AMA | July 2022
What this covers
This monthly Ask Me Anything episode ranges across physics fundamentals, cosmology, democracy, philosophy, and epistemology. Sean Carroll addresses a mix of technical questions from Patreon supporters, selecting those where he has substantive material to contribute. The episode announces a scholarship initiative in partnership with Bold.org for students pursuing fundamental physics.
The physics content spans several domains. Carroll discusses Reversibility vs Arrow of Time, clarifying that T-violation in particle physics is not a true arrow because it is reversible under CPT symmetry, whereas the real arrow emerges from the low entropy of the Big Bang. He treats Space Emerges from the Wave Function, arguing that quantum entanglement is simply non-local once you recognize the wave function is a function of all configurations at once. On cosmology, he addresses whether a de Sitter-like far future universe can support structure, explains why dark matter planets are implausible, and evaluates Stephen Wolfram's hypergraph project as intellectually worthwhile but unlikely to succeed. Carroll also examines fine-tuning arguments, Boltzmann brain problems in eternal universes, and Pascal's wager, using Bayesian Credence to show why the wager fails given multiple possible deities.
Beyond physics, Carroll defends Fuzzy Boundaries Are Still Useful, using concepts like "chair" and "free speech" to argue that imprecision does not undermine utility. On democracy, he emphasizes that stability depends on citizens accepting electoral losses and warns that American mechanisms like the electoral college could enable minority rule—a structural threat he identifies as underappreciated. He also discusses persuasion strategy, physics pedagogy (criticizing the rubber-balloon analogy for cosmic expansion), and methodological questions in climate science, where he notes that political pressures can suppress legitimate critiques.
Carroll argues that many seemingly precise concepts in physics, politics, and philosophy actually rest on fuzzy boundaries and self-consistency rather than rigorous foundations, and that this imprecision does not undermine their usefulness; meanwhile he warns that American democracy faces a real, underappreciated structural threat from minority-rule mechanisms.
- Useful concepts (life, death, 'chair', free speech limits) can lack bright-line definitions yet still convey real information
- T-violation is not an arrow of time because it is reversible (CPT-preserved), whereas the true arrow comes from the low-entropy Big Bang
- Democracy's stability depends on people accepting losses, and electoral-college/legislative mechanisms could enable minority rule
The term 'qualia' is too ambiguous to use precisely in consciousness discussions, and democratic values are universal human ideas not inherently Western, as democracies have existed globally and autocracy dominated the West for most of history.
- Carroll avoids the word 'qualia' because its use is ambiguous: sometimes it implies experience is by definition non-physical and sometimes it doesn't, so it refers to different things for different people and lacks the precision needed for careful discussion of consciousness.
“the definition or the use of the word qualia is a little bit ambiguous as to whether it is supposed to represent something that is by definition non-physical, and that's why I don't like to use it, because it's just a vague word”
- It is bogus to frame democratic values as inherently 'Western' and contrast them with 'Eastern' or 'Chinese' values; the Western world has been mostly autocratic over its 3000-year history, democracies have existed all over the world, and the foundational idea of democracy—that authority comes from the will of the people rather than a special elite—is a universal human idea that can be implemented in many cultural forms, as Japan demonstrates.
“it's just completely bogus to think of democratic values as Western and to contrast them with Eastern or Chinese or Asian values”
The idea of inverted color spectra is meaningless because seeing red is just detecting corresponding wavelengths with no objective fact beyond that to invert.
- The idea of one organism's color spectrum being 'inverted' relative to another's is meaningless, because there is no objective fact to 'seeing red' beyond detecting the wavelengths corresponding to red, so there is nothing that could be inverted.
“I don't think there's any objective meaningfulness to saying I'm seeing red, other than the fact that I'm seeing those wavelengths of life corresponding to red, so I don't know what it could possibly mean to be... A color spectrum being inverted”
Democracy's core is everyone accepting outcomes when they lose, and its fundamental instability is that widespread rejection of democracy can end it from within with almost no remedy.
- The most important and underappreciated part of democracy is not that the majority gets what it wants when it wins, but that everyone agrees to live with the outcome when they lose; democracy is fundamentally about living with people you disagree with, and its core instability is that if too many people stop buying into democracy, they can end it from within with almost no available remedy.
“there's another part of democracy that is equally important, maybe more important, which is that if you lose, you have to live with it. That's the part that matters”
Quantum entanglement is non-local because the wave function is of all possible configurations at once and cannot decompose into location; the real puzzle is why the universe appears local.
- Quantum entanglement is simply non-local because the wave function is a function of every possible configuration of everything at once and cannot be decomposed into 'here' and 'there'—space emerges from the wave function rather than the wave function living in space. The genuine puzzle is therefore not how to live with non-locality but why the universe appears local at all, a question made even sharper by quantum gravity.
“Wave functions are not local... The wave function is a function of every possible configuration of everything in the universe all at once. Space emerges from the wave function, not the wave function lives in space”
Fine-tuning argument probabilities are entirely bogus because probability distributions over physical constants are made up, treated as independent unjustifiably, and assume unjustified things about life's possibility.
