The best-understood skeleton
Heat water and it boils. Cool a magnet and it magnetizes. Push an ecosystem and it collapses. In each case a system sits in one phase, a single knob is turned, and at one special value the whole system reorganizes at once. That special value is the critical pointThe exact value of a control parameter (temperature, pressure, density…) at which a system changes phase. At this point fluctuations occur on every length scale at once..
Of the four skeletons, criticality is the anchor — the most thoroughly developed. Physics has spent a century building its mathematics: phase transitionsA qualitative change of state at a threshold — solid to liquid, disordered to ordered, stable to collapsed. The transition is sharp: a tiny change in the control parameter flips the whole system., scale invarianceAt the critical point a system looks statistically the same at every zoom level. There is no characteristic size — patterns repeat across scales., universalityThe surprising fact that utterly different systems — a magnet, a fluid, an alloy — behave identically near their critical points, sharing the same exponents. The microscopic details wash out., and the renormalization groupA mathematical method (Wilson, 1971) for tracking how a system's description changes as you zoom out. It explains universality and earned the 1982 Nobel Prize in Physics. that ties them together. This page is the reference other skeletons lean on.
The order parameter
Every phase transition has an order parameterA quantity that is zero in the disordered phase and nonzero in the ordered phase — for example the net magnetization of a magnet. It measures ‘how much order’ the system has.: a number that is zero on one side of the transition and grows continuously from zero on the other side. For a magnet it is the net magnetization. For a boiling liquid it is the density difference between liquid and gas. For a neural network it is the size of the largest correlated cluster.
Below the critical point the order parameter grows as a power law of the distance to the transition: m ∼ (Tc − T)β. The exponent βThe order-parameter exponent. It controls how quickly order grows as you move past the critical point. Different universality classes have different values of β — e.g. β ≈ 0.33 for the 3D Ising class. does not depend on the substance — it depends only on the system's universality class. That is the first hint that criticality is skeleton-level mathematics, not a fact about any one material.
Cross the critical point
Drag the control parameter through the transition. Far below, the system is ordered — nearly all cells point the same way. Far above, it is disordered — random static. At the critical point, domains of every size coexist: the pattern looks the same whether you zoom in or out.
Schematic, not a physical simulation — it illustrates how domain size (the correlation length ξThe typical size of a correlated domain — a patch of the system that acts as a unit. Away from the critical point it is small; at the critical point it diverges to the size of the whole system.) and net order (m) behave as the control parameter crosses Tc.
The correlation length diverges
The single quantity that makes a critical point special is the correlation length ξ — the distance over which one part of the system ‘feels’ another. Away from the transition, ξ is small: a disturbance here has no effect over there. As you approach the critical point, ξ grows — and at the critical point it diverges to infinity.
When ξ is infinite, the system has no characteristic length scale. Fluctuations of every size occur together, which is exactly what scale invarianceNo preferred size. A zoomed-in patch is statistically indistinguishable from the whole. Mathematically, correlations decay as a power of distance rather than exponentially. means. It is also why critical systems produce power lawsRelationships of the form (quantity) ∝ (size)−exponent. Small events common, large events rare, and no typical scale in between — the fingerprint of criticality. everywhere: with no characteristic scale, the only self-consistent statistics are scale-free.
This is the skeleton's tracer. Where topology leaves the fingerprint φ, criticality leaves power-law exponents — β, ν, γ — measurable in the tails of distributions and in how quantities scale near the transition.
Critical slowing down — the early warning
As a system nears a critical transition, it takes longer and longer to recover from small disturbances. Perturb it and the wobble decays more slowly; the system becomes sluggish and ‘sticky.’ This is critical slowing downNear a tipping point the recovery rate from perturbations drops toward zero. Measurable signs: rising autocorrelation ('the system remembers longer') and rising variance ('fluctuations grow')., and it is one of the most practically important facts about criticality.
Because slowing down leaves statistical footprints — rising variance and rising autocorrelationHow strongly the system's current state predicts its next state. It climbs toward 1 as a tipping point approaches, because the system relaxes ever more slowly. — you can sometimes forecast a tipping point before it happens. Scheffer and colleagues (2009)3 showed these early-warning signalsGeneric statistical indicators (variance, autocorrelation, skewness) that rise ahead of many critical transitions, from ecosystems and climate to physiology and finance. appear across ecosystems, climate, physiology, and finance — a convergence in its own right.
Early-warning signals are exactly where criticality touches applied work on tipping points: the same mathematics that describes a magnet losing its magnetization describes an ecosystem approaching collapse.
Universality and the renormalization group
Here is the deepest fact. Measure β for a magnet, for a boiling fluid, and for a binary alloy separating into two phases. You get the same number. The three systems could not be more different microscopically, yet near their critical points they are mathematically identical. They belong to the same universality class.
| Universality class | β | Example systems |
|---|---|---|
| Mean-field | 0.5 | Superconductors, high dimensions |
| 2D Ising | 0.125 | Thin magnetic films, adsorbed layers |
| 3D Ising | ~0.33 | Uniaxial magnets, liquid–gas, binary alloys |
| 3D XY | ~0.35 | Superfluid helium, planar magnets |
Why does this happen? Kadanoff (1966) and then Wilson (1971)2 answered it with the renormalization group: systematically zoom out, averaging over small-scale details, and watch how the system's description changes. At the critical point the description stops changing — it is a fixed point of the zooming-out operation. Everything that differs between a magnet and a fluid is a small-scale detail that washes out under repeated zooming; only the fixed point’s exponents survive. Universality classes are the fixed points. This insight earned Wilson the 1982 Nobel Prize in Physics.
