What this covers

Sean Carroll answers Patreon-submitted questions across physics, philosophy, economics, and technology. The episode works through several fashionable intellectual puzzles—the simulation hypothesis, AI singularity risk, consciousness as substrate-dependent, proposals to replace quantum mechanics—arguing that many dissolve when examined carefully rather than deepening into genuine mysteries. Carroll's recurring theme is that coarse-graining and entropy are objective features of the world once a coarse-graining scheme is chosen, not observer-dependent or mystical; that Many-Worlds interpretation remains better supported than classical realist alternatives; and that artificial general intelligence is physically possible (since brains are physical systems) even if current LLM architectures won't achieve it alone.

The breadth spans quantum mechanics and determinism, where Carroll defends Many-Worlds as locally sensible and discusses alternatives like Gerard 't Hooft's classical realism; thermodynamics and time, including how coarse-graining resolves the microscopic reversibility paradox; and mental phenomena, where Carroll argues that physicalism survives even strong emergence. On economics, he examines how system optimization creates perceived powerlessness and whether algorithmic advances might one day make centralized planning competitive with markets. The conversation also touches Peter Singer's drowning-child argument, the conceptual gaps between Bayesian and Popperian epistemology, Sarah Walker's assembly theory, and sharp criticism of how AI-doom narratives function as collective theater—a kind of kayfabe—to mislead public discourse.

Sharpest takeaway

In this AMA, Carroll argues that many fashionable scientific and philosophical puzzles—the simulation hypothesis, the AI singularity, strong substrate-dependence, and proposals to overturn quantum mechanics—are better dissolved through clear physical and conceptual reasoning than treated as deep mysteries, while emphasizing that emergence, entropy, and coarse-graining are real but objective rather than mystically observer-dependent.

  • Coarse-graining and entropy are objective once a coarse-graining is chosen, not dependent on intelligence
  • Quantum mechanics with superpositions and many-worlds is better supported than classical-realist or particle-based alternatives
  • AGI is physically possible because brains are physical systems, but LLM architecture alone won't achieve it

The claims · ranked29 claims · weighted by value

This asset isn't compiled yet

You're seeing its claims, ranked. Compile it to build the argument threads, weight them, and check each claim against your library — the full view.

0.79

The observed increase of entropy arises from how we coarse-grain phase space over time; without any coarse-graining, Liouville's theorem implies phase-space volume is conserved and there would be no entropy increase—raising the parallel question of whether complexity too would be constant or zero absent coarse-graining.

causalhigh valueestablishednovelty 3/4durability 4/4· Sean Carroll

“the observation of entropy increasing is because of how we coarse grain phase space over time, and if there was no coarse graining, there would be no entropy increase (Liouville's theorem)”

0.78

A Bayesian approach to knowledge increases confidence that one is approaching truth without ever reaching certain truth statements, but Popperian epistemology argues Bayesian processes ultimately rest on an initial guess or conjecture forming the prior, so conjecture and refutation may be the more fundamental truth-seeking process—or alternatively, Bayesian updating could itself be construed as part of the critical testing demanded by conjecture and refutation.

normativehigh valuecontestednovelty 3/4durability 4/4· Sean Carroll

“Bayesian based processes can themselves be traced back to an original guess or conjecture which forms the prior in the Bayesian process”

0.78

Entropy is not fundamentally dependent on intelligence; while defining a coarse-graining requires choosing which microstates to group into macrostates, once that coarse-graining is specified the entropy becomes an objective matter of fact about the system, independent of whether any intelligent observer exists.

factualhigh valuecontestednovelty 3/4durability 4/4· Sean Carroll

“once a coarse-graining is defined, entropy is a matter of fact, whether or not there is intelligence in the Universe”

0.77

As price efficiency and economies of scale increase, consumers often become less happy and feel less able to influence large institutions, and this growing sense of powerlessness from system optimization may be a key reason people feel things are worse even when measurable life conditions have improved.

causalhigh valuecontestednovelty 3/4durability 3/4· Sean Carroll

“as price efficiency and economies of scale increase, consumers often become less happy and feel less able to influence large institutions”

0.77

Although free markets historically outperformed Soviet central planning because prices aggregate dispersed information more efficiently, advances in technology and algorithmic information-gathering (as used by companies like Uber) raise the question of whether centralized planning could eventually match or surpass markets in practical efficiency—and if so, whether free markets would then become morally questionable.

forecasthigh valuecontestednovelty 3/4durability 3/4· Sean Carroll

“Do you think technology could get to the point that Soviet-style centralized planning of the economy could be superior, in a practical sense, to free markets?”

