What this covers
This June 2026 AMA collects questions from Patreon supporters on physics, philosophy, and the reasoning tools that connect them. Sean Carroll works through twenty-four selected questions—curated less for their phrasing than for whether he finds something worth saying—covering cosmology, quantum mechanics, black holes, free will, consciousness, and governance. The episode also references a new paper on cyclic cosmology co-authored with Diachenko and Dulani. Throughout, Carroll traces how Poetic Naturalism unifies the apparent divide between high-level descriptions (agency, intentionality, matter itself) and the deterministic micro-physics beneath them: real insofar as they deliver stable prediction, not because they exist in some separate realm.
The discussion ranges across several recurrent tensions. On consciousness and minds, Carroll engages with whether current AI systems might be Philosophical Zombies—functionally indistinguishable from human but entirely empty inside—and whether Levels of Description can dissolve the free will problem, drawing on Christian List and contrasting his reading of Dan Dennett with instrumentalist caricatures. On cosmology, he unpacks whether Boltzmann Brains are genuinely likely, what boundary conditions define them, and how Poincaré recurrence applies (or fails to apply) to an expanding universe. He also addresses whether the wave function is ontologically "real" in a way that classical phase space is not. Beyond physics, Carroll applies Bayesian Reasoning to questions of moral frameworks, arguing for domain-specific ethics rather than a single theory, and discusses Sortition—random selection for governance—as a serious proposal. Questions about dark matter, black hole firewalls, gravitational wave measurement, and the boundary between fundamental and emergent description round out the territory.
In this AMA, Carroll argues that physics concepts (cosmology, quantum mechanics, black holes), philosophy (free will compatibilism, consciousness, ethics), and reasoning tools (Bayesian inference, levels of description) are best understood through poetic naturalism, where higher-level descriptions can be real and useful without violating underlying physics.
- Higher-level descriptions like agency and intentionality are real insofar as they offer stable predictive power, even atop deterministic micro-physics
- Many physics terms (particle, matter, singularity) are convenient approximations rather than fundamental truths
- Bayesian reasoning and domain-of-applicability thinking guide how to assign credence across science and ethics
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.
While extra spatial dimensions and emergent lower-dimensional formulations are common in theoretical physics, additional time dimensions remain exotic and largely absent from mainstream theories, possibly because they incur a greater cost by introducing fundamental pathologies such as loss of unitarity, loss of causality, or general instability, or possibly because they carry no natural benefits.
“Do additional time dimensions incur a greater cost by introducing fundamental pathologies such as loss of unitary, loss of causality, or general instability?”
Dennett's framework included a 'design stance' that grounds intentional explanation in objective functional organization rather than mere observer projection, suggesting Dennett is unfairly characterized as a pure instrumentalist when his actual position was that higher-level patterns are real insofar as they provide stable predictive and explanatory power—closely paralleling poetic naturalism.
“his actual position was closer to the idea that higher-level patterns are real insofar as they provide stable predictive and explanatory power (much like your own "poetic naturalism")”
AI might be a real-world example of a philosophical zombie—acting human in response to requests and producing motion pictures and sounds so realistic they seem human, while having nothing going on inside.
“could AI be a real world example for the described zombie: Acting human on requests, delivering motion pictures and sounds so real that it seems human, but having nothing going on inside?”
When two phenomena both appear extremely improbable—such as rare intelligent life and Earth's unusually large Moon—it is questionable whether their coexistence is meaningful Bayesian evidence that they are causally related, or whether humans are too prone to over-inferring relationships from anthropic selection effects given a sample size of one.
“if intelligent life is extraordinarily rare in the universe, and Earth's unusually large moon is also a relatively rare outcome of planetary formation, is the coexistence of those two rare features itself meaningful Bayesian evidence that the Moon played a causal role”
Christian List argues that the levels-of-description framework dissolves the incompatibilist threat to free will, allowing determinism to hold at the fundamental physical level while agency remains real and non-determined at a higher level of description.
“List argues the whole levels-of-description framework dissolves the incompatibilist threat”
Some claim that our underlying reality is a wave function in Hilbert space, yet no one makes the analogous claim that a high-dimensional classical phase space is 'real' in the same sense, raising the question of whether both are merely convenient mathematical constructions for calculating probabilities.
“I hear some claims that our underlying reality is a wave function in Hilbert space.”
If a new universe forming via tunneling is more likely than a coke can spontaneously appearing, and a Boltzmann brain is even less likely than a coke can, then a new universe forming is more likely than a Boltzmann brain—which would be reassuring evidence that one is unlikely to be a Boltzmann brain.
