What this covers

This monthly Ask Me Anything episode brings together questions from Patreon supporters on a wide range of topics, with Carroll selecting them based on what he finds substantively interesting to address rather than the merit of the questions themselves. The session spans physics, philosophy, and adjacent territories, touching on quantum mechanics, cosmology, relativity, epistemology, evolution, and ethics. Throughout, Carroll applies a consistent philosophical approach: rejecting simplistic rules and embracing the messiness of what actually works in both science and reasoning.

The scope runs from technical quantum mechanics — clarifying that the Heisenberg uncertainty principle is not about measurement disturbance but a mathematical fact of Hilbert space, and defending Many-Worlds Interpretation as a genuine physical theory rather than mere interpretation — to foundational cosmology and the Arrow of Time. Carroll revisits Karl Popper's falsifiability as a tool scientists misuse as bumper-sticker philosophy rather than an accepted criterion, and addresses how Bayesian epistemology works in practice: priors are necessarily subjective, but likelihoods are not, and the hope is that sufficient data washes out disagreement. He treats the simulation hypothesis and theism seriously as scientific theories, arguing they fail on Bayesian grounds and that misapplying the Principle of Mediocrity through anthropic reasoning amounts to unwarranted overconfidence about what typical observers resemble. Beyond physics, Carroll discusses kin selection and Gene's-Eye View of Evolution, why entropy does not track complexity or knowledge, and how moral reasoning should flow from intuitions refined into systems rather than systems imposed against intuitions.

Sharpest takeaway

Carroll argues that across physics and philosophy, the deepest progress comes from refusing simple slogans and bright-line rules — embracing subjectivity in epistemology, treating 'interpretations' of quantum mechanics as distinct physical theories, judging theism and simulation hypotheses by Bayesian prediction rather than dismissal, and recognizing that intuitions trained on the everyday world mislead us about relativity, quantum mechanics, and cosmology.

  • Bayesian priors are necessarily subjective; there is no rock-hard foundation for knowledge
  • Many-worlds is a physical theory, not an 'interpretation', and locality is the surprising thing to explain, not non-locality
  • Theism and the simulation hypothesis are legitimate scientific theories that fail because they predict a universe unlike ours

The argument · threads79 threads · 80 claims
0.70

Intellectual progress requires assuming some things correct and building on them; Carroll assumes many-worlds is right while acknowledging this bet could be wrong, and Western media attention to tragedies is unfairly biased toward white victims.

2 points
  • Intellectual progress requires taking some things as assumed-correct and building on them; if every academic started every project from zero assumptions they would never make progress, which is why Carroll assumes many-worlds is right and builds foundational work on it, while acknowledging this is a bet that could be wrong.

    “If every single academic, whenever they did a research project started with zero assumptions... you would never make any progress.”

  • The West tends to pay more attention when white people are attacked than when people of color are attacked, so media coverage of tragedies worldwide is not given fairly — though this does not detract from the genuine tragedy in Ukraine or the inspirational courage of its people.

    “we in the West tend to pay a lot more attention when white people are being attacked than when people of color are being attacked, so we do not give fair media coverage to these kinds of tragedies”

0.35

The physical world is objectively real, but human knowledge-development necessarily contains subjective features; there is no objective rock-hard foundation for building unbreakable pure knowledge despite our wish for one.

1 point
  • The physical world is objectively real, but the process by which human agents develop knowledge about it necessarily contains a subjective feature; there is no objective handle or rock-hard foundation on which to build an unbreakable edifice of pure knowledge, despite our wish for one.

    “there's not a rock-hard foundation on which to build an unbreakable edifice of pure knowledge, maybe we'd like that to be true, but I don't think it is”

0.35

Consilience cannot capture weak emergence because consilience is possible with one reality level, whereas weak emergence asserts additional graspable patterns using only macroscopic information.

1 point
  • Consilience cannot capture the meaning of weak emergence because consilience would be possible even with only one level of reality, whereas weak emergence asserts the extra fact that there are additional patterns graspable using only macroscopic information.

    “consilience would be completely possible if there were no higher levels”

0.35

Bayes' theorem is a mathematical necessity given axioms, but Bayesian epistemology claims updating credences from priors is how we should gain knowledge; new insights count as data that change credences.

1 point
  • Bayes' theorem is a mathematical theorem you cannot disagree with given the axioms, but Bayesian epistemology is a broader claim that updating credences from priors is how we should gain knowledge; new insights, not just new empirical data, count as new data that can change your credences.

    “Bayes' theorem says that when we have some credences and some propositions and then new evidence comes to bear, here is a formula that tells you how you change your credences.”

0.35

Everyone is entitled to their own priors but not their own likelihoods; priors come from human brains, and Bayesian epistemology hopes sufficient data washes out differing priors.

1 point
  • Everyone is entitled to their own priors but no one is entitled to their own likelihoods; priors come from human brains embedded in the physical world that simply have opinions, and the hope of Bayesian epistemology is that enough data collection washes out differing priors.

    “everyone is entitled to their own priors, no one is entitled to their own likelihoods”

0.35

Without the past hypothesis (early universe had very low entropy), retrodicting from current macrostate alone would predict past macroscopic momenta reversed, eliminating past-future asymmetry.

1 point
  • Without the past hypothesis (that the early universe had very low entropy), retrodicting from the current macrostate alone would predict the past to look exactly like the future with macroscopic momenta reversed — meaning entropy would increase toward the past, radiation would converge on the sun rather than radiate away, and the asymmetry between past and future would vanish.

    “without the past hypothesis, it would also increase toward the past in exactly the same way”

0.35

Carroll assigns inflationary cosmology substantially lower probability than many-worlds interpretation, while noting many physicists flip that ordering; differing expert estimates illustrate how progress is made.

1 point
  • Carroll assigns inflationary cosmology a substantially lower probability of being correct than the many-worlds interpretation of quantum mechanics, while noting many other physicists would flip that ordering — illustrating that different experts holding different probability estimates is how progress is made.