0.73

Even if one modified the Standard Model to allow strong emergence, any such strongly emergent behavior could itself be mathematically encapsulated in the laws, so no matter how surprising it would still pose no threat to materialism about the mind.

factualhigh valuecontestednovelty 3/4durability 3/4· Sean Carroll

“any such strong emergence could be mathematically encapsulated and as such, no matter how surprising, it would still not be any threat to materialism about the mind”

0.73

Substrate dependence of a phenomenon like consciousness can be graded analogously to weak versus strong emergence: a weak form merely restates physicalism (material properties enable behavior but can be replicated in other materials with similar properties), while the strongest form posits that the behavior is uniquely dependent on the specific material such that it cannot be replicated in any other substrate.

definitionhigh valuecontestednovelty 3/4durability 3/4· Sean Carroll

“the strongest form of substrate dependence would posit that the system behavior is *uniquely* dependent on its material properties, such that the behavior cannot be replicated in other materials”

0.73

Systems that appear dynamic, such as the interior of a nucleon, can be described as static (stationary) solutions to the Schrödinger equation, because a stationary state has a time-independent probability distribution even though the underlying quantum object is not classically at rest.

factualhigh valueestablishednovelty 2/4durability 4/4· Sean Carroll

“things which appear dynamic (such as the inside of a nucleon) are in fact static solutions to the Schrödinger equation”

0.73

Some infinities are genuinely larger than others because Cantor's diagonal argument demonstrates that the real numbers cannot be put into one-to-one correspondence with the natural numbers—any proposed enumeration can be defied by constructing a new number differing in each diagonal digit—so the reals are uncountably infinite while the naturals are countably infinite.

factualhigh valueestablishednovelty 2/4durability 4/4· Sean Carroll

“some infinities are larger than other infinities”

0.73

Because the fundamental laws of physics are time-reversal symmetric, if every particle's momentum in an isolated room were exactly reversed at one instant, the system would in principle retrace its evolution, recomposing structures—such as a decomposed human—running backward, illustrating the conflict between microscopic reversibility and macroscopic irreversibility.

factualhigh valueestablishednovelty 2/4durability 4/4· Sean Carroll

“reverse the momentum of every particle in the room, would the inside of this room play itself back, recomposing a human who lives backwards and gets younger over time?”

0.71

The claim that the most fundamental formulation of physical law is a dynamical law plus boundary or initial conditions is challenged by the fact that nearly all interesting General Relativity solutions (Schwarzschild, de Sitter) were not found that way and some (the Gödel universe) cannot even be formulated that way, just as Kepler's laws were stated without initial conditions before Newton recast them.

factualhigh valuecontestednovelty 3/4durability 4/4· Sean Carroll

“nearly all interesting solutions in General Relativity were not found that way (e.g. Schwarzschild or de Sitter), and many cannot even be theoretically formulated that way (e.g. Godel universe)”

0.68

Artificial general intelligence is achievable in principle because human brains are purely physical systems obeying the laws of physics, so there is no in-principle barrier to replicating their information-processing capabilities in another substrate—even though the current large language model architecture alone will not get there.

factualhigh valuecontestednovelty 2/4durability 3/4· Sean Carroll

“I tend to agree with you that the llm architecture alone will never reach any “agi” level of intelligence. And I also agree with you when you say that reaching agi is totally possible”

0.68

A system can be deterministic in principle yet scale beyond what we can call practically deterministic, as with large language models: we know exactly how they work in principle, but analyzing their internal state to determine why they produce specific outputs is so difficult—especially with self-feedback mechanisms—that practical predictability is lost.

factualhigh valuecontestednovelty 2/4durability 3/4· Sean Carroll

“while we know exactly how they work in principle, analyzing their state to determine why they do certain things is incredibly difficult”

0.66

Complexity is a distinct concept from entropy: whereas entropy monotonically increases over time, complexity characteristically starts low, rises to a peak in the middle of a system's evolution, and then declines again as the system approaches equilibrium.

factualhigh valuecontestednovelty 3/4durability 3/4· Sean Carroll

“complexity is a distinct concept from entropy. Whereas entropy only increases, complexity usually increases but then decreases”

0.66

In the Many-Worlds interpretation, the measurement of distant entangled particles can be understood as a local process of branching, so that if Many-Worlds renders entanglement local while Bell-inequality tests show other interpretations require nonlocality, this locality should count as significant experimental evidence favoring Many-Worlds.

factualhigh valuecontestednovelty 3/4durability 3/4· Sean Carroll

“If MW does make entanglement local, shouldn’t the non-locality of other QM interpretations, shown in tests of Bells inequality, be large experimental evidence in favor of MW?”

0.66

Work by Wondrak, van Suijlekom, and Falcke suggests Hawking radiation does not require an event horizon and can arise in vacuums under different geometries of curved spacetime, which if correct would raise the possibility of detecting such radiation without needing an actual black hole.

factualhigh valuecontestednovelty 3/4durability 3/4· Sean Carroll

“Hawking radiation does not require an event horizon, and can arise in vacuums under different geometries of curved space”

0.65

Peter Singer argues that if it is in our power to prevent something very bad from happening without sacrificing anything of comparable moral significance, we ought to do it; the drowning-child case shows we must save the child despite ruined clothes, and by parity of reasoning, spending money on luxuries instead of donating to prevent distant starvation is morally equivalent to letting the child drown.

normativehigh valuecontestednovelty 2/4durability 4/4· Sean Carroll

“if it is in our power to prevent something very bad from happening, without thereby sacrificing anything else morally significant, we ought, morally, to do it”