- The fine-tuning argument's claimed probabilities (like 10^-223 for a life-supporting universe) are entirely bogus because the probability distributions for physical constants are simply made up, treated as independent when they are not, and assume both that we can calculate the probability of a universe like ours and that we know the probability life can exist—all unjustified; particle physicists make the same error.
“they completely pull out of their butts some probability distribution for these numbers”
Technology contributes to democratic decline by showing people vastly different lives without letting them live them, increasing alienation and opening doors to populist anti-democratic figures amid population size and inequality.
- Technology contributes to democratic decline by giving people a vivid, accessible taste of others' very different lives (via social media) without giving them the ability to live those lives, increasing alienation and feelings of powerlessness; combined with sheer population size and rising inequality, this opens the door to populist, anti-democratic figures who claim to stand up for the ignored individual.
“what technology is giving us is a taste for the multiverse. Because we can look at other people's lives... without giving us much more ability to live those lives ourselves”
The omniscient yes-or-no question thought experiment poorly models knowledge because without understanding the world's fundamentals we cannot even know which questions to ask, and deep truths require elaborate theoretical frameworks.
- The thought experiment of asking an omniscient being one yes-or-no question about physics is a poor model of how knowledge works, because without understanding how the world works at a fundamental level we do not even know which questions to ask; deep truths like quantum mechanics or how the universe began require elaborate theoretical frameworks, not answerable yes/no questions.
“without knowing how the world works at the fundamental level, we don't even know which questions to ask”
Life is a temporary dynamical process like a way of characterizing physical things, not a substance; death is the stopping of that process, and vague boundaries do not make the concept useless.
- Life is a dynamical process, not a substance or essence; like energy it is a way of characterizing physical things rather than a thing itself, but unlike energy it is temporary and not conserved. Death is the stopping of this process, and the fact that the boundaries of life and death are not cut-and-dried does not make the concept any less useful.
“I think that life is a process, I don't think it is an essence”
Inferring low early-universe entropy from observations is not circular but self-consistent: assuming entropy increases lets us use today's light observations to infer the early state, producing a coherent fruitful picture.
- Inferring the low entropy of the early universe from observations is not circular reasoning but self-consistent: we assume the broad principle that entropy has been increasing, then use observations of light reaching us today to infer the specific low-entropy early state; this assumption could be wrong (we cannot rule out being Boltzmann brains or a last-Thursday creation) but making it produces a coherent, fruitful picture, which is the nature of science.
“we're not assuming the specific low entropy early state, we're assuming the basic rough idea that entropy is increasing, then taking the observations that we see and inferring a specific kind of low entropy state. So it's not a completely circular argument, but it is a self-consistent argument”
There is no infinite regress for the universe's origin because fundamental physics is reversible patterns, not cause-then-effect, so an eternal universe simply always existed requiring no prior explanation.
- There is no infinite regress problem for the origin of the universe because the everyday principle that effects must be explained by prior causes does not apply to fundamental physics; the laws of physics are not cause-then-effect but a reversible pattern that, given a state at one moment, can be pushed forward or backward, so an eternal universe with no initial time simply always existed and requires no prior explanation.
“there's no infinite regress problem because there's no principle that says I must explain the conditions at one moment in terms of previous moments”
The arrow of time is not imported into relativity but assumed when applying it; physicists put 'future' in by hand despite equally valid past-focused solutions, reflecting the thermodynamic arrow not relativity itself.
- The arrow of time is not imported into relativity, but physicists implicitly assume it when applying relativistic causality: when a physicist says poking the electromagnetic field sends consequences into the future light cone, the word 'future' is put in by hand, because there is an equally valid solution where ripples focus in from the past—the choice reflects the thermodynamic arrow of time, not relativity itself.
“the relativistic notion of causality is ingrained in relativity. There are light cones, but the way that it is used is very, very much colored by the fact that physicists know there's an arrow of time”
Quantum mechanics already unifies with classical mechanics as the unifying theory; classical is the limit where QM applies, and QM is always true while classical is only a good approximation under certain conditions.
- There is no need to unify classical and quantum mechanics because quantum mechanics is already the unifying theory: classical mechanics is a proper subset, the limit of quantum mechanics applicable under certain circumstances. The correct statement is not that QM applies only to small things but that QM is always true while classical mechanics is a good approximation only under a subset of conditions.
“we don't need a unifying theory theory for classical and quantum, because we have one. It's called quantum mechanics. Classical mechanics is the limit of quantum mechanics”
US system instability arises from mechanisms protecting minority rights being hijacked for minority rule, creating a self-sustaining positive-feedback loop pushing away from democratic representation.
- The instability of the US system is that mechanisms designed to protect minority rights (Senate, electoral college, Supreme Court) can be hijacked to enable minority rule, for example state legislatures planning to override popular vote results in the electoral college, creating a self-sustaining positive-feedback loop that pushes the country away from democratic representation.