    “I think that the chance that inflation is right is substantially less than the chance that many-worlds is right.”

0.35

In general relativity, coordinate systems have nothing to do with observers; unlike special relativity where observers construct coordinates via rods and clocks, in GR you generally cannot.

1 point
  • In general relativity, coordinate systems have nothing to do with observers; while in special relativity an observer can construct a coordinate system via rods and clocks, in GR you generally cannot, so the origin of a coordinate system need not correspond to any observer's location.

    “Coordinate systems had nothing to do with observers.”

0.35

Freedom of speech was never freedom to say whatever you want on any platform; platforms can restrict behavior on themselves, and legitimate banning for misbehavior requires hard balancing with no simple rule.

1 point
  • Freedom of speech was never the freedom to say whatever you want on any platform you choose; platforms can only restrict what you do on the platform, not free speech itself, and it is legitimate to ban people for misbehavior — the real work is the hard balancing of free speech, social welfare, and platform owners' rights, with no simple rule available.

    “Freedom of speech was never supposed to be the freedom to say whatever you want on any platform you chose.”

0.35

We are almost certainly not missing a spatial depth dimension; we have gone far beyond two-eye perception, inverse square law gravity depends on three-dimensionality, and missing dramatic features would break independent scientific understanding.

1 point
  • We are almost certainly not missing something analogous to depth in our understanding of space, because we have gone far beyond what two eyes tell us — the inverse square law of gravity and electromagnetism depends crucially on space being microscopically three-dimensional, and any missing dramatic feature like a new sense of depth would have caused many independent pieces of our scientific understanding to go wrong.

    “we've gone far, far, far beyond what our two eyes can tell us about space”

0.35

Schrödinger equation treats time and space differently and is compatible with special relativity but spiritually in tension; choosing a reference frame makes Lorentz invariance non-manifest, an arbitrary choice, but underlying dynamics and physical results don't depend on it.

1 point
  • The Schrödinger equation treats time and space differently and is compatible with special relativity but spiritually in tension with it; choosing a reference frame to write a quantum state makes Lorentz invariance non-manifest, an arbitrary choice we always make, but the underlying relativistic dynamics do not depend on it, so no physical result depends on the arbitrary frame.

    “Schrödinger's equation is a formulation of quantum dynamics that does treat time and space differently”

0.35

Weak force time asymmetry is probably not the matter-antimatter asymmetry explanation because the violation is numerically too small; Sakharov's three conditions must work simultaneously at right place and time.

1 point
  • Time asymmetry in the weak force (seen in CPLEAR's kaon decays) is probably not the explanation for the matter-antimatter asymmetry because that violation is numerically too small, and as Sakharov pointed out you need three conditions simultaneously — baryon number violation, time (CP) asymmetry, and departure from thermal equilibrium — working together at the right place and time.

    “to explain the matter-antimatter asymmetry as Sakral pointed a long time ago, you need more than just time asymmetry, you need time asymmetry, Baryon asymmetry... and you need departure from thermal equilibrium”

0.35

Philosophy is continuous with physics; both attempt understanding the world, and physicist and philosopher both asking about quantum mechanics formulation do the same thing regardless of department.

1 point
  • Philosophy is continuous with physics — both are attempts to understand the world — and a philosopher and a physicist both asking what the correct formulation of quantum mechanics is are doing the same thing regardless of department; physics can teach philosophers not to mistake familiar features of our world (like three-dimensional space) for logical necessities.

    “philosophy is continuous with physics, they're not really separate things, they're both attempts to understand the world”

0.35

Imaginary numbers are never strictly necessary in physics but are convenient in quantum mechanics because complex numbers allow wave functions to interfere while maintaining constant probability magnitude.

1 point
  • Imaginary numbers are never strictly necessary in physics since you could always use separate real and imaginary parts, but they are convenient in quantum mechanics because the wave function must interfere (go positive and negative) while keeping total probability equal to one — complex numbers achieve this by circling with constant magnitude in the complex plane rather than oscillating between positive and negative.

    “it's not really going between positive and negative, it's circling around with constant magnitude in the complex plane”

0.35

Humans are built to appreciate risks on timescales shorter than lifespans, so 1% per century of billion deaths is something we prevent poorly; we should worry more about catastrophic non-extinction events like nuclear war, pandemics, and solar flares.

1 point
  • Human beings are not built to appreciate risks occurring on timescales longer than a human lifetime, so a 1% chance per century of a billion deaths is something we are bad at preventing — we should worry more about catastrophic-but-not-extinction events like nuclear war, pandemics, and solar flares, spending a small but prudent fraction of resources rather than large fractions of GDP.

    “Human beings are just not meant, not built to appreciate risks that occur on time scales longer than a human lifetime.”

0.35

Semi-classical gravity is general relativity quantized in weak-field regime; it agrees with all accessible experiments but doesn't solve quantum gravity because genuine problems lie in black holes and Big Bang where it doesn't apply.

1 point
  • Semi-classical gravity is general relativity quantized but restricted to the weak-field regime; it agrees with all experiments accessible to us, but it is not a step toward solving quantum gravity because all the genuine problems lie in regimes like black hole singularities and the Big Bang where semi-classical gravity does not apply.

    “no, it is not a step forward towards solving the problems, 'cause all the problems are in the regimes where semi-classical gravity doesn't really apply”

0.35

Comparing ideas like simulation hypothesis to theology to dismiss them is guilt by association; theism, simulation argument, multiverse, and many-worlds are legitimate scientific ideas judged on right or wrong, not dirty words.

1 point
  • Comparing an idea such as the simulation hypothesis to theology as a way to dismiss it is guilt by association; theism, the simulation argument, the multiverse, and many-worlds are all perfectly legitimate scientific ideas that should be judged on whether they are right or wrong, not dismissed with dirty words.

    “Don't tell me whether something is like theology, tell me if it's right or wrong.”

