The Simulation Hypothesis: Why It’s Almost Certainly Wrong

A 38-source analysis that presents the best arguments for the simulation hypothesis in their strongest form, then dismantles them with physics, logic, and the critics most pro-simulation content skips over.

By Scott Covert — April 2026

Last Updated: May 24th, 2026
Polaroid photograph of nested photographs developing in a darkroom tray — the visual metaphor of recursive worlds-within-worlds that the simulation hypothesis invokes

In This Analysis

  1. The Steel-Man Case (What Proponents Actually Argue)
  2. Why The Physics Kills It
  3. Why The Logic Kills It
  4. The Cherry-Picked Evidence
  5. It’s Functionally Identical to Religion
  6. The 38 Sources — What They Actually Say
  7. The Bottom Line

Elon Musk says the odds we’re in base reality are “one in billions.” Neil deGrasse Tyson once gave it better than 50-50. TikTok and YouTube are saturated with videos presenting quantum mechanics as “proof” that reality is a video game.

When something looks too trendy to be true, I go to the sources: thirty-eight of them this time, from Bostrom’s original 2003 paper to Nobel laureate physicists, cognitive scientists, Reddit threads that turned out sharper than the peer-reviewed papers, and YouTube explainers ranging from brilliant to delusional.

The best arguments for the simulation hypothesis turned out to be interesting, and they deserve to be heard in their strongest form, because that’s where their problems become most obvious.

I went into this as a skeptic, and I built my research process to challenge that instinct rather than confirm it. After 38 sources, I’m more skeptical than when I started.

Part I

Before I Take This Apart, Here’s the Strongest Case Anyone Has Made For It

If you’re going to dismantle an argument, you start by taking its strongest version seriously. Here is the pro-simulation case at its best.

The 2003 Paper That Started All of This

Oxford philosopher Nick Bostrom published the original version of this argument in 2003, years before Musk ever mentioned it. He didn’t claim outright that we’re simulated; he framed it as a trilemma, where one of three propositions has to be true.

(1) Civilizations almost always go extinct before reaching the technological capability to run detailed ancestor simulations.

(2) Advanced civilizations that could run ancestor simulations almost never choose to.

(3) We are almost certainly living in a simulation.

If even a tiny percentage of posthuman civilizations run ancestor simulations, simulated beings would vastly outnumber real ones. You’d be statistically foolish to assume you’re one of the rare “real” ones.

This is sound logic if you accept one critical assumption: that consciousness is substrate-independent, meaning it can arise from silicon just as well as from neurons. That assumption is completely unproven.

“I don’t see any strong argument for which proposition is true.”

— Nick Bostrom, who distributes his credence roughly equally across all three options (~20% each for simulation being true)

Musk Says the Odds Are One in Billions. Here’s His Actual Math.

Musk’s version is simpler and more intuitive: “If you assume any rate of improvement at all, games will eventually be indistinguishable from reality.” We went from Pong to photorealistic VR in 40 years, and extrapolated forward, he estimates the odds of being in base reality at “one in billions.”

The appeal is obvious, but it conflates visual fidelity with physics simulation. A video game renders what you’re looking at. Reality requires simulating everything, everywhere, simultaneously, down to the quantum level. That gap is a mathematical impossibility, not a technology problem, and the next section shows the actual math.

The Universe Looks Rigged. That’s the Real Hook.

The universe’s fundamental constants appear suspiciously tuned for life. Change the strong nuclear force by 1% and atoms don’t form. Change the cosmological constant slightly and the universe either collapses instantly or expands too fast for galaxies to coalesce, making it look designed.

Simulation proponents argue someone “set the parameters,” but fine-tuning has at least four explanations: coincidence, the multiverse, God, or simulation. It doesn’t uniquely support any of them. The anthropic principle offers a simpler answer: we can only observe a universe compatible with our existence. If the constants were different, we wouldn’t be here to notice.

The Physicist Who Says He Found Actual Proof

Melvin Vopson at the University of Portsmouth claims to have found empirical evidence. His “Second Law of Infodynamics” argues that information entropy in physical systems decreases over time, the opposite of what thermodynamics predicts, suggesting built-in “data compression” like a simulation would use. He’s applied this to SARS-CoV-2 mutations and proposed that information itself has mass.

