What this covers

Sean Carroll and Kate Jeffery conduct a discussion—rather than a conventional interview—on how complexity emerges across cosmological and biological timescales. Carroll, a theoretical physicist, argues that complexity arises not in spite of the second law of thermodynamics but because of it, emerging in a narrow band between low-entropy simplicity and high-entropy equilibrium. Jeffery, a behavioral neuroscientist, extends this framework into biology, tracing how life builds complexity through discrete evolutionary transitions—the origin of multicellularity, the emergence of nervous systems—each of which expands an organism's reach across space and time. The two explore the deep structure underlying these increases in organized structure: why the early universe began in a steeply improbable low-entropy state, how free energy available above thermal equilibrium fuels all organized activity, and why this window of possibility is temporary.

The conversation ranges across how biological evolution escalates not merely in structural elaboration but in information processing, from RNA to brains to language. Jeffery introduces a darker implication: the very innovations that extend biological reach—photosynthesis, burrowing, the colonization of land—have repeatedly triggered mass extinctions by reshaping planetary chemistry too rapidly. Carroll addresses technical puzzles around entropy and cosmology: why the early universe appeared smooth despite gravity, how gravitational entropy differs from entropy in simpler systems, and the eventual foreclosure of all complexity when the universe reaches heat death. The episode also clarifies conceptual confusions—how physicists and information theorists define entropy in opposite ways, why the second law of thermodynamics weakens as particle count drops, and how coarse-graining connects entropy to measurable descriptions of the world.

Sharpest takeaway

Carroll argues that complexity is not in tension with the second law of thermodynamics but is generated by entropy increase itself, emerging only in the intermediate regime between low-entropy simplicity and high-entropy equilibrium, while Jeffery adds that biological complexity grows in fragile step-transitions that extend organisms' reach across space and time, making the same mechanisms that build complexity capable of catastrophic collapse.

  • Complexity is low at very low entropy and very high entropy, peaking only in between, and arises because of (not despite) entropy increase given long-range correlations.
  • Biological evolution proceeds through discrete transitions (multicellularity, nervous systems) that extend correlation over space and time, but these same advances drive mass extinctions.
  • The universe is in a temporary low-entropy window where complex life is possible; eventual heat death forecloses complexity permanently.

The argument · threads13 threads · 35 claims
0.78

Entropy is a measure of microscopic arrangements consistent with observed macroscopic states, and systems naturally tend toward higher entropy by the second law.

3 pointscentrality 5/5
  • Entropy, following Boltzmann, is a measure of how many microscopic arrangements of a system's constituents are consistent with its observed macroscopic appearance; because there are more ways to be high-entropy than low-entropy, natural undirected evolution tends to move systems toward higher entropy, which is the second law of thermodynamics.

    “Entropy is basically a way of counting how many ways there are to arrange the microscopic constituents to give you that macroscopic look.”

  • In edge cases with few moving parts, such as a DNA molecule versus another long polymer, what counts as the relevant microscopic constituents and how they are coarse-grained into macrostates can legitimately depend on functionality; a careful physicist should use whichever coarse-graining suits the context and remain consistent, rather than insisting that physically similar molecules have identical entropy.

    “yeah, what exactly you count as the microscopic constituents and how you course grain them into macro-states can matter, can be different. And it might depend on functionality”

  • Entropy is subjective only in the same sense as any emergent description: a perfect Laplace's-demon observer wouldn't need coarse-graining at all, but the macroscopic description is real and useful because only certain ways of coarse-graining (those tracking local, robust patterns) yield simple predictable dynamics, while arbitrary coarse-grainings produce no usable laws.

    “The nice thing about emergence is you throw away an enormous amount of information, the locations and velocities of every individual molecule, but what you keep is still somehow representing a true pattern in the dynamics.”

0.78

The early universe was anomalously low-entropy despite being small, smooth, and subject to gravity, making our existence and cosmic evolution possible.

3 pointscentrality 5/5
  • When matter is very densely packed and mutual gravity between particles becomes important, a high-entropy configuration is no longer smooth but lumpy: a single black hole at our galaxy's center has more entropy than all the particles in the known universe, so the early universe being smooth was itself a low-entropy fact, compounding its low entropy from being packed into a tiny volume.

    “when gravity, the gravitational pull of the particles on each other was really important... When matter is very densely packed, high entropy doesn't mean smooth anymore... you increase entropy by making things lumpier.”