0.35

Academia should have fewer postdocs rather than fewer grad students; PhDs are education and degrees, but postdocs are mere training, and five-year PhD plus six-to-eight postdoc years makes field-switching hard.

1 point
  • The defensible tweak to academia is to have fewer postdocs rather than fewer grad students; a PhD is an education and degree, but postdocs are merely researcher training programs, and spending five years on a PhD plus six-to-eight years across multiple postdocs makes switching fields hard — professors love postdocs because they boost productivity, but it serves the postdocs poorly.

    “we should have fewer postdocs. As you say, in fields like theoretical physics, anyway, you spend five years getting a PhD and then you get a post-doc”

0.35

Mixing angles in the Standard Model have physical interpretation: up quark emitting W boson converts to linear combination of down, strange, and bottom quarks specified by mixing angle; same applies to neutrino mass-state mixtures.

1 point
  • Mixing angles in the Standard Model have a physical interpretation: when an up quark emits a W boson it converts not into a pure down quark but into a linear combination of down, strange, and bottom quarks, and the mixing angle specifies precisely which combination; the same applies to neutrino mass states being mixtures of electron, muon, and tau flavors.

    “the up quark is actually going to change into some linear combination of all the negatively charged quarks”

0.35

Cosmic microwave background is observable because it comes from everywhere in space, not a single location; it is glowing recombination surface 380,000 years after Big Bang reachable by any light cone backward in time.

1 point
  • The cosmic microwave background is observable because it does not come from a single location but from everywhere in space — it is the glowing recombination surface 380,000 years after the Big Bang — so tracing any light cone backward in time always reaches a glowing point of last scattering regardless of how far apart we now are.

    “the cosmic microwave background doesn't come from some location in space, it comes from everywhere in space”

0.35

Total energy of closed universe in general relativity is exactly zero; positive matter-radiation contribution cancels negative gravitational field energy, so creating baby universe from black hole costs nothing with zero conserved quantities.

1 point
  • The total energy of a closed universe in general relativity is exactly zero — informally a positive contribution from matter and radiation cancels a negative contribution from gravitational field energy — so creating a baby universe from a black hole costs nothing because it has zero conserved quantities, allowing unbounded entropy creation.

    “the energy of the universe is zero, at least in this model... in a closed universe... the total energy is zero”

0.35

Quantum mechanics is best taught hybridly: quantize classical particle to get wave functions in infinite Hilbert space, then use smaller two-dimensional qubit spaces for conceptual ideas; qubits are conceptually easier but disconnect from classical equations.

1 point
  • Quantum mechanics is best taught with a hybrid approach: start by quantizing a classical particle to get wave functions in an infinite-dimensional Hilbert space, then notice that much smaller two-dimensional qubit Hilbert spaces exist and use qubits to introduce conceptual ideas — because qubits are conceptually easier but disconnect from classical equations of motion, while classical quantization connects to prior training but is mathematically harder.

    “conceptually, qubits are a lot easier way to introduce quantum mechanics, the problem with them is they don't connect with anything you've heard of in classical mechanics”

0.35

Deep puzzle is not why quantum mechanics has classical limit but why quantum theories are found by quantizing classical theories; reverse-engineering full theory from limit is non-obvious, and gravity appears a case where this procedure fails.

1 point
  • The deep puzzle is not why quantum mechanics has a classical limit, but why the quantum theories of nature can be found by quantizing a classical theory — reverse-engineering the full theory from its limit; in principle a limit gives only a hint of the general theory, and not every quantum theory arises from quantizing a classical one, with gravity appearing to be a case where this procedure fails.

    “how can you ever reconstruct theory A just given theory B? Reconstruct the full theory from the limit. In principle, there's no reason at all why you should be able to do that.”

0.35

With ordinary probabilities and moderate consequences, multiply probability by utility; but with tiny probabilities and cosmic consequences, simple multiplication breaks down because of probability error bars and unknown possibility spaces.

1 point
  • When probabilities are ordinary (1-99%) and consequences moderate, you can multiply probability by utility and act as a consequentialist; but when probabilities become tiny (10^-6 or 10^-12) and consequences become cosmically good or bad, this simple multiplication breaks down because of error bars on the probability and unknown spaces of possibility — we lack a well-formulated theory for rational decisions under extreme uncertainty and extreme risk.

    “When the probabilities become 10 to the minus six, or 10 to the minus 12 or something like that, but the consequences become cosmically good or bad, it's no longer just a matter of multiplying the probability by the utility.”

0.35

Phase transition in cognitive capacity occurs when brain or system gains enough capacity to be Turing machine; finite resources mean some computations remain impossible despite ability to create new primitives.

1 point
  • There is a phase transition in cognitive capacity when a brain or information-processing system gains enough capacity to be a Turing machine, after which it can run essentially any computation; but finite resources in space and time mean some computations remain impossible, so the ability to create new primitives ad infinitum is overstated.

    “there is this kind of phase transition in reasoning, in cognitive capacity when a brain or an information processing system of some sort gains enough capacity to be a Turing machine”

0.35

Evolution gives weak in-group bias to favor similar people; genes cannot tune themselves to precise quantitative degrees, so kin selection explains general relative preference without literal cost-benefit calculation.

1 point
  • The 'die for two brothers or eight cousins' rubric should not be taken quantitatively because our unconscious is not good at math; rather, evolution gives us a weak in-group bias to favor those more like us, but genes cannot tune themselves to such precise quantitative degrees, so kin selection explains general preference for relatives without literal cost-benefit calculation.

    “our unconscious is not that good at math”

0.35

Pornography is dangerous in bounded way cars and alcohol are; degrading women depiction is symptom not root; real fundamental problem is women not treated equitably, so pornography addressing is band-aid misidentifying cause.

1 point
  • Pornography is dangerous in the same bounded way cars and alcohol are dangerous, and its degrading depiction of women is a symptom rather than the root problem; the real fundamental problem is that women are not treated equitably in society, so addressing pornography as a band-aid misidentifies the cause.

