Entropy Increasing Doesn't Mean Things Getting Messy: Oil and Water Separating, Frost Crystals Forming Are All Spontaneous Order
The second law of thermodynamics is often told as "things get messier and messier," but oil separating from water, beach sorting, and frost flowers on a window are all spontaneous order. Carlo Rovelli says low entropy may not be in the universe, but in the way we look at the universe.
The video won't play here. Listen to the audio instead:
The argument · tap a timestamp to hear it
Oil separating from water isn't anti-entropy; it's entropy producing order
Rovelli uses a set of everyday images to dismantle the popular claim that "entropy = disorder": water and vinegar mix into a uniform blend, but water and oil separate into layers after mixing; on a beach, fine sand and pebbles sort themselves apart; frost on a window spontaneously grows regular structures. He says that under a properly defined "order," many things in nature become spontaneously ordered when left alone. This does not overturn the second law; it says popular science draws the correspondence between "order" and "entropy" too crudely. He specifically points out that the complexity of biology and the structure of the Earth's surface are often misread as "anti-entropy," when in fact entropy growth itself is constantly producing order.
— Carlo RovelliLow entropy may not be in the universe, but in how you coarse-grain
Rovelli proposes a hypothesis he has put into a paper himself but is not sure is right or wrong: low entropy means "special," but what is special is not the state of the universe, but coarse graining itself. Coarse graining is a perspective — looking only at some variables, not all the details. If this hypothesis holds, the universe can be in a completely arbitrary state, but from our perspective, which sees only some variables, it appears low-entropy in the direction we call "the past." This turns the "initial low-entropy boundary condition" from a fact about the universe into a question about how observers partition information.
— Carlo RovelliYou can be in two places at once, just not to yourself
Rovelli pushes quantum superposition onto people: if quantum mechanics is universal, Brian can be in a superposition of Los Angeles and San Francisco, and to Carlo this interference can be measured. But to Brian himself, quantum mechanics never tells you that you are in a superposition — you look around and see either San Diego or Los Angeles. So "where you are" is two different things for you and for me. He compares this to Galileo discovering that velocity is relative: saying "velocity is zero" actually omits "relative to the Earth." What quantum mechanics says is that all properties must be understood relative to something.
— Carlo RovelliGalileo used the principle of relativity and got the tides wrong
Rovelli uses Galileo's tidal model to show that "having an idea" and "pushing the idea to the end" are two different things. Galileo thought the combination of the Earth's revolution around the Sun plus its rotation stirred the seas and produced the tides. But Rovelli points out that revolution is, to first order, uniform straight-line motion, and rotation alone cannot produce an effect either, so the tides cannot come from this — what actually explains the tides is Newton. Galileo himself precisely violated the principle of relativity he had discovered. Rovelli says Einstein was the same: after writing general relativity in 1915, he later wrote a paper denying gravitational waves (and corrected it after a referee sent it back), and also insisted that the Schwarzschild singularity was the end of the world, an error he never corrected until his death.
— Carlo RovelliWithout dissipation you can't record anything
Rovelli says that already in classical mechanics he found a fact that made him "see the light": if you want to model a measuring device with classical mechanics — a barometer, say, or a device that records whether a ball is on the left or the right — you can't do it unless you introduce dissipation. A ball dropped into a box will bounce back up; it must dissipate in order to stop and leave a record. So measurement is essentially a statistical process requiring dissipation, and this was already true before quantum mechanics; it is not peculiar to quantum mechanics. This directly answers the puzzle of "unitary evolution is reversible, so where does irreversibility come from": recording itself requires that something be thrown away.
— Carlo RovelliA chess program deliberates, and so do you
Rovelli uses his chess computer to take apart free will: the computer is deterministic, and given its internal state and the opponent's move, its output is necessarily predetermined. But why does it still spend time enumerating all possible moves and evaluating them one by one? Because the only way to reach that result is to go through this evaluation process, and that process is called deliberation. He says it would be foolish to say it is not deliberating, because deliberation is exactly this. Brian is the same: if he did not go through this complex process, he would not make the decision to start a project or choose a career. So-called freedom is not violating the laws of nature; it is precisely the fact of "having gone through deliberation" itself.
— Carlo RovelliIn principle yes, not now, and within two years he doubts it
Asked whether machines could have Einstein's "happiest thought" (the equivalence principle in free fall), Rovelli answers in three layers: in principle yes, he does not think humans are so special as to be unreproducible; not now; and within two years he is skeptical, but has no hard evidence that it is impossible. He mentions that last week he had Claude do something his PhD student had not finished in a month, and Claude finished it in half an hour, so the capability is beyond doubt. But he says Claude did not invent the equivalence principle, and doing that within two years he finds hard. At the same time he states plainly that there is a lot of hype in AI.
— Carlo RovelliTen years and unlimited funding, he'd build a roomful of Josephson junctions
Asked what he would do with "ten years and unlimited funding," Rovelli's answer is not a telescope or a collider, but a large room filled with Josephson junctions, plus an engineering team. The reason: loop quantum gravity predicts particles of Planck mass, which could be dark matter candidates, with a mass roughly equal to a small piece of hair or a flea egg. He says we already know the direction, mass, and interaction strength of dark matter clouds; although it is only an extremely weak Newtonian gravity, Josephson junctions could catch it with simple physics. He thinks it is technically feasible within ten years, and what is missing is only money.
— Carlo RovelliIn their own words · checked verbatim
It's not true that under any definition of order or disorder, things get disordered. In fact, very often in nature, things left alone get ordered.
Carlo Rovelli2:03
Low entropy means special. But what is special is not the state of the universe. It's the coarse graining.
Carlo Rovelli7:12
You Brian can be in a superposition. You can be in LA and also in San Francisco.
Carlo Rovelli12:16
So there's more time up here and less time down here. Your head, it's older than your feet, right?
Carlo Rovelli26:29
you cannot record anything without some dissipation so measurement is an intrinsic statistical process
Carlo Rovelli31:06
To imagine that to be free, you have to violate the laws of nature is stupid. I mean, we're free even without violating the laws of nature.
Carlo Rovelli38:53
the planck mass is almost microscopic thing is the mass of my a little piece of my hair
Carlo Rovelli48:15
Figures
| Physical size corresponding to the Planck mass | Roughly equal to the mass of a small piece of hair or a flea egg | 48:15 |
| Number of candidate explanations for dark matter | Five to six | 48:15 |
| Top-voted option in the quantum gravity survey | "No opinion" got the most votes, followed by string theory, then loop quantum gravity, and finally gravity is not quantized | 50:21 |
| Time Claude took to finish a task a PhD student had not completed in a month | Half an hour | 40:56 |
Glossary
- coarse graining
- A way of observing that looks only at some variables and ignores all the details; the definition of entropy depends on it.
- loop quantum gravity
- A quantum gravity theory co-founded by Rovelli, which holds that space itself is discrete.
- relational quantum mechanics
- Rovelli's interpretation: the quantum state is relative to an observer, not absolute.
- Josephson junction
- A superconducting device that Rovelli thinks could be used to detect the weak gravity of Planck-mass particles.
- equivalence principle
- In free fall you feel no gravity; Einstein called it his happiest thought.
How to listen
Engineers and researchers interested in time, entropy, and the foundations of quantum mechanics; anyone who wants to hear a physicist take apart the popular claim that "entropy increase = things getting messy."
The Abacus AI ad segment from 34:48 to 36:50 can be skipped.