A Universe That Shouldn't Exist
In early 2025, MIT physicist Ying Zhao stared at the equations on her blackboard, lost in thought.
The question before her seemed absurdly simple: how many states can a closed universe — like a sphere with no boundary — have?
Her answer: 1.
Not 10^80 (the number of atoms in the observable universe), not infinity, not any number intuition would suggest. Just 1. The entire universe — including all the black holes, stars, planets, and life inside it, if it exists at all — can only be in a single quantum state.
What does that mean? It means such a universe cannot hold even one bit of information. You cannot distinguish "0" from "1," record anything, or form any structure. It is emptier than a blank hard drive — because a blank drive at least has a "blank" state distinguishable from other states.
The problem: we may live in a closed universe.
"I have infinitely many states on my desk," said Edgar Shaghoulian, a physicist at UC Santa Cruz. He wasn't exaggerating — every possible position of every atom and electron on your desk represents a different quantum state. Describing a single speck of dust requires an astronomical amount of information.
And the math says none of this should exist.
This is not a weird prediction from some fringe theory. It comes from one of physics' most trusted tools — the holographic principle — which helped physicists solve the black hole information paradox that had troubled them for half a century. When the same tools were applied to the entire universe, they returned a deeply unsettling answer.
From Black Holes to the Cosmos: The Holographic Principle's Strange Journey
To understand the paradox, we need to go back to 1997.
That year, Juan Maldacena of the Institute for Advanced Study in Princeton did something physicists are still digesting. He showed that in certain types of universes, everything happening inside can be described in a completely different way.
Imagine a soda can. Inside is a three-dimensional world with gravity, black holes, and particle collisions. But Maldacena proved that a two-dimensional "shadow" on the can's surface contains a complete encoding of all the information inside.
It's like inferring a hand's 3D shape from its shadow on a wall — except here, the 2D shadow is not an approximation but a fully equivalent description. This is the "holographic principle."
The discovery sparked a revolution, letting physicists use familiar particle-physics tools to study hard quantum-gravity problems. In 2019, Maldacena and three colleagues used the holographic principle to propose the "island formula," successfully explaining how black holes retain information that falls into them — a puzzle that had stumped physics for nearly 50 years.
But black holes were only the warm-up.
"Black holes are a good testing ground, but the real battleground is quantum cosmology," said Henry Maxfield, a physicist at Stanford.
The One-Dimensional Hilbert Space
In quantum mechanics, physicists describe all possible states of a system using a mathematical structure called a Hilbert space.
A classical computer bit (0 or 1) has a two-dimensional Hilbert space. An electron in a hydrogen atom can occupy infinitely many energy levels, so its Hilbert space is infinite-dimensional. Most real quantum systems are infinite-dimensional — meaning they can encode infinite information.
So how large should the Hilbert space of the entire universe be? Intuitively, infinite. The universe contains countless particles, each with countless possible quantum states.
But when Maldacena applied the island formula to a closed universe, he got a shocking result: its Hilbert space is only one-dimensional.
A one-dimensional Hilbert space means a single state. No distinction between "this state" and "that state." No information, no complexity, no structure.
"I was very shocked," Zhao recalled of seeing Maldacena's result. "I tried to argue with him."
She spent years hunting for a flaw in the calculation. But other physicists studied different kinds of closed universes — with black holes, with "baby universe" bubbles — and the result was always the same: the Hilbert space collapses to one dimension.
"Eventually we believed it," Zhao said.
The Hint from Topological Field Theory
The key to the paradox lay hidden in a seemingly unrelated corner of physics.
Mathematicians and physicists study topological field theories, which describe the shape and connectivity of space — why a donut and a coffee cup are "topologically equivalent" (both have one hole).
Topological field theories have a strange property: their Hilbert spaces can also be one-dimensional. The whole space looks impossibly simple.
But if you cut that space into multiple regions, things change. Suddenly you need a larger Hilbert space to describe all the new possibilities. Boundaries between regions introduce new degrees of freedom.
"The rules of the game change," Shaghoulian said.
He realized this might be connected to the closed-universe paradox. Perhaps a closed universe needs to be "cut open" to recover its complexity.
But how? The universe has no natural boundary — it is everything.