    “the problem is that the women are not treated equitably or thought of equitably in our society. That's the real fundamental problem.”

0.35

Modern quantum foundations do not have 'interpretations'; many-worlds, pilot-wave, and superdeterminism are distinct physical theories with definite ontology and rules; equivalent formulations like Lagrangian-Hamiltonian can be chosen by convenience.

1 point
  • Modern foundations of quantum mechanics do not decide between 'interpretations'; many-worlds is a physical theory with definite ontology (the wave function) and rules (the Schrödinger equation), as are pilot-wave and superdeterminism theories — these are distinct physical theories, and the word 'interpretation' is old-fashioned and no longer applies. Genuinely equivalent formulations like Lagrangian vs Hamiltonian mechanics can be chosen freely by convenience.

    “It's a theory. It's not an interpretation of anything. It's a physical theory.”

0.35

Nuanced patriotism appreciates both good and bad aspects of country's history; teaching national harms makes country strong, and United States is mediocre at this while rightfully proud of aspirational founding values.

1 point
  • A nuanced patriotism appreciates both the good and the bad aspects of a country's history; teaching how badly a nation has treated the rest of the world is part of what makes a country strong, and the United States is mediocre at this, being too sanguine about its harms while still rightly proud of its aspirational founding values.

    “I think that teaching how horrible we have treated the rest of the world is really, really important. And I think that's part of the ability to do that... is part of what makes a country strong.”

0.35

Whether rewinding time replays identical events depends on quantum mechanics formulation: Copenhagen and collapse have randomness so differs; pilot-wave replays identically; many-worlds evolves identically but produces many universes.

1 point
  • Whether rewinding time replays identical events depends on your formulation of quantum mechanics, not the uncertainty principle: Copenhagen and objective-collapse models have genuine randomness so it would differ; pilot-wave (Bohmian) mechanics with the same hidden variables replays identically; many-worlds evolves the universal wave function identically but produces many universes so 'your universe' is ambiguous.

    “it depends on your favorite formulation of quantum mechanics”

0.35

Euthanasia and self-ending should be allowed but controlled, ensuring sound mind, doctor discussion, and clear-day action rather than temporary chemical imbalance; legitimate place for incurable terrible disease sufferers.

1 point
  • Euthanasia and ending one's own life should be allowed but controlled, ensuring the person is of sound mind, has discussed it with doctors, and is acting in the clear light of day rather than from a temporary chemical imbalance — there is a legitimate place for it in a well-formulated society for people facing terrible incurable diseases.

    “I think that it should be allowed, but I think it should be controlled. You really wanna make sure that people are doing this in the clear light of day”

0.35

Decoherence seems misnamed because electron correlated with many particles intuitively seems more coherent; term arose because entangled entities become disentangled or 'decohered' when entangled with environment.

1 point
  • Decoherence seems misnamed because an electron becoming correlated with many particles intuitively seems more coherent, but the term arose because within a set of entangled things (like qubits in a quantum computer), when one becomes entangled with the environment it becomes disentangled or 'decohered' from the others, ruining the intended entanglement among the set.

    “within the set of entangled things, measuring one of them or entangling one of them with the environment, decoheres it or disentangles them with everything else”

0.35

Academy has responsibility informing prospective PhD students of long odds to tenure; solution is not admitting fewer students because that lets school decide for applicants; PhD has value beyond professorship.

1 point
  • The academy has a responsibility to inform prospective PhD students of the long odds (roughly one in four from the best places becoming tenured professors), but the solution is not for graduate schools to admit fewer students, because that makes the school decide for applicants; a PhD has value beyond a professorship and the choice should be the informed student's to make.

    “I don't think the solution is to just let fewer people get PhDs.”

0.35

There is no single good distance notion between observers in expanding universe; 'more than 13.4 billion light-years away' refers to calculable but unmeasurable constant-time slice distance, not physically essential like light cones.

1 point
  • There is no single good notion of distance between two observers in an expanding universe; the figure of being 'more than 13.4 billion light years away' refers to distance along a constant-time spatial slice, which is calculable but not measurable and not physically essential — only light cones are robust and observer-independent in relativity.

    “there's not really any good such thing as the distance between two different observers in an expanding universe”

0.35

Bell's theorem shows no local definite-values theory can be as correlated as quantum mechanics; if you accept many-worlds and wave function describes reality, Bell's theorem is unsurprising because wave functions are inherently non-local.

1 point
  • Bell's theorem shows that no theory of locally definite values can be as correlated as quantum mechanics, which is interpreted as evidence against locality; but if you already believe many-worlds and that the wave function describes the world, Bell's theorem is unsurprising because wave functions are inherently non-local, describing what happens everywhere at once.

    “wave functions are not local, that's just how they are, they describe what is going on all over the place at the same time”

0.35

Ultimate theory with multiple non-overlapping models differs from current quantum mechanics and gravity because QM and gravity overlap in regimes like the sun where both are essential and cannot be separately consistently described.

1 point
  • The hypothetical of an ultimate theory with multiple non-overlapping models differs from today's quantum mechanics and gravity because QM and gravity do overlap in regimes like the sun, where heat and light (quantum processes) and gravity are both essential and cannot be consistently described without including both, even if different languages are used.

    “My favorite regime is the sun. The sun causes gravity, it also causes heat and light, both of these are very important to our lives”

0.35

Explaining suicide requires no dualism or non-determinism; dualist sees mind choosing to destroy body; physicalist accepts physical decision-making account where body decides to destroy itself.

1 point
  • Explaining suicide does not require dualism or non-determinism; the dualist sees a mind choosing to destroy its body, but the physicalist who accepts a physical account of free will and decision-making sees the body deciding to destroy itself, so the physicalist need not interpret it in dualistic terms.

    “If you're a dualist, like you say, you can imagine the mind choosing to destroy the body, if you're a physicalist, the body is deciding to destroy itself.”