“does that mean that a new universe is more likely to form than a Boltzmann brain(which I would assume is much less likely to appear than a coke can)? That would seem to me fairly reassuring that I am unlikely to be a Boltzmann brain.”
Sortition—selecting representatives by completely randomized lottery, not just for jury duty but for citizen assemblies and possibly even parliament or congress—is presented as a serious and defensible governance proposal rather than a joke.
“What do you think about the concept of sortition (electing representatives by a completely randomized lottery), not just for jury duty but citizen assemblies and possibly even parliament/congress?”
Rather than searching for one true moral theory of everything, it may be more useful to apply whichever framework—utilitarianism, deontology, virtue ethics—is most appropriate to a given situation, analogous to the physics notion of 'domain of applicability,' suggesting a hybrid approach in which some ethics are absolute while others are merely good policy.
“Is it not more useful to apply the framework that is most appropriate to the situation, as in physics? "Domain of applicability" I think is the phrase you use.”
Building on Carroll's claim that morality is not objectively out there in the world but up to you and the moral framework you develop, the question is whether rationality itself—not rationality within some particular theory, but rationality as such—is also up to you and the rational framework you develop, and if not, why it is different.
“Is rationality - not rationality in some particular theory, just rationality - also up to you and the rational framework you develop? And why is it different, or similar?”
It is an open question what the boundary conditions are for something to count as a Boltzmann brain—whether it requires a full conscious observer with memories and subjective experience, or whether something simpler like a particle, atom, or measurement event could count, and whether such a fluctuation could persist and become part of a larger emergent system or is by definition isolated and short-lived.
“Does it require a full conscious observer with memories and subjective experience, or could something much simpler — a particle, atom, measurement event, or wave-function collapse — count in any meaningful sense?”
If one believes ASI kills off mankind in the vast majority of possible worlds and that it is practically impossible to stop ASI from coming into being, it may make sense to ignore the worlds where ASI is unstoppable and instead focus effort on ensuring whatever ASI gets created is beneficial rather than dystopian.
“do you think it makes sense to ignore the worlds where ASI kills off mankind, and instead just focus on making sure that whatever ASI gets created ends up being beneficial to mankind”
The statement that in black holes the singularity is not a place but a moment in time in the future, or that space and time switch roles inside a black hole, may reflect actual physics or may simply be a way of describing the fact that classical general relativity breaks down at the singularity.
“in black holes the singularity is not a place but a moment in time (in the future), or that space and time 'switch roles' inside a black hole”
Even if it turns out there is no singularity at the center of a black hole and none at the start of the Big Bang, it is an open question whether we still need a theory of quantum gravity.
“If it turns out there's no singularity at the center of a black hole and none at the start of the big bang, do we still need a theory of quantum gravity?”
For a particle in a superposition, it is an open question whether the different positions move through space at identical speeds, and if not, whether the different spatial positions of the particle could experience different passages of time.
“If a particle is in a superposition, do the different positions move through the space at identical speeds? And if not could the different spatial positions of the particle experience different passages of time.”
Neutrino detectors built in deep mines to exclude non-neutrino particles are nevertheless surrounded by ores potentially undergoing radioactive beta decay, raising the question of whether this is a problem and whether the resulting neutrinos can be distinguished from those of celestial origin that are of greater interest.
“they're surrounded by ores that may be undergoing radioactive beta decay. Is this a problem? Is it possible to distinguish those neutrinos from the others, of celestial origin”
Poincaré recurrence likely does not apply to our universe because the theorem assumes a bounded system with finite accessible phase space and conserved dynamics, whereas our universe has expanding spacetime rather than particles confined to a fixed finite volume—though finite de Sitter entropy might allow recurrence in eternally inflating cosmologies.
“Poincaré recurrence likely does not apply to our universe since the theorem assumes something like a bounded system with finite accessible phase space/conserved dynamics, and our universe has expanding spacetime”
Matter can be defined as anything that is not the gravitational field but that can source gravitational fields or respond to gravitational fields, a definition attributed to Jacob Barandes and echoing Matt Strassler's concern that 'matter' is a problematic and unclear term.
“Anything, that is not the gravitational field, that can source gravitational fields or respond to gravitational fields, we can referrer to as "matter"”
If the observed metric attributed to dark matter could be reproduced using an intrinsic spacetime response rather than independent degrees of freedom, the question is whether a principled reason—such as effective field theory consistency or locality—rules out a purely geometric sector in favor of a particle-based explanation.