The serious problems with this claim show up in the cherry-picked evidence section below.

Every YouTube Video Uses the Same Quantum Trick

This is the argument in every YouTube video on the topic: the observer effect means reality only “renders” when you look at it, quantum entanglement lets particles share information like variables in the same memory address, quantization packages reality into discrete pixels, and the Planck length serves as the “minimum resolution.”

These analogies are thought-provoking, which is why they spread so well. But analogies aren’t evidence, and every single one has an explanation within standard physics that doesn’t require invoking a simulator.

The Weirdest Finding: Web-Browser Code, Buried in String Theory

Theoretical physicist James Gates Jr. (University of Maryland) discovered what appear to be error-correcting codes, specifically doubly-even self-dual linear binary block codes of the type used in web browsers, embedded in the equations of supersymmetric string theory.

Gates himself is careful to say this could be mathematical coincidence or deep structure. But it’s the single most cited “evidence” in pro-simulation videos, and unlike most claims, it’s genuine peer-reviewed physics.

Part II

Now Watch the Physics Take It Apart

The steel-man case sounds compelling right up until you check it against actual physics.

You’d Need More Computing Power Than the Universe Contains

Simulating the quantum state of just 300 particles requires more classical bits than there are atoms in the observable universe. 2300 is vastly larger than 1080, the total number of atoms in the observable universe.

Scale that to an actual universe and the computational resources required to simulate quantum mechanics at full fidelity would exceed the information content of the thing being simulated. At equivalent-scale physics, that makes it a mathematical impossibility rather than a matter of needing better computers.

The only escape hatch is that the parent reality has fundamentally different, more powerful physics than ours. But then you’re no longer making a scientific argument; you’re speculating about an unknowable realm with unknowable properties, the exact same move theology makes about heaven.

No Competent Programmer Would Build It This Way

Nobel laureate Frank Wilczek raises what may be the strongest single objection: the laws of physics contain “hidden complexity that is not used for anything.” Physical laws are local, unchanging, and apply everywhere uniformly. None of that is necessary in a simulation.

A competent programmer wouldn’t build unnecessary hidden structure. You don’t code features nobody will ever see or use. The universe has enormous amounts of structural complexity in its physics that serves no apparent purpose, exactly what you’d expect from something unguided rather than engineered.

The “Rendering Optimization” Theory Has It Exactly Backwards

The popular claim is that quantum mechanics shows reality only “renders” when observed, like a video game saving resources. But the “observer” in quantum mechanics is any physical interaction, not conscious observation. A detector, a stray photon, and a wall all count.

More importantly, the unobserved quantum state isn’t simpler than the observed one. A particle in superposition is in more states simultaneously, not fewer, making the wave function more computationally expensive than the collapsed state. If someone were optimizing a simulation, quantum mechanics is the opposite of how you’d do it.

Physicists Went Looking for the Glitch. They Found Nothing.

Physicists have tested this. The lattice test proposed by Beane, Davoudi, and Savage (2012) would detect if space-time is discretized on a grid, because ultra-high-energy cosmic rays would show directional bias aligned with grid axes. No such bias has been observed.

Tom Campbell proposed testing whether the double-slit pattern changes based on whether which-path information is theoretically recoverable versus actually recorded. His claimed preliminary results have not been independently replicated or published in major journals.

Every proposed test has either come back negative or remains unconfirmed. The simulation hypothesis makes no unique predictions that have been verified.

“The simulation hypothesis is pseudoscience and religion. It is physically impossible to simulate the universe without producing measurable inconsistencies.”

— Sabine Hossenfelder, theoretical physicist (2021). She later softened to “5 out of 10 on the bullshit meter” in 2026, which is still hardly a ringing endorsement.

Part III

Set the Physics Aside. The Logic Falls Apart on Its Own.

Even if you set aside the physics, the argument defeats itself logically.