0.35

Not hypocritical to accept emergence for chairs but reject for consciousness; consciousness with ineffable first-person 'what it is like' is claimed fundamentally different than describing chairs; a coherent position.

1 point
  • It is not hypocritical for philosophers to accept emergence for chairs but reject it for consciousness, because the whole point of the hard problem is that consciousness — with its ineffable first-person 'what it is like' experience — is claimed to be a fundamentally different kind of problem than describing a chair; Carroll disagrees with the argument but finds it a coherent failure of imagination, not cheating.

    “the whole point of drawing the distinction between the hard problem of consciousness and the easy problems, is that the hard problem is supposed to be a different kind of problem than describing a chair”

0.35

Difference between different infinity cardinalities is mostly irrelevant to physics; only in corners like quantum field theory do countable versus uncountable matter, with no known experimental results depending on it.

1 point
  • To a first approximation the difference between different cardinalities of infinity is irrelevant to physics; only in mathematically careful corners like quantum field theory does countable versus uncountable matter, with no known experimental result depending on it, and infinities larger than the cardinality of the real numbers have no known use in physics.

    “physicists don't care about this stuff at all... the difference between different kinds of infinity is just irrelevant to physics”

0.35

In many-worlds there is set of worlds each with own spacetime and quantum fields; fields are in worlds, not worlds in fields; you cannot say where other worlds are because no space exists in which they exist.

1 point
  • In many-worlds, there is a set of worlds and each world has its own spacetime and its own quantum fields, so the fields are in the worlds, not the worlds in the fields; this is why you cannot say where other worlds are located, since there is no space in which they exist.

    “In many-worlds, there's a set of worlds and each world has its own space-time and its own quantum field, so the fields are in the worlds, not the other way around.”

0.35

Using Bayesian reasoning asking what universe looks like if simulated versus self-contained, Carroll concludes our universe is not a simulation because its resolution far exceeds simulation needs.

1 point
  • Using Bayesian reasoning — asking what the universe would look like if it were a simulation versus a self-contained naturalistic universe — Carroll concludes our universe does not look like a simulation, because its resolution is far better than any simulation would need, and a good Bayesian asks 'is this what I expect' rather than 'can I explain this'.

    “the world that we observe around us, does not look like it would if we lived in a simulation”

0.35

In expanding universe, total matter and radiation energy is not conserved because spacetime changes underneath; redshifted photons lose energy absorbed by spacetime while dark energy gains energy as universe expands.

1 point
  • In an expanding universe the total energy of matter and radiation is not conserved because spacetime is changing underneath it; redshifted photons lose energy that is absorbed by spacetime, while dark energy gains total energy as the universe expands. A conserved energy can be defined only if you include the energy of the gravitational field itself.

    “energy is not conserved. In an expanding universe, the total energy of stuff, of matter and radiation, etcetera, is not a constant number because spacetime is changing underneath it.”

0.35

Distinction between what matter does and what matter is makes no sense; matter is physical universe sui generis with nothing to explain it in terms of; best current description is vector in Hilbert space.

1 point
  • The distinction between what matter does and what matter is makes no sense; matter is the physical universe, which is sui generis with nothing to explain it in terms of, so all of science describes what matter does, and the desire to say matter is something else is a leftover from pre-scientific thinking. Carroll's best current description is a vector in Hilbert space.

    “I don't think this distinction makes any sense between what matter does and what matter is. What matter is is matter. The stuff of the universe.”

0.35

Artificial intelligences generally not embodied matter greatly because bodies are crucially important to who we are; AGI language and intelligence use may differ from ours in ways worth testing rather than assuming.

1 point
  • The fact that artificial intelligences are generally not embodied — existing only in software without complicated biological bodies — matters a lot, because our bodies are crucially important to making us who we are as human beings; we are not just software that can run on any machine, so AGI's use of language and intelligence may differ from ours in ways worth testing rather than assuming.

    “the fact that artificial intelligences are generally these days, not embodied, they don't have bodies, they're just in software... I think this matters a lot”

0.35

Carroll deliberately never repeats podcast guests; breadth and avoiding usual suspects adds more value than institutional memory from favorites, countering rich-get-richer dynamic favoring smooth talkers.

1 point
  • Carroll deliberately never repeats podcast guests because the value added by forcing breadth and avoiding the usual suspects exceeds the value of institutional memory from favorite guests; this counters a rich-get-richer dynamic where smooth talkers get over-invited while less polished but interesting people are overlooked.

    “the value added just by forcing myself to spread out and not just have the usual suspects over and over again, is more than the value added that I would get by having a favorite set of guests”

0.35

Manifest image is under-appreciated and unchanged as scientific image improves; we never apprehend fundamental world directly but always through manifest image of tables, chairs, and photographs.

1 point
  • The manifest image is under-appreciated and does not change as the scientific image improves; we never apprehend the fundamental world directly but always through the manifest image of tables, chairs, and photographs, constantly comparing features of our best scientific practice against that manifest image even when looking through a microscope.

    “Everything that we see is moderated through the Scientific Image, even sorry... The Manifest Image rather.”

0.35

Finite Hilbert space cannot have position-momentum commutator equal to nonzero identity multiple; you can construct analog operators whose commutator approaches canonical as dimensionality grows.

1 point
  • Although a finite-dimensional Hilbert space cannot have a position-momentum commutator equal to a nonzero multiple of the identity, you can construct analog operators whose commutator approaches the canonical one as the dimensionality grows; the universe's Hilbert space dimension is so large (~10^(10^120)) that the finite/infinite distinction is irrelevant for local purposes.

    “in a finite-dimensional Hilbert space, you can construct analogs of position and momentum that are not exactly the same, but they have the same kind of commutator”

0.35

Carroll's interview question skill stems from being sole curator motivated by genuine curiosity plus academic training understanding deeply; deeper than reading for test, not satisfied with words merely sounding like understanding.

1 point
  • Carroll's ability to ask incisive interview questions stems less from special talent than from being the sole curator motivated by genuine curiosity, plus academic training to understand things well enough to do research on them — a deeper level than reading for a test, which means he is not satisfied with words that merely sound like understanding.