“you have to distinguish between protecting minority rights and allowing minority rule. So the obvious instability of a setup like ours is if the mechanisms that have been put in place to make sure the minority has a voice, are hijacked by the minority to take power and keep it”
Carroll spends roughly one day per week on the podcast including reading guests' books, varying by interview, mainly to avoid missing genuinely interesting material rather than controlling what guests say.
- Reading guests' books is included in the roughly one day per week Carroll spends on the podcast, though this is an average since some interviews require more reading than others, and the main reason to read is to avoid missing genuinely interesting material rather than to control what the guest says.
“when I say it's one day per week, that's an expectation value, that's an average”
The block universe does not imply eternal life because it means only that every moment is equally real, not real simultaneously; immortality would require living through all moments, which you do not in a block universe.
- The block universe does not imply eternal life in any sense, because it only means every moment of time is equally real, not real simultaneously; different moments are real at different moments. Immortality would require living through all moments, which in a block universe you do not, so there is no reason to connect the block universe to immortality.
“the block universe just means we treat every moment of time as being equally real... different moments of time are real at different moments of time. They're just equally real”
You cannot formulate Many-Worlds while agnostic about branch existence, because accepting a realist wave function, unitary evolution, and no hidden variables commits you to predicting the worlds' existence.
- You cannot formulate Many-Worlds while remaining agnostic about whether the other branches exist, because if you accept a realist wave function, unitary Schrödinger evolution, and no hidden variables, the theory predicts the existence of the worlds; agnosticism is not a coherent scientific stance—you either accept the theory and its prediction of worlds or you must change the theory.
“either you believe that theory and you believe in the worlds or you don't believe in the theory, so you have to change the theory somehow. You can't just be agnostic about what the theory is predicting”
A coexisting dark matter world is very unlikely because dark matter does not interact strongly enough to build structures; observations show it diffusely spread, and exotic possibilities like dark atoms remain limited intellectual exercises.
- A coexisting world of dark matter planets, life, and spacecraft is very unlikely because dark matter is dark in the sense of not interacting strongly with anything, and building structures requires particles that can interact and dissipate effectively; observations show dark matter is diffusely spread rather than concentrated in planets and stars, though exotic possibilities like dark atoms (made of two heavy oppositely-charged particles) remain an interesting but limited intellectual exercise.
“in order to build stars and planets and spacecraft, you need the dark matter particles to be able to interact with each other and dissipate fairly effectively. And dark matter doesn't do that”
Wolfram's hypergraph project deserves support as fundamental physics needs varied approaches, but unlikely to succeed because non-QM programs either rediscover QM or fail, and major leaps usually come from empirical nudges not bold guessing.
- Stephen Wolfram's hypergraph project is worth supporting because fundamental physics needs more variety of approaches, but it is unlikely to succeed because any program starting from something other than conventional quantum mechanics will either just rediscover quantum mechanics, or differ from it and probably fail since QM is so successful, and historically major leaps come from being nudged by empirical data rather than from guessing the right bold idea.
“when you have a program that starts from something other than conventional quantum mechanics... there are two possibilities. One is what he will find is that they just rediscover quantum mechanics... Or, alternatively, they will not discover quantum mechanics... but then probably it's not gonna work, 'cause quantum mechanics is super-duper successful”
Deep learning can contribute scientifically like data collection, but scientific understanding requires principles and equations you understand, not just correlations; a deep learning model cannot predict new results from learned patterns.
- Deep learning methods can contribute to scientific progress the way Tycho Brahe's data collection did, but they are not the whole story because true scientific understanding requires principles, equations, and relationships you understand—like Kepler's laws—not just correlations; a deep learning model trained on planetary motions could not predict a new planet's orbit because it only fits curves rather than capturing the underlying law.
“true scientific understanding is not just a correlation between bits of data that you can discover by doing a deep learning algorithm, it's a principle, it's an equation, it's a relationship that you understand why it is like that”
Observed T-violation in particle physics is not an arrow of time because it is perfectly reversible under CPT symmetry; the true arrow comes from irreversibility arising from the Big Bang's low entropy.
- The observed T-violation in particle physics (such as the B meson oscillating faster than its reverse) is not an arrow of time because it is perfectly reversible: there exists a time-reversal-like operation (CPT) that is exactly preserved by the laws of nature, reflecting underlying microscopic reversibility, whereas the true arrow of time is the irreversibility arising from the low entropy of the Big Bang.
“it's not narrow of time, and the reason why is because it's perfectly reversible”
GPT-3's haiku attempt shows it is nowhere near conscious or smart because it violated syllable structure and ignored deeper haiku conventions, merely correlating texts without understanding.
- GPT-3's attempt to write a haiku in Carroll's style demonstrates it is nowhere near conscious or even smart, because it produced a three-line poem that violates the syllable structure and ignores the deeper haiku conventions about seasons and the punchy turning word; all it does is correlate pre-existing texts without understanding anything.
“it's a wonderful example of how things like GPT are nowhere near conscious or even sensible or even all that smart”
Carroll's worldview rests on self-consistency with two pragmatic assumptions: the world is intelligible and sense data tells us something true about it, justified pragmatically by working rather than with perfect certainty.