0.64

The Everett Many-Worlds interpretation assumes a deterministic universe, so if our universe is in fact fundamentally probabilistic as it appears (with the uncertainty principle preventing knowledge of an electron's path), the deterministic assumption would be undermined—raising the question of whether Many-Worlds should then be abandoned and what justifies assuming determinism in the first place.

factualhigh valuecontestednovelty 2/4durability 3/4· Sean Carroll

“Everett’s “many worlds” interpretation assumes a deterministic universe”

0.62

Gerard 't Hooft argues that quantum superpositions of states are not real—claiming that if you look carefully things never actually superimpose—and that the underlying real world is classical even at the level of quantum particles, which is the direction he believes physics research should take.

factualhigh valuefringenovelty 3/4durability 3/4· Sean Carroll

“I'd argue that superpositions of states are not real. If you look very carefully, things never superimpose.”

0.62

Sarah Walker's assembly theory holds that objects above a fairly low threshold of complexity cannot spontaneously appear from random fluctuations, which appears to conflict with the idea that given enough time random fluctuations will eventually produce every possible structure, and thereby suggests a non-randomness in how the universe builds complex things.

factualhigh valuecontestednovelty 3/4durability 2/4· Sean Carroll

“according to assembly theory nothing above a certain incredibly low level of complexity can just spontaneously pop into existence”

0.61

Framing mental disorders as 'neurodivergence' implying everyone lies on a spectrum may, despite good intentions, whitewash public perception and mask the genuinely dysfunctional and painful reality experienced by affected people, analogous to euphemistically calling heart disease 'cardiodivergence'.

normativehigh valuecontestednovelty 2/4durability 3/4· Sean Carroll

“Does labeling mental disorders as “neurodivergence” truly honor the suffering of affected persons, or does it whitewash public perception and mask its disfunctional reality behind well-meant respect?”

0.61

When a telescope detects a photon emitted by a quasar billions of light years away, the photon's wave function prior to detection was spread out throughout the universe, and the Born rule probabilities determine the chance of detecting that photon at the telescope's particular location.

factualhigh valuecontestednovelty 2/4durability 3/4· Sean Carroll

“when your telescope detects a photon that was emitted by a quasar billions of light years away, until you detected it, it's wave function was spread out all throughout the universe”

0.53

Anarcho-communism may be untenable because the only routes to it pass through strong attractor states—either abolishing private ownership via communism (which entrenches centralized government) or abolishing the state via anarcho-capitalism (which entrenches capitalism)—and these attractors would trap the system before it could reach the anarcho-communist end goal.

factualhigh valuespeaker onlynovelty 3/4durability 2/4· Sean Carroll

“centralized government and capitalism both seem like attractors of sorts”

0.48

Much of the AI doom discourse may function like professional wrestling's 'kayfabe'—a collective pretense that something is real which the participants know is not—because AI company executives publicly promote alarming narratives about malicious AI and specious studies even though they understand that is not actually how AI works, which corrupts reasonable public discourse.

causalhigh valuespeaker onlynovelty 3/4durability 2/4· Sean Carroll

“Do you think that all of the AI Armageddon talk (promoted mostly by the AI companies themselves) is a kayfabe?”

0.41

Drawing on Koch and Harris, much of human thought may consist of subconscious 'blurtations' that we justify after the fact rather than genuinely reasoned conclusions, raising the question of how to distinguish higher-quality, carefully considered ideas from mere post-hoc rationalizations of subconscious impulses.

factualcontestednovelty 2/4durability 3/4· Sean Carroll

“we are basically all doing that all the time and then justifying it ex post facto”

0.38

Sam Harris argues parents should not lie about Santa Claus because the potential sense of betrayal and loss of trust outweighs the added childhood excitement, treating it as a small but meaningful deception that promotes magical thinking over critical thinking and may undermine a child's relationship with truth and reason.

normativecontestednovelty 1/4durability 3/4· Sean Carroll

“the potential sense of betrayal and loss of trust isn’t worth the added excitement”

0.34

The simulation hypothesis shares the structure of creationism by positing an intelligent creator outside the universe, but it earns more scientific respectability because it grounds that creator in physical computation rather than supernatural divinity, making it at least in principle a naturalistic hypothesis.

factualcontestednovelty 1/4durability 2/4· Sean Carroll

“trading a mythical divine Creator for an equally mythical cosmic coding Creator”

0.34

Cuts to NASA, NSF, and other US government STEM funding may have serious long-term implications, to the point that junior STEM researchers might be advised to pursue their careers abroad.

forecastcontestednovelty 1/4durability 2/4· Sean Carroll

“Some are saying that if one is a STEM researcher one should probably plan to pursue one's career abroad”

0.26

The academic professor hiring process may resemble the NBA draft in being heavily focused on superstars, or alternatively professors in a given department may all be at roughly the same level, with the question of how often genius-level talent actually matters being unclear.

factualspeaker onlynovelty 1/4durability 2/4· Sean Carroll

“Is the professor hiring process like the NBA draft (superstars focused), or are professors in a given department all at pretty much the same level?”