    “you get trained to understand things in such a way that you could do research on them”

0.35

Spatial locality is probably not fundamental; philosophers wrongly assume you cannot have interactions, structure, or memories without spatial locality; even space itself may not be fundamental.

1 point
  • Spatial locality is probably not fundamental; many philosophers wrongly take for granted that you couldn't have interactions, structure, or memories without spatial locality, but this is an area where listening to physicists reveals that even the existence of space itself may not be fundamental.

    “you will hear them say that we couldn't have, I don't know, interactions or structure or memories without spatial locality, which I think is obviously not true, 'cause I don't think spatial locality is fundamental”

0.35

Physics curricula lag behind times, teaching what was needed thirty years ago; undergraduate quantum mechanics neglects information, decoherence, entanglement, computing while over-emphasizing perturbation theory.

1 point
  • Physics curricula lag behind the times, teaching now what was needed thirty years ago; undergraduate quantum mechanics textbooks neglect quantum information, decoherence, entanglement, and quantum computing while over-emphasizing perturbation theory and Clebsch-Gordan coefficients, and general relativity should at least be optional for all undergraduates.

    “Things like quantum information, decoherence, entanglement, quantum computing are not mentioned in a lot of undergraduate quantum textbooks, and there's no good reason for that”

0.35

War cannot be solved by smart people as vaccine was because it arises from incompatible desires, not a puzzle; prosperous democratic trading countries less likely to war as shown by France-Germany.

1 point
  • War cannot simply be solved by smart people the way a vaccine was, because war arises from people having incompatible desires rather than from a puzzle to be solved; prosperous, democratic, trading countries are less likely to go to war, as shown by France and Germany no longer being plausible mutual invaders despite centuries of conflict, so war can be minimized through constructing an international order even if never eliminated.

    “The reason why we have wars is because people have incompatible desires”

0.35

Intuitions should mean little for beginning of universe; but intuitions train by learning theories beyond experience, and rational stance with little theory is open-mindedness with priors not wildly different between eternal and beginning universe.

1 point
  • Our intuitions should mean very little when reasoning about the beginning of the universe, but intuitions can be trained by learning theories beyond everyday experience, and Bayesian reasoning still requires placing priors; with little theoretical guidance, the rational stance is open-mindedness with priors not wildly different between an eternal universe and one with a beginning.

    “I don't think our intuitions should mean very much honestly, when we are talking about the beginning of the universe”

0.35

Pair-of-particles popping out picture of Hawking radiation is colorful metaphor not BS; deeper accounts are thermodynamic preservation of conserved quantities while raising entropy, or quantum field instability near horizon.

1 point
  • The pair-of-particles popping out picture of Hawking radiation is a colorful metaphor, not BS, but the deeper accounts are: thermodynamically, exchanging a black hole for a slightly smaller one plus outgoing radiation preserves all conserved quantities while raising entropy; mechanistically, quantum fields near the horizon live in unstable superpositions of particle numbers that leak particles as the black hole shrinks under Schrödinger evolution.

    “that configuration, slightly smaller black hole plus outgoing radiation, has all of the same conserved quantities... And a higher entropy.”

0.35

When talking about interests others may not share, be proud and unapologetic about enthusiasms while recognizing wrong room if no overlap; strategy identical whether knitting or quantum mechanics.

1 point
  • When talking about interests others may not share, you should simultaneously be proud and unapologetic about your enthusiasms while recognizing that if others aren't interested this isn't the right room — the strategy is identical whether the interest is knitting or quantum mechanics, and inability to find any overlap means you're either the problem or in the wrong room.

    “you should simultaneously be proud of your interests, you should not apologize for them... and at the same time, recognize that if other people aren't interested, then this just isn't the room to talk about that”

0.35

Fock space is not different space from Hilbert space but particular Hilbert space summing zero, one, two particle subspaces; only relevant in perturbative regime where field fluctuations look like particles.

1 point
  • Fock space is not a different kind of space from Hilbert space — it is a particular Hilbert space constructed by summing the zero-particle, one-particle, two-particle subspaces, and it is only relevant in the perturbative regime where field fluctuations look like particles; in general circumstances like the Higgs field evolving smoothly, a particle description does not apply.

    “Fock space is still a Hilbert space. It's not a different kind of thing.”

0.35

Simulation argument probabilistic step misuses anthropic reasoning assuming we are typical observers; assuming typical observers like you infers all-observer facts from parochial experience, an unwarranted move.

1 point
  • The simulation argument's probabilistic step misuses anthropic reasoning by assuming we are typical observers; this only seems humble but is actually maximally arrogant because you do not know what a typical observer looks like, so assuming typical observers are like you infers facts about all observers from your parochial experience — an unwarranted move.

    “when you're assuming that you are just a typical observer, you're actually assuming the typical observers are like you, and that's not humble at all... that's sort of maximally arrogant”

0.35

Heisenberg uncertainty principle is not about measurement disturbance; it is feature of any quantum system states whether observed or not, arising because position and momentum are complementary wave function observables.

1 point
  • The Heisenberg uncertainty principle is not about measurements or observations disturbing particles; it is a feature of the states of any quantum system whether observed or not, arising because position and momentum are complementary observables of a wave function — compressing the wave in position spreads out the momentum, a mathematical fact independent of how we observe.

    “The Heisenberg uncertainty principle has nothing to do with measurements or observations.”

0.35

No right or wrong operating system; some suit tinkerers enjoying system guts while others better served by ecosystems allowing focus on work over technical engagement.

1 point
  • There is no right or wrong operating system; some are better for different purposes and users, so Linux suits people who enjoy tinkering with the guts of their system while those who just want to work and focus intellectual energy elsewhere are better served by something like the Mac ecosystem.

    “there's no right or wrong. Some are better than others for different purposes, but different users and different sets of needs are going to be better served by different operating systems.”