- Rather than foundationalism, Carroll's worldview rests on self-consistency, starting from two pragmatic assumptions: that the world is intelligible (there are patterns and ways to make sense of it) and that our sense data tells us something true about the world (not the unvarnished whole truth, but not completely false either). These are justified pragmatically by working by their own lights, never with perfect certainty.
“the two things I kind of start with are the idea that the world is intelligible... And the other is that our sense data, our input tells us something true about the world”
De Sitter space being maximum-entropy is unsurprising once recognizing high-entropy states are smooth, it is quantum-mechanical, and entropy is assigned phenomenologically via Hawking's formula before fully quantizing gravity.
- De Sitter space being a maximum-entropy state is not counterintuitive once you recognize that high-entropy states are smooth (like a gas spread through a box), that the de Sitter state is quantum-mechanical rather than classical, and that the entropy is assigned phenomenologically via Hawking's relation between horizon area and entropy—we know the macroscopic entropy without yet knowing the underlying microstates because gravity is not fully quantized.
“Our current understanding of the entropy of a thing like De Sitter space or a black hole is similar to that phenomenological entropy. We know the macroscopic rules and we can assign an entropy to it, but we don't know the complete microscopic theory”
Framing arguments around mainstream media complaints signals not to take them seriously, because non-mainstream media typically has more biases and errors; credibility depends on source reliability and likely biases.
- When people frame arguments around complaints about the mainstream media, that is a signal not to take them seriously, because while mainstream media has real biases and makes mistakes, non-mainstream media typically has more biases and is more prone to error; the right question is not mainstream versus non-mainstream but how credible any given source is given its likely biases.
“when people start complaining about the mainstream media, I almost immediately know to not take seriously what they're saying”
To dissuade from myths, lead with the true statement rather than the myth, because people tend to remember the myth when you present it first rather than its refutation.
- Studies suggest that to dissuade people from believing a myth you should lead with the true statement rather than the myth, because when you lead with the mistake people tend to remember the myth rather than its refutation; you should state the truth first and present the myth only as a contrast.
“you're leading with the myth, with the mistake, and very often... when you do that, the thing that people remember is the myth, not the refutation of the myth. So you have to lead with the true thing first”
Boltzmann's formula alone does not give the second law's content because it is time-symmetric and says nothing about the future; the content requires the past hypothesis plus non-rare-entropy-decrease assumption.
- Boltzmann's formula alone does not give the second law's content, because the formula is time-symmetric and says nothing about the future; the real content of the second law—that entropy was lower yesterday and will be higher tomorrow—requires augmenting Boltzmann's formula with the past hypothesis plus a requirement that we are not in the rare entropy-decreasing states.
“Boltzmann's formula doesn't say anything about the future, it just says that the entropy is proportional to the log of the number of micro-states... Anything you say about the future, if that's the only thing you know... you should also say about the past”
The US Constitution was good for its time but imperfect; while a better one could be written, current political climate makes success unlikely, so fixing flaws through amendments is preferable to wholesale replacement.
- The US Constitution was amazingly good for its time, but it is not perfect and its original flaws can worsen as society changes; while it would be possible in principle to write a much better constitution from scratch, Carroll has low credence that attempting it under the current political climate would succeed, so he favors fixing flaws through the existing amendment mechanism rather than wholesale replacement.
“it would be absolutely possible to sit down from scratch and write a new constitution... that would be much better than the one we have right now. On the other hand, I have very low credence that that would actually happen if we tried”
Physics can discover fundamental laws with no obstacles in sight, but physics will never finish because manifesting those laws in the messy macroscopic emergent world is almost infinitely varied.
- Physics can in principle discover the fundamental laws of nature with no obstacles in sight, but physics as a whole will never be finished because the manifestation of those laws in the messy, emergent macroscopic world (including all of biology as a subset) is of almost infinite variety, so even with the fundamental laws known, working out their implications would take effectively forever.
“the way that those fundamental laws manifest in the messy macroscopic world, in the emergent world... is of almost infinite variety”
Explaining consciousness fully requires biology and neuroscience, but whether known physics laws must change for consciousness is a legitimate physics question; Carroll holds existing physics is sufficient, doubting those who think otherwise.
- While explaining consciousness fully requires biology and neuroscience, the specific question of whether the known laws of physics must be changed to account for conscious experience can legitimately skip biology; Carroll holds the laws of physics are perfectly up to the task and that those who think otherwise vastly underestimate how hard it is to change the laws of physics.
“I think that the laws of physics are perfectly up to that task, but number two, people who don't think that they are vastly underestimate the difficulty of changing the laws of physics”
A white hole is a black hole played backward in time with a past singularity; white holes do not exist around us because they decrease entropy, but the Big Bang is highly analogous to one.
- A white hole is exactly a black hole played backward in time: a region of spacetime things can come out of but not enter, with a singularity in the past rather than the future. White holes don't exist around us because they decrease entropy and are thermodynamically crazy, but the Big Bang itself is highly analogous to a white hole because it starts with low entropy.
“A white hole is exactly a black hole, played backward in time”
Supreme Court justices should have term limits rather than lifetime appointments, because lifetime terms incentivize ideologically compatible young justices, while fixed eight-to-ten-year terms would remove that incentive.