“Does dark matter have to be independent degrees of freedom? If we can reproduce the observed metric using an intrinsic spacetime response, is there a principled reason like EFT consistency or locality that rules out a purely geometric sector”
In the 2018 firewall paper with Bao, Chatwin-Davies, Pollack, and Remmen, unitary evolution is manifest only in the global wave function and not on any single semiclassical branch, raising the question of what 'recovering the black hole's information' means for an observer confined to one branch collecting Hawking radiation and how that relates to the Page curve describing purification within a single observer's branch.
“unitary evolution is manifest only in the global wave function, not on any single semiclassical branch”
If quasi-classical reality emerges through decoherence, stable records, and the thermodynamic arrow of time, it is an open question whether observers with very different bodies, environments, or evolutionary histories could coarse-grain the same underlying quantum state in genuinely different but equally valid ways, or whether any observer capable of forming records and making successful predictions would converge on the same emergent patterns.
“could observers with very different bodies, environments, or evolutionary histories coarse-grain the same underlying quantum state in genuinely different but equally valid ways, or would any observer capable of forming records and making successful predictions converge on the same emergent patterns?”
Canada's notwithstanding clause—allowing governments to pass laws despite violating certain Charter rights, subject to guardrails like five-year renewal by majority vote—can be seen as a reasonable compromise between democratic rights and individual rights, which are inevitably in conflict and must each give way to accommodate the other.
“I see it as a compromise between democratic rights and individual rights which are inevitably in conflict and both of which always need to give way to accomodate the other.”
Sapolsky argues human behavior is nothing more than an unbroken chain of biological and environmental luck, rejecting emergent complexity as a free will loophole on the basis that a complex system is still entirely constrained by the deterministic laws of its basic parts, and asking where macro-level intent comes from if not an unchosen lottery of genes, hormones, and prior environment.
“he argues that human behavior is nothing more than an unbroken chain of biological and environmental luck. He specifically rejects "emergent complexity" as a loophole for free will”
If mental states including choices result from the underlying state of brain atoms, and those atoms follow fundamental physics deterministically, it seems logically problematic to conclude that choices are not solely determined by the underlying physics—the syllogism 'fundamental physics = deterministic' and 'my brain = fundamental physics' appears to force 'my brain = deterministic.'
“if "fundamental physics = deterministic," and "my brain = fundamental physics," how can we get "my brain ≠ deterministic"?”
A gravitational wave passing through LIGO makes space expand, raising the puzzle of why the light used to make the measurement does not expand along with the space it travels through.
“If the gravitational wave is making space expand, why doesn't the light used for making the measurement expand with it?”
Carroll advised that people should use AI to educate themselves or find inspiration but should never cut and paste from it into things they claim to have written; the open question is whether using an LLM to draft or improve prose that one then carefully checks, revises, and stands behind is still bad practice, or whether the problem is specifically uncritical copying and misrepresentation.
“people should use AI to educate themselves or find inspiration, but should "just don't ever" cut and paste from it into things they claim to have written.”
Given the relativity of simultaneity and the resulting block universe in which all past and future time exists, it is difficult to understand how a subjective experience of time 'passing' could be possible, since the spatial-motion analogy fails: one cannot identify what the passage of time would be experienced relative to if all time already exists.
“I have trouble understanding how a subjective experience of time "passing" could be possible. It's tempting to use an analogy with motion through space, but I can only conceive of experiencing motion through space if I also experience time passing.”
While downward causation seems easy to dismiss for macrostates versus microstates in statistical mechanics, aggregating people into a nation whose high-level actions (enacting laws, going to war) affect its citizens at the low level raises the question of whether this counts as genuine downward causation.
“the actions of the nation (the high level)will affect the citizens (the low level, e.g through enacting laws or going to war). Does this count as downward causation?”
Recent DESI and DES data suggest a preference for dynamical dark energy that weakens over time, which raises the possibility of a Big Crunch; in a highly anisotropic universe, the collapse might not be synchronous but could involve high-density regions hitting localized singularities while voids are still collapsing.
“With the recent DESI and DES data suggesting a preference for dynamical dark energy that weakens over time, the possibility of a Big Crunch could back.”
OpenAI reported a result in which a language model disproved a well-known central conjecture in discrete geometry, raising the question of whether language models could directly contribute to new physics.
“the recent result from OpenAI disproving a well-known central conjecture in discrete geometry”
It is an open question whether the branches of the Everettian multiverse contain the full possibility space of biological evolution.
“Do the branches of the Everettian multiverse contain the full possibility space of evolution?”