0.35

Don't expect reversed arrow in open entropy-decreasing system like cooling champagne in refrigerator because decrease mechanism differs from natural increase; cool water cools further not reforming ice plus warm water.

1 point
  • You should not expect a reversed arrow of time in an open system that decreases entropy (like a champagne bottle cooling in a refrigerator), because the way entropy decreases there differs from natural entropy increase — the cool water just cools further rather than spontaneously reforming an ice cube plus warm water, so reverse-direction life cannot exist in entropy-decreasing regions.

    “There is no sense in which the arrow of time is reversed inside that bottle of champagne”

0.35

AI behavior always accountable by cost-benefit system but need not be recognizable as values to us, like rock following gravity; stay open to both AIs having values and being merely cost-benefit-explicable things.

1 point
  • An AI's behavior can always be accounted for by a cost-benefit system, but that need not be recognizable to us as a set of values — like a falling rock following gravity rather than valuing lower elevation; Carroll stays open both to AIs being sensibly describable as having values and to them being merely things that do things explicable by cost-benefit analysis.

    “that kind of cost-benefit system that the AI might end up having need not be recognizable to us as a set of values”

0.35

Contrary to knowledge and complexity increasing with entropy, they come and go in pockets that disappear; no law forcing them to increase like entropy does; burning book increases entropy without increasing knowledge.

1 point
  • Contrary to the claim that knowledge and complexity increase in lockstep with entropy, complexity and information come and go in pockets that can disappear, and there is no law of nature forcing them to increase the way entropy does — burning a book increases entropy but does not increase knowledge or complexity.

    “If I take a book and I burn it, entropy is increased, but knowledge is not increased”

0.35

Real democracy sign is opposition party can win and leadership changes; Russia same leader across decades while Ukraine changes; non-democracy attacking democracy is bad but long-term solution spreads democracy.

1 point
  • The sign of a real democracy is that the opposition party can win and leadership changes, whereas Russia has had the same leader across decades and Ukraine a different leader each period; the Russian invasion of Ukraine is in part a non-democracy attacking a democracy, which is bad, but the long-term solution is to spread democracy worldwide rather than to erect barriers against autocratic invasion.

    “The opposition party never wins, that's what means you're not really in a democracy.”

0.35

Time and spacetime exist in thermodynamic equilibrium universe without arrow; arrow is behavior feature of matter differentiating past from future, distinct from time itself like space without preferred direction.

1 point
  • Time and four-dimensional spacetime would still exist in a universe at thermodynamic equilibrium, just without an arrow of time; the arrow is a feature of the behavior of matter that differentiates past from future, distinct from time itself, just as space exists without any preferred spatial direction — in equilibrium, macroscopic features are static but particles still move microscopically.

    “I always try to distinguish between time as a coordinate on the space time and the arrow of time, which is the feature of the behavior of matter”

0.35

Philadelphia 76ers came out great in Harden for Simmons trade because Harden, Embiid, Maxey complement perfectly on offense; defensive learning curve exists but Sixers have very good championship chance.

1 point
  • The Philadelphia 76ers came out great in the James Harden for Ben Simmons trade because Harden, Embiid, and Maxey complement each other perfectly on offense, though there will be a defensive learning curve, giving the Sixers a very good chance to win the championship.

    “I actually think that the Sixers have a very good chance of winning it all this year.”

0.35

If information conserved escaping black holes and string theory correct, black hole describable as string configuration degrees of freedom evolving under Schrödinger equation; general idea on right track.

1 point
  • If information is conserved as it escapes black holes and string theory is correct, then a black hole must be describable in string theory as some configuration of degrees of freedom (Mathur's fuzzball of wound-up strings) that evolve under the Schrödinger equation; the proposal is not fully fleshed out for generic black holes, but the general idea has to be on the right track if information is conserved.

    “There have to be a set of degrees of freedom in Hilbert space that interact with the outside world”

0.35

Once accepting more than one universe, no meaningful difference between two and 10^120 universes; only meaningful step is one to two universes, so calling numbers 'absurd' reacts to human brain sound not real objection.

1 point
  • Once you accept more than one universe, there is no meaningful difference between two universes and 10^120 or any larger number; the only meaningful step is from one universe to two, so calling a number of universes 'absurd' is just reacting to how it sounds to a human brain rather than a real objection to the multiverse.

    “once you have two, then there's no difference in my mind between two and 10 to the 120”

0.35

Every particle having antiparticle defined by annihilation is too simplistic; fundamental is particles interacting by Standard Model symmetries and conserved quantities; particle-antiparticle separation is convenient not fundamental.

1 point
  • The idea that every particle has an antiparticle defined by what annihilates it is too simplistic; what is fundamental is a set of particles interacting in accord with the symmetries and conserved quantities of the Standard Model, and separating them into particles and antiparticles is convenient but not fundamental — a neutrino has no electric charge but does carry conserved lepton number that can be positive or negative.

    “this idea that every particle has an anti-particle that is defined by the thing that annihilates it, is way too simplistic”

0.35

Scientific community will reach consensus on quantum foundations within 50 years partly because more intellectual capacity trained on non-intractable problem and partly flippantly because 50 years prevents accountability.

1 point
  • Carroll predicts the scientific community will reach consensus on the foundations of quantum mechanics within 50 years, partly because more worldwide intellectual capacity is trained on the problem than in past decades and it is not an intractable problem, and partly (flippantly) because 50 years is far enough that he cannot be held responsible for the prediction.

    “I still think that we have a lot more IQ points worldwide trained on the problem of understanding the foundations of quantum mechanics than we had in previous decades. And I don't think it's an intractable problem either.”

0.35

Special relativity paradoxes are not real; they arise from habitual pre-relativistic speech treating space and time separately; thinking spacetime first from start makes paradoxes evaporate.