  • Many professional cosmologists wrongly believe the early universe had high entropy because it was smooth, hot, and glowing like a black body in apparent thermal equilibrium; this is entirely wrong because it was both a very small box of gas and smooth despite gravity being important, making it an anomalously low-entropy state by any measure.

    “when gravity, the gravitational pull of the particles on each other was really important... When matter is very densely packed, high entropy doesn't mean smooth anymore... you increase entropy by making things lumpier.”

  • Unlike classical statistical mechanics where gas is confined to a fixed box, in general relativity space itself can always expand, so one cannot fix the universe's size when asking how much entropy is possible; the early universe's entropy was genuinely low and is much larger now, but it is not that the maximum allowed entropy increased—rather the early universe was never close to accessing the entropy allowed even at that time.

    “if your box is the universe you have to allow the box to expand. That's just part of the dynamics.”

0.78

Complex structures emerge only in intermediate-entropy regimes and require specific mechanisms like gravity and electromagnetism creating correlations across scales.

3 pointscentrality 5/5
  • Complexity is low both when entropy is very low (everything organized and simple) and when entropy is very high (everything uniformly mixed and simple); complex structures can only exist in the intermediate regime, so as entropy increases from low to high, complexity starts low, grows, and then fades away again.

    “as entropy goes up complexity starts low grows, and then fades away again.”

  • What matters for complex structure formation is not gravity per se but the existence of different forces acting in different directions, allowing competition and equilibrium between push and pull; having gravity and electromagnetism both acting over long ranges seems to be a minimal requirement for interesting complex structures to form.

    “what is important is not gravity per se, but the fact that there are different forces that work in different directions, so there can be a competition between them.”

  • Whether complex structures actually form as entropy increases is not guaranteed; in simple simulations complexity arises only when there are correlations in how components mix, meaning it cannot be that each element interacts only with its immediate neighbor—there must be rigidity or long-range forces pushing things together in concert so that structure forms across many scales.

    “in cases where you wanted to get complex structures forming that was allowed as long as there were correlations in how the cream and coffee mixed into each other.”

0.71

The accelerating universe will eventually reach heat death equilibrium where no complexity or consciousness can exist, ending all possibility for life.

4 pointscentrality 4/5
  • Because the accelerating universe has a finite amount of stuff within our horizon but an infinite time to evolve, it will reach maximum-entropy equilibrium: longest-lived stars burn for trillions of years then exhaust their fuel, matter falls into black holes, and black holes evaporate over roughly 10^100 years (Hawking, 1970s), leaving empty space forever with no room for complexity—the heat death of the universe.

    “there's a finite amount of stuff in the universe, and there's an infinite amount of time for it to evolve. So like the cream in the coffee, it will reach equilibrium. It will reach a point where everything is smoothed out into its highest entropy configuration.”

  • Energy is conserved but entropy is not—entropy goes up—so the real constraint on life is free energy: the gap between the current state and thermal equilibrium. As the universe approaches maximum entropy this free energy goes to zero, leaving less and less room for anything complex to happen; this gap, not space per se, is the ultimate fuel crisis that ends life.

    “Energy is conserved, but entropy is not. Entropy goes up. And so, the difference between where we are and thermal equilibrium just decreases with time.”

  • Even in theory an advanced civilization cannot escape the heat death: while one could conceivably build an artifact recording its deeds that lasts arbitrarily long, no one could ever read it, because the act of reading and remembering itself increases entropy, which is impossible at maximum entropy—so consciousness, thought, and life cannot exist once equilibrium is reached.

    “the act of reading and remembering increases the entropy of the universe, and if you're at maximum entropy, that can't happen. So, consciousness itself, thought, and life cannot exist once you've reached equilibrium.”

  • The Poincaré recurrence theorem says that in a universe with a truly finite number of possible states you would eventually fluctuate back into a low-entropy state, but this seems clearly not to describe our universe, causes problems like Boltzmann brains, and would not help anyway because no messages could be sent across the point of equilibrium from one part of time to another.

    “There is something called the recurrence theorem that if you wait long enough... you will fluctuate back into a low entropy state. But that causes all sorts of problems with things like Boltzmann brains”

0.64

Life and complexity exploit the universe's high initial information content as entropy increases, manipulating pre-existing order rather than creating new information.

