe appears flat and why widely separated regions have nearly identical temperature, and it has observational support in the structure of the cosmic microwave background.
Here is the part that produces a multiverse. In many versions of the theory, inflation is eternal: it never stops globally. It ends locally, in pockets that cool and settle into ordinary universes, while the space between those pockets continues inflating faster than the pockets can fill it. The result is an unlimited number of bubble universes, each one internally infinite, embedded in a space that goes on producing more.
Nobody designed this to explain fine-tuning. It fell out of a theory built for other reasons, which is exactly the pedigree a hypothesis wants. If the constants vary between bubbles, the apparent fine-tuning of our own has a candidate explanation that requires no intention.
That is the position at full strength, and it is a serious one.
The problem: what does probable mean here?
Now try to get a prediction out of it.
The multiverse is supposed to explain why we observe the constants we do: most bubbles are inhospitable, ours is one of the rare hospitable ones, and observers necessarily find themselves in hospitable ones. For that to be an explanation rather than a story, it has to say something about what a typical observer sees.
Which requires counting. And counting is where it fails.
There are infinitely many bubbles. Infinitely many of them contain observers who see one set of constants; infinitely many contain observers who see another. To say that one outcome is more probable than another, you need a ratio of two infinite quantities, and that ratio is not defined by the physics. It depends entirely on how you count: what order you take the bubbles in, where you draw a cutoff before letting it go to infinity, which slicing of an infinite spacetime you treat as the natural one.
Different choices give different answers. Not slightly different: arbitrarily different. A prescription for how to count is called a measure, and the theory does not come with one.
This is the measure problem, and I want to be precise about its severity. It is not that the calculation is hard. It is that without a measure, the sentence "this outcome is probable in the multiverse" does not have a determinate meaning, and therefore cannot be true or false yet.
Boltzmann brains
The problem is not abstract, and here is the result that made it urgent.
Under some measures, a strange prediction appears. Over sufficiently long times, thermal and quantum fluctuations in a vacuum can produce, extremely rarely, an organized structure. Extremely rarely, over an unbounded time, still produces unlimited numbers. Among those structures are ones complex enough to constitute an observer with a moment of experience and a set of false memories: a brain that fluctuated into existence, had an experience, and dissipated.
Under some measures, such observers vastly outnumber observers like us, who arose through the long, thermodynamically expensive route of stars, planets and evolution.
If that is what the theory predicts, the theory is in trouble, and not for aesthetic reasons. If typical observers are fluctuations with false memories, then your memories are probably false, including the memories of evidence that led you to accept the theory. The theory would be undermining its own evidential basis. This is called cognitive instability, and it is a genuine reason to reject a measure that produces it.
What the physicists actually did about it
Here is the part that matters most for how you should read this whole article.
The people who built inflationary cosmology treated this as a problem to be solved rather than as an embarrassment to be managed. Work by De Simone, Guth, Linde, Vilenkin and colleagues examined whether particular measures avoid Boltzmann brain domination, and found that a specific proposal, the scale-factor cutoff measure, does avoid it, provided certain requirements about the landscape of possible vacua are satisfied.
Read that result carefully, because both halves are load-bearing.
The problem is not fatal. There is at least one measure under which the pathology does not arise.
And the result is conditional. It holds if assumptions about the landscape hold, and those assumptions are not currently established. It is progress on an open question rather than a closure of it.
That is what genuine work on a hard problem looks like, and anyone reporting only the first half or only the second is misrepresenting it.
What this does and does not show
Four things, stated separately because they get run together.
It does not refute the multiverse. A framework with an unsolved technical problem is a normal framework. Almost every productive theory in physics had one.
It does mean multiverse explanations of fine-tuning are currently incomplete. If you cannot say what is probable, you cannot yet say that our constants are unsurprising given many trials. The explanatory move requires the measure, and the measure is what is missing.
It applies to the competing explanation too. Design arguments claiming our constants are improbable by chance also require a probability distribution over possible universes, and no such distribution has been established either. The difficulty is symmetric, and using it against one side while ignoring the other is borrowing half an objection.
It is being worked on, from inside. This is not a stalled question. It is an active research area whose participants include the framework's founders.
Why this belongs in a philosophical argument
Because it is a case study in something rarer than either the enthusiasts or the critics of science suggest.
A community adopted a promising framework, followed it to a place where it produced an absurd result, named the absurdity precisely, and set about determining under what conditions it could be avoided, publishing conditional results rather than declaring victory. That is intellectual seriousness, and it is worth admiring regardless of what one thinks about multiverses.
It also means the honest current position on cosmological fine-tuning is that the leading naturalistic explanation is incomplete in a specific, identified way, and so is its main rival. Anyone offering you a settled answer on either side is not reporting the state of the field.
Sitting with an unresolved difficulty for a hundred pages is a strange thing to ask of a reader, and it is what Letter Six of the Dear Musk series does, at US$ 7.99. Whether that is bearable is worth testing on something free first: Letter One is given whole, with nothing to enter.
Dear Musk is an independent work of philosophical writing addressed to a public figure, in the long tradition of open letters. It is not affiliated with, endorsed by, sponsored by, or connected to Elon Musk or any company associated with him. See the publisher's Terms & Conditions.Continue reading
- Fine-Tuning, Stated Without a Thumb on the Scale — The same difficulty applied to both explanations
- Materialism, Multiverse, and Nihilism Under One Standard — How this compares with the other positions
- Science, Scientism, and the Limits of an Instrument — What a method can and cannot settle