- Supreme Court justices should have term limits rather than lifetime appointments, because lifetime terms create a huge incentive for presidents to choose ideologically compatible young justices who will serve for decades; a fixed term of eight or ten years would remove that incentive and encourage choosing the best people, with no loss of judicial independence.
“term limits for Supreme Court justices make absolute 100% sense. There's zero reason why Supreme Court Justices should have lifetime terms”
An eternal universe need not be repetitive; like infinite integers appearing once each, an eternal universe can have infinitely many distinct unrepeating moments, which Boltzmann brain considerations suggest we observe.
- Even if the universe is eternal, it need not be repetitive; like the infinite integers where each number appears only once, an eternal universe can have infinitely many distinct moments that never recur, and Boltzmann brain considerations suggest a genuinely recurring universe would mostly produce small fluctuations near equilibrium rather than the large divergence from equilibrium we observe.
“the integers, the numbers zero, one, two, three, four... an infinite number of integers, and they never repeat... And the universe could be like that, it could be eternal, but no one moment ever repeats”
The 'axial Higgs boson' or 'magnetic Higgs' is not a new standard model particle but a Higgs mechanism in condensed matter, completely different from fundamental physics, misrepresented as beyond the standard model.
- The 'axial Higgs boson' or 'magnetic Higgs' that got press coverage is not a new particle in the standard model; it is an instance of a Higgs-type mechanism appearing in a condensed matter system, completely different from fundamental particle physics, and the way it was reported was misleading PR that implied a discovery beyond the standard model.
“It's not a new particle in the standard model... it was a condensed matter system, nothing like the fundamental particle physics”
Pascal's wager is bad reasoning because it only works assuming atheism or traditional Christian God, but many possibilities exist including a God rewarding intellectual atheism, making probability assignment impossible.
- Pascal's wager is bad reasoning because it only makes sense if the sole choices are atheism or the traditional Christian God, when in fact there are many more possibilities—including a God who rewards intellectually honest atheism and punishes hypocritical belief, or non-heaven-hell deities—so without a probability distribution over all these possibilities the wager cannot be accurately made.
“it would make sense if you thought that the only two choices were atheism or the traditional Christian God, but of course there are way more choices than that”
Chalmers's objection that the multiverse bumps fine-tuning up a level is wrong; a single simple generating principle like string theory plus inflation produces all universes without needing fine-tuning itself.
- Chalmers's objection that the multiverse merely bumps the fine-tuning problem up a level (requiring the multiverse itself to be fine-tuned) is wrong, because the whole point of a multiverse is a single simple generating principle—like string theory plus inflation—that produces all sorts of universes, just as the single operation of adding one generates all the infinite integers without needing to be fine-tuned.
“The whole idea of a multiverse is there is some simple generating principle that creates all sorts of different universes”
De Sitter space fluctuations could create universes but do not plausibly explain ours because eternal fluctuating space creates individual people vastly more often than entire universes, predicting we should observe a smaller universe.
- While fluctuations in de Sitter space could in principle create an entire universe like ours, this is not a plausible explanation for our universe's origin because it leads to the Boltzmann brain problem: in an eternal fluctuating space you would create individual people, planets, or small subsystems vastly more often than entire universes, so the theory predicts we should live in a much smaller universe than we observe.
“you're gonna make individual people or planets way more often than you're gonna make entire universes, so that's a theory that could explain the origin of our universe, but that theory makes predictions, and the prediction is, we don't live in the whole universe like that, we live in a much, much, much smaller universe”
The electoral college should be eliminated because original justifications no longer apply; population disparities are far larger and small states are already ignored under the current system anyway.
- The electoral college should be eliminated because the original conditions justifying it no longer apply: the population disparities between states are far larger than in the 1790s, and the counterargument that small states would be ignored is bogus since under the current system any state with a predictable outcome is already ignored, whereas a direct popular vote would give equal reason to campaign everywhere.
“the rules are dumb. The rules don't work anymore. They don't apply to... The conditions that originally made people want to put these rules in place, no longer apply”
When you cannot identify direct causation for personal risk, indirect statistical methods across other people can detect hidden dangers and inform risk accounting despite uncertainty.
- When you cannot identify a direct causal mechanism for a personal risk (such as whether running caused a stroke), you can still use indirect statistical methods—looking at whether the relationship holds across other people's experiences—to detect that a danger could be present and take it into account; the best you can do under uncertainty is minimize that uncertainty.
“even in the absence of a direct mechanism to explain where the danger is coming from, you can still use indirect methods to see that there could be a danger there and take it into account”
Apparently sudden mind changes are surface manifestations of gradual underlying change; conscious beliefs appear constant while sub-beliefs shift until triggering apparent phase transitions, explaining why persuasion studies miss long-term effects.
- Apparently sudden changes of mind are actually the surface manifestation of gradual underlying change; like a slow-cooking pork shoulder whose temperature plateaus while its moisture content changes invisibly, our conscious beliefs can appear constant while sub-beliefs shift gradually until they trigger an apparent phase transition—which is why short-timescale studies of persuasion may miss effects that linger and change minds years later.