3 pointscentrality 4/5
  • The formation of complex structures like the biosphere and organisms is not merely despite entropy increasing but because of it: if entropy weren't increasing we would already be in high-entropy equilibrium with no living structures, so our existence as organized low-entropy creatures is part of the universe's general tendency to increase entropy overall.

    “the formation of complex structures like the biosphere and individual organisms here on earth is not just despite the fact that entropy's increasing, it's because of entropy increasing.”

  • The evolution of life can be seen as an escalating sophistication in handling information, not just structural complexity: early RNA/DNA carried information, eukaryotic life differentiated it into mitochondrial and nuclear DNA, multicellularity added further differentiation, brains enabled thought and self-awareness, and language enabled symbolic thinking—each a leap forward in information processing.

    “one way of thinking about the evolution of life is not only an evolution of complexity in terms of the structures that make the physical organism but an evolution of the sophistication with which we deal with information.”

  • Because the universe began in a very low-entropy state, it started with a great deal of information built in; life, thought, and books do not increase the universe's information content but degrade it as entropy increases—what we do is manipulate and take advantage of that pre-existing information in increasingly interesting ways, so the universe goes from high-information-low-interest to lower-information-higher-interest.

    “We haven't been increasing the information content of the universe by having life or thoughts or books, we've been degrading it, because entropy is increasing, but we've been using, manipulating, taking advantage of the fact that... information is all around us in more and more interesting ways.”

0.36

Multicellularity and other phase transitions enlarged possibility space, and our universe might be one of many created in low-entropy states allowing renewed complexity.

3 pointscentrality 2/5
  • Complexity can build upon itself: when there is no life, or only unicellular life, the space of possibilities is limited, but a phase transition like the origin of multicellularity enormously enlarges that possibility space, and 500 million years later we are still nowhere near done exploring it, with technology opening up entirely new dimensions of possibility.

    “complexity can build upon itself. This is certainly the lesson both of the origin of life and of the origin of multi-cellularity that there's only so many things that can happen if there's no life around or if all the life is unicellular and the space of possibilities becomes enormously bigger as this phase transition happens.”

  • While no information can pass from our universe to escape the heat death, it is possible that a baby universe will be born from ours, or that our universe is part of a larger multiverse continually creating new universes; the way such universes are created naturally starts them in low-entropy states from which complexity can grow, and Carroll has proposed a theory in which our own universe arose this way.

    “it's possible that what we think of as the universe is just part of a bigger multi-verse and more universes are constantly created, and the way that those universes are created naturally starts them in a low entropy state from which they can grow”

  • The journey from the low-entropy early universe to late-universe equilibrium is one of exploring the vast space of possibilities; in equilibrium every configuration is roughly equally probable, but because the set of configurations is so enormous, even in the 10^100 years to equilibrate the universe will never come close to actually exploring every arrangement—so the specific way that exploration unfolds is the crucial question for life and complexity.

    “The space of possibilities is really large... the journey from the early universe where entropy was low to the very, very late universe where we're in equilibrium is one of exploring that space.”

0.35

Creationist objections to evolution violate no physical law because Earth is open to incoming low-entropy solar energy that balances local entropy decrease.

1 pointcentrality 3/5
  • The creationist argument that complex biospheres violate the second law is wrong because the second law only requires entropy to increase in closed systems, and Earth is nowhere near closed: it receives low-entropy energy from the sun and radiates higher-entropy energy back, so Earth's net effect is to increase the entropy of the universe, fully consistent with the second law.

    “the second law only says that entropy increases in closed systems, and the earth and its biosphere are nowhere near a closed system. We're getting wonderfully low entropy energy from the sun”

0.29

Entropy behaves as an approximation whose reliability depends on particle count, making fluctuations critical at cellular scales but irrelevant for engines.

2 pointscentrality 2/5
  • The second law of thermodynamics is not a fixed law but an approximation whose quality scales with particle number: with three moving parts there is no second law, with about 100 it is a reasonable approximation, and with 10^10 it is extremely good; consequently fluctuations and unlikely events (like an ice cube spontaneously forming) are hugely important at the cellular-biology scale and completely irrelevant for an internal combustion engine.

    “if you have three moving parts, there's no such thing as the Second Law of Thermodynamics... If you have 100, then there's a reasonable approximation to it, and if you have 10 to the 10, then it's really, really, really good.”

  • Physics strongly prefers Markovian descriptions in which the state at one moment depends only on the immediately previous moment and not on the entire past history; if the history over time had to be tracked, the physical description would become enormously more complicated, though it is not impossible.