It’s Turtles All the Way Down, and That’s a Problem

If Bostrom’s statistical reasoning applies to our reality, it applies to every reality above us too. The simulators are probably simulated, and so are their simulators all the way up, leaving no base reality at all. The argument proves too much.

That creates a cascading failure problem: if any reality in the chain fails, whether from a shut-down supercomputer, a collapsed civilization, or a power outage in some meta-reality, our world ends. As the probability of being simulated approaches 1, the probability of existing in an infinite chain also approaches 1, and the probability of at least one link failing in an infinite chain is essentially 1.

The theory is self-defeating: the more likely simulation is, the more likely our imminent nonexistence, which makes our continued existence evidence against simulation.

Sean Carroll Found the Hole in the Argument’s Own Premise

Cosmologist Sean Carroll points out a clean logical objection: the simulation argument assumes we’re “typical” observers. But if we’re typical and can’t currently perform full universe simulations, we can’t conclude that other civilizations probably can, which defeats the premise.

Nothing Could Ever Prove This Wrong. That’s the Problem, Not the Strength.

No evidence would convince a simulation believer they’re wrong, because every observation gets read as confirmation no matter which way it breaks:

An unfalsifiable hypothesis is a worldview wearing a lab coat, not a scientific theory. As philosopher Preston Greene pointed out, even trying to test it carries a paradox: if the simulators notice us looking for evidence, they might end the simulation.

Every Piece of “Evidence” Already Assumes the Conclusion

Every piece of “evidence” cited for simulation presupposes the simulation to explain perfectly normal phenomena. Quantum weirdness, deja vu, the Mandela Effect: whatever the input, the output is always the same word. It assumes the conclusion in order to interpret the evidence.

The Guy Who Invented This Argument Gives It One-in-Three Odds

YouTube videos never mention that Bostrom himself distributes his credence roughly equally across all three propositions of his trilemma, giving the “we’re simulated” option about a one-in-three (roughly 20%) probability. Even the architect of the argument doesn’t think it’s likely.

When Musk says “one in billions,” he’s citing his own gut rather than Bostrom’s math. He’s an engineer working from intuition, not a philosopher building a formal argument.

Part IV

The “Evidence” Only Works If You Don’t Look It Up

The pro-simulation case relies heavily on phenomena that sound mysterious until you learn the actual science behind them.

The Mandela Effect Is Just Your Brain Misremembering a Book Title

Large groups of people misremember the same things: the Berenstain/Berenstein Bears, Nelson Mandela dying in prison. Simulation proponents claim these are “patches” to the simulation where the programmers changed something and most people’s memories updated, but some didn’t.

This requires believing an advanced civilization capable of simulating reality would be sloppy enough to leave memory inconsistencies. Cognitive science has already documented how human memory works: confabulation, social reinforcement, schema-based recall. We don’t need simulation theory to explain why people misremember the spelling of a children’s book. No serious physicist or philosopher has ever endorsed this argument.

“Glitches in the Matrix” Are Just Faces in the Clouds

Strange coincidences, unexplained feelings, doppelgangers, objects turning up where they shouldn’t, all get reframed as a “glitch” by people who’ve already decided reality is a simulation and are just looking for confirmation. It’s pattern recognition applied to random data, the same cognitive process that makes people see faces in clouds and Jesus in toast.

Deja Vu Isn’t a Clue. It’s a Temporal Lobe Doing Its Job Badly.

Deja vu is a well-studied temporal lobe phenomenon that correlates with epileptic activity, fatigue, and stress, and can be reliably triggered with electrical stimulation. It’s your brain misfiring a familiarity signal, not the universe buffering.

Fine-Tuning Has Four Explanations. Simulation Is Just One.

As noted in the steel-man section, fine-tuning has multiple explanations: coincidence, multiverse, God, or simulation. It doesn’t uniquely support simulation over the alternatives. It’s structurally identical to the teleological argument for God’s existence, using the same evidence repackaged in tech vocabulary.

The “Proof” That Turned Out to Be Circular Reasoning in a Lab Coat

Vopson’s “Second Law of Infodynamics” has attracted significant popular attention, but the critical analysis shows major flaws:

The reasoning is circular: he defined his own terms, measured by them, then claimed the measurements supported his theory. Popular Mechanics called it “cherry-picking one very specific measurement.”