1 point
  • Special relativity puzzles and paradoxes are not real paradoxes; they arise only because we habitually speak in pre-relativistic terms that treat space and time as separate — saying 'a rod one meter long' already cheats by ignoring that real rods are four-dimensional worldvolumes — and as soon as you think in spacetime terms from the start, the paradoxes evaporate.

    “As soon as you start thinking in space time terms from the start, most of these special relativity puzzles and paradoxes just evaporate.”

0.35

Real democracy danger is not invasion by autocracies but internal collapse as populist forces use rhetoric, technology, misinformation, resentment for power disrespecting institutions.

1 point
  • The real danger to democracy is not being invaded by autocracies but internal collapse, as powerful populist forces use rhetoric, technology, misinformation, resentment, and alienation to gain power with no respect for democratic institutions; being in favor of democracy requires accepting that you might not get what you want and persuading others rather than using other mechanisms to seize power.

    “the things that worry me about democracy are not being invaded by autocracies, but internal collapses”

0.35

Democracy and capitalism combination can leave people in lurch not prospering despite GDP rise; healthy democracy must address this by giving vote plus economic opportunity and human rights.

1 point
  • The combination of democracy and capitalism can leave some people in the lurch — not prospering even as GDP rises — and a healthy long-term democracy must address this by giving people not just a vote but economic opportunity and human rights; this does not happen automatically and requires persuasion, cooperation, compromise, and coalition building.

    “the combination of democracy and capitalism can leave some people in the lurch, can leave some people not prospering, even though the GDP goes up”

0.35

Science is methodologically empiricist judging by predictions, not methodologically naturalist; God is judgeable theory but not good theory because world would look noticeably different if God existed.

1 point
  • Science is not methodologically naturalist (assuming natural explanations) — that idea is totally bogus — but methodologically empiricist, judging things by the predictions they make for the world; God is a perfectly good theory in the sense of being a judgeable statement about the world's ontology, but it is not a good theory because if God existed the world would be noticeably different.

    “the idea that science starts by assuming natural explanations for phenomena, 100% wrong, that is a totally bogus”

0.35

Giving all animals same human rights would be crazy; no hard bright line exists but different rights deserved by different groups; crucial human special feature is ability to conceptualize, bargain on future rewards.

1 point
  • Giving all animals the same rights as humans would be crazy (ants should not vote or drive), but there is no hard bright line between animals deserving rights and not, so different rights are deserved by different groups; the most crucial thing making humans special is the species-level ability to conceptualize the future intricately, bargain on future rewards, and be disturbed when their future is threatened.

    “there's no hard bright line there either between animals that deserve rights, and animals that don't, and there could be different rights that are deserved by different groups of animals”

0.35

Popper's falsifiability as science-demarcation criterion was never widely accepted by philosophers; demarcating science from non-science is subtle, and 'beyond falsifiability' point is misused lazily as bumper-sticker.

1 point
  • Popper's falsifiability as a demarcation criterion between science and non-science was never accepted by most philosophers of science, because demarcating science from non-science turns out to be more subtle; the 'beyond falsifiability' point is not that we are discarding a once-agreed idea, but that current scientists misuse Popper lazily as a bumper-sticker rule.

    “Philosophers of science never had a period of time where most of them were onboard with this idea, because it turns out that demarcating science from non-science is more subtle than that.”

0.35

Fundamental GR-QM time issue is problem of time: naively quantizing closed zero-energy GR yields no evolution, so time must emerge; distinct from GR time-parameter freedom which is not fundamental problem.

1 point
  • The fundamental issue connecting GR and QM time is the problem of time: naively quantizing general relativity for a closed universe with zero energy yields no time evolution at all, so time must be made to emerge — a tricky undertaking, distinct from the choice-of-time-parameter freedom in GR which is not itself a fundamental problem.

    “if you just naively quantize general relativity for a closed universe with zero energy, like we were talking about, there's no time evolution at all in that theory”

0.35

Morality comes from starting with intuitions and systematizing them; mistaken to invent rational system then stick with it disagreeing intuitions because system's point was systematizing those intuitions.

1 point
  • Morality comes from starting with moral intuitions and trying to make them into a rational system; it is a mistake to invent a rational system and then stick with it when it disagrees with your intuitions, because the whole point of the system was to systematize those intuitions in the first place.

    “morality comes about by starting with some moral inclinations or intuitions that we have and trying to make them into a rational system”

0.35

Low-entropy early universe looks extraordinarily smooth and homogeneous with same stuff everywhere; gravity-important low entropy because particles clump under gravity becoming inhomogeneous.

1 point
  • A low-entropy early universe looks extraordinarily smooth and homogeneous, with the same amount of stuff at every point; this is low entropy because gravity is important — particles left to evolve under gravity clump together and become more inhomogeneous, so a high-entropy early universe would have been wildly fluctuating rather than smooth.

    “the thing that makes it low entropy is that it is extraordinarily smooth, right? It's homogeneous, the same amount of stuff at every different point”

0.35

Wrong to say time passage varies at different space points; passage is always one second per second; true fact is different time elapses along different paths between same events.

1 point
  • It is wrong to say the passage of time varies at different points in space; the passage of time is always one second per second, and the true fact in relativity is that different amounts of time can elapse along different paths between the same two events — curvy versus straight paths in special relativity, or passing through strong gravitational fields in general relativity.

    “The passage of time is always one second per second, it is never two seconds per second. I don't even know what that would mean.”

0.35

Foundations over-emphasizes Bell's inequality as non-locality evidence; from quantum-gravity perspective you cannot have locality; real question is why physics looks local when quantum gravity shouldn't be.

1 point
  • The foundations community over-emphasizes Bell's inequality as profound evidence for non-locality, but from a quantum-gravity perspective you cannot and should not have locality — a path integral summing over many spacetime geometries cannot pinpoint a single spatial point across them; so the real question is not 'physics is non-local, what do we do?' but 'why does physics look local as much as it does when quantum gravity should not be?'

    “The question we should ask is, 'Why in the world does physics look local as much as it does since gravity, quantum gravity, anyway, clearly should not be local?'”