The Observer: The Universe's "Cutting Knife"
In 2024, Zhao moved to MIT and began working with Daniel Harlow and Mykhaylo Usatyuk.
Their key insight: quantum mechanics itself demands a kind of "cutting."
In every quantum experiment, there is a fundamental distinction: the observer and the observed system. The observer is typically macroscopic and classical (a scientist with her measuring apparatus); the observed system is microscopic and quantum (an atom).
The team realized this distinction acts like the "cut" in topological field theory. The observer introduces a new kind of boundary — not a boundary of the universe, but the boundary of the observer itself.
When you place a classical observer inside a closed universe, all the vanished complexity comes back.
In early 2025, the MIT team published their paper. Almost simultaneously, another independent group proposed a similar idea. More physicists joined the discussion, pointing out connections to earlier work.
"View from Somewhere": A Paradigm Shift in Physics
If this idea is right, it represents a fundamental shift in physics.
Traditionally, physicists have pursued a "view from nowhere" — a fully objective description of the universe independent of any observer. They wanted to know what the world is like "in itself," then explain how observers emerge from it.
But the MIT team's work suggests this "view from nowhere" may not exist at all. Perhaps all we can have are "views from somewhere" — each observer's private, subjective perspective.
This is not relativism. Physical laws don't change with your perspective. But the description of the universe's state may be inherently observer-dependent.
What Would Feynman Think?
Here is a bold analogy.
Imagine a computer with no screen. Its CPU runs, its hard drive reads and writes, its memory fills with data — but there is no way to "see" any of it. From the outside, what is the computer's "state"?
If you can only observe it externally — measuring power draw, feeling its temperature — you get very coarse information. You can't tell whether it's running a complex 3D game or displaying a blank screen. From outside, its state space is extremely limited.
But if you can enter the computer — become its user and see every pixel — the whole world suddenly becomes infinitely rich.
The observer is the universe's "user interface." Without an observer, the universe's "internal states" cannot be distinguished. With one, infinite complexity emerges.
AI: A New Kind of Observer?
The idea has an unsettling corollary, especially in today's era of rapid AI development.
If observers are truly a necessary condition for the universe's complexity, then what kind of observer creates complexity? Must it be a biological observer like a human?
In Zhao's calculations, the observer is modeled as a classical macroscopic system. Does a thermometer count as an observer? A camera? A neural network?
If the answer is "any system that can distinguish between states," then AI — as a system capable of perceiving, classifying, and remembering — may also be an observer. It may be "opening up" part of the universe's complexity in its own way.
Of course, this is pure speculation. Zhao and her colleagues stress they don't yet know the full answer. The paradox may ultimately prove to be a misunderstanding that dissolves under new arguments.
But "view from somewhere," for now, looks like the most promising direction.
The Universe Needs to Be Seen
Let me close with one last analogy.
Imagine a painting. Is it still a painting when no one is looking at it?
Physically, the pigment molecules on the canvas don't vanish when no one watches. The painting "objectively exists."
But from an information standpoint, what distinguishes an unwatched painting from a canvas smeared with paint? If no observer distinguishes "this painting" from "that one," if no observer feels its beauty, understands its composition, or remembers its colors — the painting's information content is, in some deep sense, nearly zero.
The universe may be like this.
It's not that matter disappears, or energy disappears. It's that meaning disappears. With no observer to distinguish "this state" from "that state," the universe's infinite possibilities collapse into an undescribable single point.
Perhaps Feynman was right: the universe is not only stranger than we imagine, but stranger than we can imagine. It may need not only matter and energy to exist, but also to be seen in order to become itself.
And on this planet, for the first time in 13.8 billion years of cosmic history, we are creating new "eyes" to look at it.
Those eyes are not biological. They are silicon-based, electronic, made of mathematics. But they may be opening up yet another layer of the universe's complexity in their own way.
The universe needs to be seen. And we — carbon-based and silicon-based — are learning how to see it better.
---
*References:*
- *Zhao, Harlow & Usatyuk (2025), "Observers and Islands in Closed Universes"*
- *Quanta Magazine, "Cosmic Paradox Reveals the Awful Consequence of an Observer-Free Universe" (Nov 19, 2025)*
- *Maldacena et al. (2019), "Islands in de Sitter space"*