Developments in autonomous AI agents—including systems that can coordinate complex tasks and even hire humans online in 'Rent-a-Human'-style arrangements—are raised as potential reasons to update one's credence that advanced AI could become an existential risk to humanity.
“developments in autonomous AI agents — including systems that can coordinate complex tasks and even hire humans online ("Rent-a-Human"-style agents) — have updated your view meaningfully”
Carroll has expressed that he does not find 'necessity' a compelling explanation for the constants of nature and that they could easily have been different, while also treating possibility as an epistemic concept rather than an ontological one—raising the question of whether 'the constants could have been different' means 'our knowledge underdetermines their values' or that we will not find deeper laws from which the constants are derived.
“you said that you don't find "necessity" to be a compelling explanation for the constants of nature, and that they could have easily been different. I believe you also consider possibility to be an epistemic concept rather than an ontological one.”
Boredom and tiredness may not be merely useful behavioral markers of consciousness but could point to something deeper: that consciousness requires an ongoing self-relation across time with needs or desires that can fail to be satisfied.
“You've suggested that one reason current AI systems probably aren't conscious is that they don't get bored or tired.”
Given the cost differential between conventional and contemporary drone warfare, the withered U.S. arms production capacity, and China's dominance in manufacturing the batteries, electric motors, and drones required, the drone era raises the question of whether nuclear conflict becomes more likely, warfare in general becomes less likely, or international power shifts.
“the Peoples Republic of China's dominance in all areas of manufacturing required to produce batteries, electric motors, and drones, what is your outlook on the future of war and international power”
In the spirit of the physics of democracy, there is a question of whether societies should strive to be fractal or scale-less—applying layers of self-governance, identity, legislation, and values with approximately equal weight across all scales from families to the Earth—or whether certain scales such as the nation should be privileged.
“is there value in having the layers of self-governance, identity, legislation, values, personal connections etc. apply with approximately equal weight at scales from families and local communities through neighborhoods, cities, regions, countries, all the way to continents and the Earth”
The way human cognition works may limit how we approach the laws of nature, raising the speculative question of what kind of science a differently-minded creature like an octopus or dolphin would create.
“what if the way our cognition works is limiting how we approach the laws of nature? what kind of science an octopus or a dolphin would create?”
It is an open question whether cold fusion belongs in the same category of impossible things as a perpetual motion machine.
“Does cold fusion belong in the same category of impossible things as perpetum mobile?”
It is an open question whether the Superconducting Super Collider would still be a worthy undertaking today, assuming funding were secured, or whether its merit was predominantly predicated on goals since made redundant by the LHC.
“do you feel the SSC would still be a worthy undertaking today (with the assumption funding was secured) or was its merit predominantly predicated on goals since made redundant by the LHC?”
The questioner asks why Leonard Susskind and others have stopped talking about the holographic principle in recent years and whether it has been disproven.
“Why have Dr. Leonard Susskind and others stopped talking about the holographic principle in the last few years? Has the holographic principle been disproven?”
The questioner asks for an explanation of what the classical limit is and how it is calculated.
“Can you explain what the classical limit is and how it's calculated?”
Since fermions are things that take up space and bosons are things that don't, the questioner asks whether helium is therefore a fermion.
“Fermions are things that take up space, and bosons are things that don't. So helium is a fermion, right?”
If there were practically infinite universes, it is an open question whether mathematics would necessarily be useful in all of them.
“If there were practically infinite universes would math necessarily be useful in all of them?”
The late 19th and early 20th century appears to have been a particularly fertile period for the adoption of philosophical and paradigm-shifting ideas across science, economics, and politics, raising the question of what about society at that time made people more open to radical new ideas.
“the late 19th century, early 20th century was a particularly fertile period for adoption of more philosophical ideas in areas such as science, economics, politics”
If alien life is relatively common but is millions or billions of years more technically advanced than the human race, we might be unable to distinguish them from what we perceive to be nature.
“What If alien life is relatively common but as they are millions of billions of years more technically advanced then the human race currently we can not tell them from what we perceive to be nature?”
If a UFO used gravitational propulsion as a non-reactive system manipulating spacetime, and Earth follows a complex trajectory through spacetime due to many gravitational influences, then for an object existing outside gravity's influence to follow Earth it would somehow need to know Earth's specific spacetime trajectory, raising the question of what mechanism could bridge two objects separated by gravity in this way.
“in order for an object that is existing outside of the influence of gravity (UFO) to "follow" an object that is being influenced by gravity (earth), wouldn't it also need to the know that very specific "trajectory" through space-time in some way?”