“we think our beliefs are constant, but there's a set of sub-beliefs... lying beneath the surface that change gradually... and then what it appears to you like is a sudden phase transition”
The Boltzmann brain problem asks whether a satisfied condition arose as random fluctuation or thermodynamic evolution; in eternal fluctuating universes, almost any condition most likely arose as random fluctuation.
- The Boltzmann brain problem does not arise from asking what most fluctuations will be like (most won't satisfy any given condition), but from asking the reverse: given some condition is satisfied, is it more likely a random fluctuation or a thermodynamically sensible evolution? In an eternal fluctuating universe, almost any specified condition is most likely to arise as a random fluctuation rather than from a sensible low-entropy history.
“It comes from asking the other question, the other way, given some conditions, are they likely to be random fluctuations or is it likely to be thermodynamically sensible?”
Scientists at each career stage are selected for previous-stage skills rather than current ones; postdocs should deliberately develop next-stage skills rather than repeating what made them successful grad students.
- Scientists at each career stage are trained and selected for the skills of their previous stage rather than their current one: undergraduates solve problem sets, then are selected for grad school on that basis where they instead must do assigned research, then selected as postdocs where they must invent their own research problems, then as faculty must mentor, teach, and apply for grants—so a postdoc should deliberately develop the skills needed for the next stage rather than repeating what made them a successful grad student.
“just like generals are always planning for the last war, students and young scientists are always being trained for their previous position”
Inability to draw hard bright lines for concepts does not make them useless; fuzzy boundaries like 'chair' still convey real information and give useful handles on the world.
- Just because you cannot draw a hard bright line for a concept does not make it useless; like the concept 'chair,' where no precise number of atoms separates chair from non-chair, the boundary may be approximate and fuzzy yet the concept still gives real handles on the world and conveys useful information. The seductive demand for clarity (seen in games and conspiracy theories) should not override accepting necessary imprecision.
“Just because you can't draw a hard bright line doesn't mean the concept of chair is not useful”
Quantum mechanics's claim that decay time cannot be predicted with certainty even with complete knowledge is made by the well-defined theory; discovering predictability would mean quantum mechanics is wrong.
- Quantum mechanics can claim that a radioactive atom's decay time cannot be predicted with certainty even given complete knowledge, because that claim is made by the well-defined theory of quantum mechanics, not by reality; if we ever discover we can predict decay with certainty, it will be because quantum mechanics is wrong, not because we misunderstood what it says.
“it's the difference between the claim being made by quantum mechanics and the claim being made by reality... according to the rules of quantum mechanics, you cannot predict with certainty when a radioactive atom will decay”
Fine-tuning is the best God argument by Bayesian rules but still fails: God would not need to fine-tune for life, and theistic predictions diverge sharply from observations.
- The fine-tuning argument is the best argument for God because it plays by Bayesian rules, but it still fails for two reasons: God would not need to fine-tune a universe for life since God could create life in any universe (the argument only works if you are already a naturalist), and a consistent Bayesian must consider all the other data, which makes theistic predictions diverge sharply from the universe we actually observe.
“God wouldn't need to fine-tune the universe for life to exist. The fine-tuning argument only works if you're already a naturalist or a physicalist about life”
In the far future de Sitter-like universe, space expands but stops cooling because temperature becomes a vacuum property via cosmic horizon, analogous to black hole event horizons giving temperature.
- In the far future de Sitter-like universe, space continues expanding but stops cooling, because the temperature is no longer a feature of physical matter (like the cosmic microwave background photons) but a property of the vacuum itself due to the cosmic horizon surrounding every observer, analogous to how a black hole's event horizon gives it a temperature.
“the universe does not stop expanding... but it does stop cooling, and the reason why is because the temperature is no longer a feature of the physical matter”
Even with finitely many states, no smallest non-zero probability exists because quantum states are complex vectors with superposition coefficients that can be arbitrarily close to zero or one.
- Even if there are only finitely many degrees of freedom and thus finitely many states, this does not imply a smallest non-zero probability, because in quantum mechanics states are vectors and superposition coefficients are complex numbers that can be arbitrarily close to zero or one; the probability of an outcome is the square of that coefficient and can be as small as you want regardless of the number of states.
“that complex number can be anything between zero and one... that has nothing to do with the number of states. So the probability can be as small as you want, even though there's only two states”
To change minds target the persuadable middle by treating people as fellow reasoners, understanding their views, working within their beliefs, and being friendly rather than haranguing or scolding.
- To change minds you should not target ideological extremes but the persuadable middle, and you persuade by treating people as fellow reasoners trying their best, genuinely understanding why they hold their views, working within their existing belief systems, and being friendly rather than haranguing or scolding them.
“People on either side of any highly polarized debate are not the persuade ones, but there's always huge numbers of people in the middle. Those are the ones you should be thinking about reaching”
Black holes are a legitimate dark matter candidate with the advantage of existing, but less popular than WIMPs or axions because no natural mechanism produces them in the right abundance.