    “we very much like in physics to think in Markovian terms. Markovian just in the simple sense that what happens at one moment depends on what happened in the immediate previous moment, but not the entire history of the past.”

0.29

Complexity's fragility and life's dependence on specific biological channels means catastrophe is possible and human caution about environmental damage is warranted.

2 pointscentrality 2/5
  • Unlike the featureless cream-and-coffee simulation where complexity smoothly rose and fell, life's complexity is built through very specific, almost unique biological channels for order to be created and to build upon itself, and the fragility of these specific mechanisms is exactly what opens the possibility for catastrophe—the same effectiveness that builds complexity also enables it to be wiped out.

    “in the specific case of life, there's sort of mechanisms for this complexity that are very specific, that there's certain biological channels for order to be created, for complexity to be created. And in some sense they're fragile.”

  • Because there is no inevitability to the increase in complexity and there is a definite empirical record of catastrophes, one should not be sanguine that life is resilient and everything will be okay when seeing carbon dumped into the atmosphere and sea levels rising—we could hurt the biosphere very badly, and the priority should be increasing the probability of a good outcome rather than merely estimating it.

    “when you see a stumping carbon into the atmosphere and the sea levels rising and stuff like that, don't be sanguine that, 'oh yeah, life is resilient and everything will be okay.' We could hurt it really badly.”

0.15

We lack sufficient data to claim commonness or rarity of life and should remain open-minded about alternative biochemistries beyond carbon-based forms.

2 pointscentrality 1/5
  • We should be open-minded about how common life is because we have essentially no data on the fraction of planets that bear life; Earth might be the only planet in the observable universe with life, or alternatively, since life arose quickly on Earth but multicellular and eukaryotic life took a very long time, those later transitions may be the hard parts, leaving other planets covered only in single-celled organisms with no detectable signals.

    “we don't have a lot of data about the fraction of other planets or other environments that have life on them, so we should be very open-minded, maybe within our observable universe the earth is the only planet that ever had life form on it.”

  • We may be unjustifiably chauvinistic in assuming life requires carbon or three-dimensional space; just as we tend to conclude life can only happen in three dimensions because we cannot see how it would work otherwise, we should be skeptical of declaring carbon necessary—if carbon's chemistry were different, life might have formed in a very different but viable way.

    “are we just so Chauvinistic about that? If the chemistry of carbon were completely different, I'm open to the possibility that life could have formed in some very bizarre to us but very different way.”

0.07

Accelerating cosmic expansion reveals repulsive vacuum energy or cosmological constant, understood within general relativity as a gravitational effect, not a new force.

1 pointcentrality 1/5
  • The accelerating expansion of the universe is still gravity, not a new force: Einstein's general relativity (1915) identifies gravity with the curvature of spacetime, of which the changing scale factor (galaxies moving apart as space grows) is one feature; the 1998 discovery of acceleration revealed a vacuum energy or cosmological constant—a new kind of stuff whose gravitational effect is repulsive—rather than a new force.

    “Einstein back in 1915 explained that what gravity is, is the curvature of spacetime... The scale factor as we call it in general relativity that tells us the relative distance between distant galaxies... Space is growing and that is a phenomenon due ultimately to general relativity which is the theory of gravity.”

0.07

Physicists and information theorists reverse the entropy-information relationship; for cosmic evolution the physicist's definition linking low entropy to high information matters.

1 pointcentrality 1/5
  • Information theorists and physicists define the entropy-information relationship in opposite ways: Shannon's communication theory links high entropy with high information content because an unpredictable (high-entropy) code conveys more new information per letter, whereas physicists say a low-entropy state carries more information because so few states look that way; for cosmic evolution, the physicist's definition is the germane one.

    “information theorists define the relationship between entropy and information in exactly the opposite way to how physicists define it because information theory grew out of Claude Shannon talking about how to send signals over wires”

0.07

Space is physically fundamental because locality of interaction—objects affect each other when proximate, not merely when sharing velocity—applies to biology and physics.

1 pointcentrality 1/5
  • Space is physically special because it is the arena in which interactions between physical systems are local: two objects interact when they are at the same or nearly the same location in space, in a way that they do not if they merely share a velocity but are far apart, and this locality of interaction should hold for biology as much as physics.

    “space is the arena in which interactions between physical systems look local”