His papers appeared in AIP Advances, a mega-journal with lower peer review standards than selective journals, and he self-published his book through his own physics institute.

The evidence is also self-defeating. A simulator capable of manufacturing an entire universe wouldn’t leave a giant footprint in data handling. If you believe the simulation is sophisticated enough to fool 8 billion people, you can’t also believe it left obvious fingerprints for a single university researcher to find.

Vopson has appeared on Bible-simulation podcasts, blending science with religious apologetics. As Popular Mechanics put it, simulation theory is “getting propped up and amplified by extremely wealthy people who have financial incentive to make people feel like they are not real,” devaluing human experiences and rights. He has essentially updated intelligent design for the tech era.

“Propping up ideas that encourage questions to be left unanswered is not particularly scientific.”

— Popular Mechanics, December 2024

Part V

Swap “Simulator” for “God” and You’ve Got a Sermon

That is the critique that makes simulation proponents most uncomfortable, because it’s the one they can’t argue against without undermining their own position.

Philosopher Duncan Clarke mapped the simulation hypothesis onto religious creation myths and found the logical structure is identical:

God = Simulator. An unfalsifiable creator entity with capabilities beyond our comprehension, whose motivations we can only speculate about.

Creation = Simulation. A designed reality, purpose-built by the creator, with rules we can discover but whose ultimate nature is hidden.

Heaven = Base Reality. A higher plane of existence where the “real” beings live, inaccessible to us except through faith or speculation.

Afterlife = Persistence through simulation layers. Death might be “logging out,” near-death experiences might be “transition sequences,” and your consciousness might simply persist in another layer.

Suffering = “They programmed it this way.” This is the problem of evil, repackaged: the simulators wanted the suffering, didn’t care, or considered it a feature rather than a bug, which is identical to theodicy in theology.

Robin Hanson argues that if you believe you’re simulated, rational behavior changes: be entertaining and praiseworthy to avoid being “turned off.” That is functionally identical to behaving morally because God is watching.

As one Reddit commenter put it: “Isn’t this just the creation philosophy used in most religions, just on Windows?”

The simulation hypothesis appeals to tech-literate atheists who would reject identical claims if framed religiously. Replace “programmer” with “God” and “simulation” with “creation,” and you have standard deism with a Silicon Valley aesthetic.

Resembling religion doesn’t automatically make the hypothesis false. But it does mean the hypothesis has no more claim to scientific status than religion does, offering the same unfalsifiable comfort, the same appeal to a higher reality, and the same answers to unanswerable questions, just in different vocabulary.

The cross-cultural universality of “this isn’t the real reality” is interesting as a psychological phenomenon. You see it in Plato’s Cave, Zhuangzi’s Butterfly Dream, Descartes’ Evil Demon, the Aztec painted world, the Gnostic Demiurge, Hindu Maya, and Buddhist dependent origination. Every civilization has some version of this intuition, but universal human intuitions are often wrong. We also universally intuited that the earth was flat and the sun moved around us.

Part VI

38 Sources. Here’s Who Actually Believes This and Who Doesn’t.

This analysis draws on 38 sources spanning academic papers, physicist interviews, podcasts, YouTube explainers, philosophy essays, and Reddit critiques, broken down below by what each one actually claims.

Where the 38 Sources Came From

Foundational papers and academic sources: Bostrom’s 2003 trilemma, Beane/Davoudi/Savage lattice test (2012), Campbell et al. testing proposals (2017), Vopson’s infodynamics papers, Wolpert’s multiverse compatibility paper (2026), the Doomsday Argument (Carter/Leslie).

Working physicists on record: Sabine Hossenfelder, Frank Wilczek (Nobel laureate), Sean Carroll, George Ellis, Marcelo Gleiser, Paul Davies, James Gates Jr., David Wolpert.

Proponents and their cases: Elon Musk (intuitive probability), Neil deGrasse Tyson (who later reversed), Rizwan Virk (RPG/NPC framing), Tom Campbell (testing proposals), Donald Hoffman (interface theory).