- Black holes are a legitimate dark matter candidate and have the advantage of being known to exist, but they are far less popular than WIMPs or axions because there is no natural mechanism to produce them in the right abundance; WIMPs naturally yield the observed density (the 'WIMP miracle') while making enough primordial black holes requires baroque, unnatural fine-tuning of small-scale density fluctuations.
“the good thing about black holes is they're known to exist... The bad news is we don't know a way of making them in the right abundance to be dark matter”
Neither Fukuyama's convergence nor Huntington's perpetual clash is correct because human systems are coupled to rapidly changing technology and environment, fundamentally out-of-equilibrium.
- Neither Fukuyama's end-of-history convergence nor Huntington's perpetual clash of civilizations is correct, because human social and political systems are coupled to rapidly changing technology and environment and are fundamentally out-of-equilibrium processes; expecting culture to settle into a permanent equilibrium is as mistaken as expecting it to freeze into current conflict patterns.
“Human culture is also an out-of-equilibrium process. It is not something you should expect to settle into an equilibrium”
A functioning dictatorship implements measures faster than democracy, seemingly advantageous early in crises, but speed does not guarantee right direction; democracy remains better despite imperfection.
- A functioning dictatorship can implement measures faster than a democracy, which seemed advantageous early in the pandemic, but speed does not guarantee the right direction; democracy is actually a good problem-solving mechanism, as evidenced by China implementing overly draconian pandemic policies that made things worse, so democracy remains better than the alternatives despite being imperfect.
“a dictatorship can move much more quickly. But the thing is, it doesn't always move in the right direction”
A radioactive decay produces a spherical wave function, but observations show straight cloud chamber tracks because decoherence and branching localize the wave function after particle interaction.
- A radioactive decay produces a roughly spherical wave function, but we observe a straight line in a cloud chamber because of decoherence and branching: as soon as the spherical wave interacts with one particle in the chamber, the universe splits, and in the branch where it interacted there, the wave function is now a localized packet that continues in a straight line, interacting with subsequent particles to form a track—so waves look like particles once observed.
“as soon as it interacts with one particle in the cloud chamber, then the universe splits... and in the branch where it interacted with that particle... the wave function of the radiated particle is no longer spherical. Now it is localized near the particle that it interacted with”
Human societies cannot be predicted deterministically unlike Asimov's psychohistory because individual people are complex systems whose averaging does not yield simple macroscopic behavior.
- Unlike Asimov's psychohistory which assumes individual idiosyncrasies average out, human societies cannot be predicted deterministically because the individual people are themselves complex systems, so averaging over them does not yield simple macroscopic behavior the way averaging over atoms yields predictable gas dynamics.
“when your collective phenomenon comes from a bunch of people who are themselves pretty darn complex, then that averaging just doesn't work”
The tolerance paradox is better addressed by responding to intolerant speech issues rather than preventing it, producing robust anti-racism through understanding; Carroll accepts limits on speech directly leading to violence.
- Popper's tolerance paradox frames the issue wrongly; the better response to intolerant speech (racist, sexist, authoritarian) is generally not to prevent it but to address the issues it raises, because a more robust anti-racism comes from understanding why people hold such views, which you achieve by listening. Carroll is not a free-speech absolutist and accepts limits on speech that directly leads to violence.
“the way to deal with that, if you want to fight against racism and sexism and fascism, etcetera, is not to prevent that kind of speech, it is to address the issues raised by that kind of speech”
The rubber sheet analogy for spacetime curvature is acceptable because misleading aspects can be flagged, but the rubber balloon analogy for expansion is bad because its misleading aspects are hard to undo.
- The rubber sheet analogy for spacetime curvature is acceptable because its misleading aspect (gravity pulling the ball down) can be clearly flagged and discarded, but the rubber balloon analogy for cosmic expansion is bad because its misleading aspects—dots growing with the balloon and the balloon expanding into surrounding space—are very hard to undo, falsely suggesting galaxies grow and the universe expands into something.
“There are other analogies which I don't like, because it is very difficult to let go of the misleading parts of the analogy. So I'm not a big fan of the rubber balloon analogy”
Society poorly supports people with serious mental illness partly by treating them as either fully responsible or requiring care; as a wealthy society we could guarantee everyone a home but choose not to.
- Society does a poor job for people with serious mental illnesses partly because we treat people as either fully responsible self-agents or as requiring care, and these fuzzy-but-useful categories leave many ill people unsupported; as a wealthy and powerful society we could give everyone a home and not allow homelessness, but we are making a choice to let the issue continue because we do not care enough.
“as a society, we're certainly wealthy and powerful enough to give everyone a home, to not let there be homeless people. And we're making a choice as a society to let this issue continue, 'cause we don't care that much”
Entanglement happens constantly everywhere in nature, not just in laboratories; particles like helium ground-state electrons are highly entangled, and only separating and measuring them breaks entanglement.
- Entanglement happens all the time everywhere in nature, not just in laboratories with lasers; for example, the two electrons in the ground state of a helium atom are highly entangled in a spin singlet state, and observing one would break the entanglement—it is only the act of separating and measuring particles that breaks entanglement, which is why we do not notice it constantly.