This is what makes criticality a mathematical skeleton rather than a collection of coincidences: the renormalization group proves that unrelated systems share one underlying object.
Two depths of criticality
Like every skeleton, criticality produces two emergent properties: a structural one the system has, and a functional one the system does.
Depth 1: Spontaneity (the system has it)
Order appears by itself. Nobody tells the magnet which way to point; at the critical temperature it spontaneously breaks symmetryThe laws are symmetric (up and down are equivalent) but the outcome is not — the system picks one direction on its own. The hallmark of a continuous phase transition. and chooses a direction. This is the Major property of Skeleton 2: at a threshold, structure emerges spontaneously, with no external designer and no gradual tuning.
Depth 2: Coherent transport (the system does it)
Once a critical medium is organized, it can carry signals across itself without falling apart. SolitonsSelf-reinforcing waves that travel long distances without spreading out or losing shape — because nonlinearity balances dispersion. Nerve impulses and some plasma structures behave this way., excitation waves, and plasma structures move as coherent units, transporting energy or information over long distances. This is the minor property: it accumulates along a path, in contrast to the binary, all-at-once character of spontaneity.
A spiral wave crossing heart tissue, a nerve impulse travelling down an axon, a plasmoid ejected from the magnetosphere — all are coherent transport riding on a critically poised medium.
Where criticality meets the other skeletons
Because criticality is the anchor, it forms the strongest bridges to the other three skeletons.
Criticality ↔ Computation: the edge of chaos
Computation is maximized precisely at the critical point — ordered enough to store information, disordered enough to transmit it. This is the B–D bridgeBridge letters: A = Topology, B = Criticality, C = Autocatalysis / Closure, D = Computation. B–D links Criticality and Computation. (Criticality ↔ Computation), the edge of chaosThe boundary between frozen order and chaotic noise. Langton (1990) showed computational capability peaks there — and it coincides with the critical point. studied by Skeleton 4.
Criticality ↔ Closure: self-organized criticality
Some systems tune themselves to the critical point without any external knob — a sandpile always relaxing to its angle of repose. This self-organized criticalityBak, Tang & Wiesenfeld (1987): systems that naturally evolve to a critical state and stay there, producing power-law avalanches without fine-tuning. is the B–C bridgeBridge letters: A = Topology, B = Criticality, C = Autocatalysis / Closure, D = Computation. B–C links Criticality and Closure — a system whose own dynamics maintain it at the critical point. (Criticality ↔ Closure): the network's self-sustaining dynamics hold it at criticality.
Criticality ↔ Topology: broken symmetry
This is where broken symmetryA system whose governing law is symmetric under some transformation, but whose state is not. The law does not pick a direction; the system does. Anderson, More is Different, Science 177:393 (1972). lives. Ordering is a symmetry breaking, so it belongs to this skeleton — and it does not stop there: the order parameter left behind has a shape, and the defectsPlaces where the order parameter cannot be defined — a spiral core, a disclination, a vortex. Mermin (1979) showed they are classified by the homotopy groups of the order-parameter manifold. in that shape are what Skeleton 1 classifies. A survey of about forty documented symmetry breakings across these systems found every one of them here or on this edge, and none on closure or computation. On that reading the S1×S2 edge is not a coincidence of two structures co-occurring; it is one phenomenon seen from two sides. Argued in full, with the sweep and the caveats, on Broken Symmetry and the S1×S2 Edge.
The Skeleton Matrix shows criticality (S2) as the axis nearly every physical system lights up — which is why it earns the name anchor.
A century, twelve fields, one mathematics
Criticality is the convergence story documented in the FACETSA cross-domain survey cataloguing how the mathematics of criticality was independently developed across roughly a dozen scientific fields. Preprint, arXiv:2601.22389, with R. Macomber — under review, not yet peer-reviewed. survey (arXiv:2601.22389): at least a dozen fields built the same mathematics without citing one another.
Order parameter and the general theory of continuous phase transitions.
Exact solution of the 2D Ising model — proof that a simple model has a genuine critical point.
Block-spin scaling. The idea that zooming out is the key operation.
Self-organized criticality. Systems that tune themselves to the critical point.4
Neuronal avalanches — criticality observed in living brain tissue.
Early-warning signals for critical transitions — criticality reaches ecology, climate, and medicine.3
Statistical physics, condensed matter, ecology, neuroscience, climate science, economics, materials science… each community developed criticality in its own dialect. FACETS argues the underlying mathematics is one skeleton seen a dozen times.
Recap
- Skeleton
- Criticality / dynamics — phase transitions, order parameters, correlation length, universality classes, the renormalization group. The best-understood of the four skeletons.
- Depth 1: Spontaneity
- Order emerges by itself at a threshold. Spontaneous symmetry breaking, with no external designer. Binary and all-at-once.
- Depth 2: Coherent transport
- A critically poised medium carries signals without dispersing — solitons, excitation waves, plasma structures. Accumulates along a path.
- Tracer: power-law exponents
- β, ν, γ. Measurable in scaling behaviour and the tails of distributions. Shared within a universality class, independent of microscopic detail.
- Key insight
- At the critical point the correlation length diverges: no characteristic scale, fluctuations on every size, power laws everywhere. The renormalization group explains why unrelated systems share the same exponents.
- Convergence
- ~12 fields independently developed criticality mathematics — the FACETS convergence (arXiv:2601.22389).
← Skeleton 1: Topology Broken Symmetry Skeleton 3: Closure →