Philosophical and cultural analysis: Duncan Clarke (religion parallels), Robin Hanson and Preston Greene (behavioral implications), historical precedents (Zhuangzi, Descartes, Plato), Eastern philosophy parallels, comparative religion.

Critical analysis and community discussion: Popular Mechanics (critique of Vopson), Reddit users who produced some of the sharpest logical objections (cascading failure, Occam’s Razor, computational crash argument), Curt Jaimungal (convergent series mathematical critique).

Believers, Skeptics, and the Physicists Who Just Laugh

Committed proponents: Musk, Virk, Vopson. None are physicists working in the relevant fields. Musk is an engineer, Virk is a game developer, and Vopson is a physicist working outside mainstream consensus.

Serious but skeptical: Bostrom (created the argument but gives it ~20%), Tyson (initially pro, then reversed after Gott’s objection), Chalmers (takes substrate independence seriously but doesn’t endorse simulation), Hossenfelder (moved from “pseudoscience” to “5/10 on the bullshit meter”).

Outright dismissive: Wilczek, Carroll, Ellis, Gleiser. These are the physicists actually working on the relevant math. Ellis called it “late-night pub discussion, not a viable theory.”

The pattern: Most serious physicists treat the simulation hypothesis as a thought experiment, not a scientific theory. Confidence runs highest among people furthest from the physics; researchers closest to the math rate it somewhere between “interesting thought experiment” and “total nonsense.”

The most underrated sources: Reddit users produced the cascading failure argument (which may be the single most devastating logical objection), the Occam’s Razor analysis, and the computational crash argument. Anonymous internet commenters, it turns out, can be sharper than best-selling authors when there’s no book to sell.

The Verdict

So Is Reality a Simulation? Here’s the Straight Answer.

The simulation hypothesis is an unfalsifiable philosophical position, not a scientific theory: it makes no unique, testable predictions, every proposed test has come back negative or stayed unconfirmed, and no conceivable observation could disprove it.

It remains intellectually interesting: Bostrom’s trilemma is elegant logic, the quantum parallels are thought-provoking, and the questions about consciousness and substrate independence are worth taking seriously.

But “interesting thought experiment” is very different from “almost certainly true.”

Anyone claiming it’s “almost certainly true” is making a faith claim indistinguishable from religion. They’re positing an unfalsifiable creator, a designed reality, a hidden higher plane, and a framework that explains everything while predicting nothing. The only difference between “we live in a simulation” and “God created the world” is the vocabulary.

The fact that smart people believe it doesn’t make it scientific. Smart people believe plenty of unfalsifiable things, because intelligence and rigor are not the same trait.

The physics kills it: simulating quantum mechanics at full fidelity would take more computational resources than the universe contains.

The logic kills it too: the argument proves too much, implying infinite nesting, cascading failure, and our own imminent nonexistence.

And the evidence doesn’t hold up: every “proof” turns out to be a misunderstanding of quantum mechanics, a well-documented cognitive glitch, or reasoning that assumes its own conclusion.

Strip away the vocabulary and it maps 1:1 onto religion, offering the same unfalsifiable comfort to a demographic that thinks it’s too sophisticated for church.

The simulation hypothesis predicts nothing, explains nothing that standard physics doesn’t already cover, and gives you no different action to take on a Tuesday. Even if it were true, as the pragmatic equivalence argument shows, nothing about your actual life would change: pain would still hurt, love would still matter, and your choices would still be yours to make.

Across the 2,400-year history of this question wearing different clothes, from Zhuangzi’s butterfly to Descartes’ demon to Bostrom’s computer, every version has gone nowhere. The question was built unanswerable from the start.

Want the Next Trendy Claim Fact-Checked Before You Believe It?

Deep dives backed by evidence, not vibes. AI, science, predictions, and who’s actually getting it right. No spam.

Unsubscribe anytime. Your email stays private.

Related on scovert.com

— Scott Covert, who actually built this. I work with these tools hands-on every day. If something here sparked a question, a project, or a “wait, can you build that for me?”, that’s the best message I get. Reach me, pitch a collab, or book a consult.

Follow Scott · Substack · Medium