“consider a helium atom that has two electrons... those two electrons are very entangled with each other. Their spins are entangled with each other”
Two valid explanations exist for why objects bend spacetime: either because the laws of physics say so, or via the action principle where actual configurations minimize total action.
- There are two valid ways to answer why objects bend spacetime: either simply because that is what the laws of physics say (we could imagine worlds where they don't), or via the action principle, where the total action has contributions from both matter and spacetime curvature and the actual configuration minimizes this action by having spacetime curve in response to energy and momentum—though this still rests on assuming the action principle holds.
“there's a contribution to the action from matter... And there's another contribution from the curvature of space-time. And you wanna minimize the overall amount of action. And the way to do that is to have space-time curve a little bit in response to the energy and momentum inside”
Climate science is unrepresentative of how science works because political pressures convince people of dopey claims and lead scientists to suppress valid methodological criticisms fearing denialism.
- Climate science is an unrepresentative example of how science works because the surrounding political pressures both convince many people of dopey anti-science claims and, because scientists are human, lead climate scientists to want to stamp out methodological criticisms (even ones with a valid point about details) for fear of opening the door to denialism—paralleling how evolutionary biologists once suppressed critiques to avoid aiding creationism.
“those political pressures both convince a whole bunch of people of really dopey things... But they also, because scientists themselves are human, they introduce non-scientific considerations into how scientists talk about it”
The claim that science and math are biased or racist because white men devised them is held by very few and is false, but the valid related point is they have been used serving biased structures.
- The claim that science and math are inherently biased or racist because they were devised by white men is held by very few influential people, and it is not even true that they were devised solely by white men; however, the related and valid point is that science and math, while objective themselves, have been used in service of biased or racist social structures, which is something worth thinking about.
“science and math, even if they are perfectly objective, have been used in the cause of biased racist points view or social structures... That's not to say that the science or the math are biased or racist, they're just science and math, but they can be put to racist or biased purposes”
Despite defending views long-term, scientists change minds rapidly once obviously valid views backed by experiment emerge; persistent disagreements usually reflect reasonable theoretical trade-off weightings.
- Contrary to the cartel critique of science, one of the great things about science is that even though scientists defend their views for a long time, they change their minds rapidly once an obviously more valid view comes along, especially when it is backed by experiment or data; persistent disagreements (like collapse theories vs Many-Worlds) usually reflect reasonable differences in weighting unconvincing theoretical trade-offs, not refusal to listen to reason.
“once an obviously more valid view comes along, they will change their minds. That's one of the great things about science. It's easiest to do when the new view is backed up by an experiment”
Since 2007 there have been no truly surprising fundamental discoveries about dark matter or dark energy; dark energy is well-explained by the cosmological constant, while dark matter remains unidentified.
- Since 2007 there have been no truly surprising fundamental discoveries about dark matter or dark energy; dark energy is well explained by the cosmological constant so there are likely no more empirical discoveries to make about it (only a better theory of its value is needed), whereas dark matter remains unidentified with no overwhelmingly plausible candidate, though Carroll remains hopeful for direct evidence within his lifetime.
“with dark matter, we don't know what it is. We don't have an overwhelming plausible candidate like we do for dark energy”
Listener numbers do not drive Carroll's topic selection because the audience is remarkably stable; famous guests roughly double average listenership while unusual topics reduce it only 10-20%.
- Listener numbers do not drive Carroll's topic selection because the audience is remarkably stable; a very famous guest can roughly double the average listenership, while an unusual topic only reduces it by 10-20%, so the asymmetric upside of famous guests outweighs any downside from weird topics, and he deliberately blends topics to keep listeners on their toes.
“having a very famous person will get you maybe twice the average number. Whereas, the decrement that you get from the average for having a less well-known topic is maybe 10% below average or 20% at most”
At fundamental physics there are no causal arrows, only reversible patterns; causality is a useful macroscopic approximation whose arrows the laws of physics constrain to run from past to future.
- At the level of fundamental physics there are no causal arrows—only an unbreakable reversible pattern—and arrows of causality are useful approximations in the macroscopic emergent world; you can legitimately demand causal networks whose arrows go only from one moment to the next, but the laws of physics only permit such sensible networks running from past to future, not to the previous moment, which is a non-trivial statement.
“there are no arrows at the level of fundamental physics. There's just a pattern that is unbreakable and perfectly reversible from past to future... The arrows and causality only come about in the macroscopic emergent world”
The other worlds in Many-Worlds have no XYZ coordinates nor exist in other space; they exist simultaneously without location because location is an emergent concept within each world, not embedding them.
- The other worlds in Many-Worlds do not have XYZ coordinates within space, nor do they exist in some other space at all; they exist simultaneously but without location, because the concept of location in space is something that exists only within each world—space is an emergent concept inside the worlds rather than the worlds being embedded in a larger space.
“they do not have an XYZ coordinate within space, nor do they have an existence in another location at all... Space is an emergent concept inside the worlds. It's not that the worlds are embedded in some bigger space”