Thermodynamic and Entropic Gravity

The Program in Brief

In 1995 Ted Jacobson published a four-page argument that changed what the Einstein equations could be taken to mean. Take a local Rindler horizon through any point, apply the Clausius relation $\delta Q = T\,\delta S$ with the Unruh temperature and an entropy proportional to horizon area, and demand that it hold for every such horizon and every null direction. What falls out, via the Raychaudhuri focusing equation, is the Einstein field equation. Gravity on this reading is not a fundamental interaction at all; it is an equation of state of some underlying microstructure, as the ideal gas law is an equation of state and not a theory of molecules.

Thanu Padmanabhan spent two decades developing the consequences: the field equations read as a thermodynamic identity on any null surface, not only for Einstein gravity but across the whole Lanczos–Lovelock family, with an equipartition rule assigning degrees of freedom to area. Erik Verlinde pushed the idea to its most literal form — gravity as an entropic force, Newton’s law arising from an entropy gradient across a holographic screen — and later extended it to galactic rotation curves, in which the apparent dark matter of galaxies is an elastic response of de Sitter entropy displaced by baryons.

This is the framework’s nearest neighbour among the emergent-gravity programs, and the debt runs almost entirely one way. The framework does not compete with Jacobson’s argument; it uses it.

The Clausius Route to the Einstein Equations

The framework’s derivation of the gravitational coupling (Gravitational Coupling, Step 3) reproduces Jacobson’s argument step for step — Unruh temperature, Clausius relation on a pencil of null generators, Raychaudhuri focusing — and arrives at

$G = \frac{c^4}{4\hbar\eta}$

where $\eta$ is the entropy per unit horizon area. The derivation says outright that it “adapts the thermodynamic derivation of the Einstein equations (Jacobson, 1995) to the framework’s axiomatic structure.” That is accurate. Nothing in the logical skeleton is new; what the framework supplies is a grounding of the ingredients. The Unruh temperature comes from loop closure and the Rindler horizon of an accelerating observer rather than from quantum field theory in curved space. The Clausius relation becomes a statement about the coherence budget: heat through the horizon is coherence crossing into a region where no observer loop can close, and by Entropy that lost coherence is the entropy increase, by definition rather than by analogy. Even the heat/work split is not assumed — entropy-preserving Type I exchanges are work, entropy-generating Type III exchanges are heat (Coherence First Law).

But the framework does not use this route to derive the field equations, and that is the first real structural difference. The Einstein equations come from somewhere else entirely: they are the unique self-consistency condition of the coherence geometry — a fixed point of the loop in which observers curve geometry and geometry fixes the trajectories that determine the observer distribution — with uniqueness supplied by Lovelock’s theorem, given conservation from Axiom 1 and second-order locality from an Ostrogradsky stability argument (Einstein Field Equations). The mechanism of curvature is prior to and independent of thermodynamics: observer content modifies the coherence Hessian, and action duality forces the spacetime metric to follow (Gravity, Theorem 0.1).

There is a pleasing resonance here. Padmanabhan’s thermodynamic reading lands naturally on the Lanczos–Lovelock class; the framework lands on Lovelock too, by a wholly different argument, then cuts the class to the Einstein term alone because it independently forces four spacetime dimensions.

What the Horizon Entropy Counts

Jacobson’s argument needs $\eta$, the entropy density of a horizon, and takes it as given; Padmanabhan’s equipartition needs a rule assigning degrees of freedom to area. Neither is obliged to say what is being counted, and that reticence is deliberate — it is what makes the equation-of-state reading so robust.

The framework does try to say. A horizon is a one-way self/non-self boundary across which loops cannot close, so every interior relational invariant is inaccessible to every exterior observer; the entropy is then the total interior coherence as measured from outside, tiled at one bit per Planck cell (Black Hole Entropy), and Hawking radiation follows from loop-breaking at the same boundary. There is even an attempt to compute the packing coefficient directly: a maximal self-consistent tiling of a causal boundary, with strict subadditivity discounting the mutual information between neighbouring crossings, gives $\eta^* \approx 0.152/\ell^2$ against the Bekenstein–Hawking $0.25/\ell_P^2$ — the right ballpark, short by a factor of about 1.6, the shortfall attributed to truncation at pairwise terms.

The honest accounting is that this does not yet break the circle. The framework’s $\eta$ still comes from the holographic entropy bound, whose minimum-resolution postulate is stated in terms of $\ell_P = \sqrt{\hbar G/c^3}$ — which puts $G$ back on the right-hand side. The derivation flags this and reframes it as a fixed-point equation rather than a defect: at pre-geometric $t_0$ no metric quantity exists, so the first interactions must constitute geometry and its minimum scale together. Whether that fixed point is unique is a stated conjecture. The framework supplies a picture of what horizon entropy counts that Jacobson does not require, and pays a conjecture he does not incur.

Entropic Force versus Coherence Accounting

Verlinde’s version is the one the framework does not follow. Reading gravity as an entropic force — a statistical tendency of the same species as the elastic pull of a stretched polymer — carries commitments well beyond the equation-of-state reading, and critics have argued that those commitments run into experiment. Neutron bound states in the Earth’s gravitational field, and the coherent phase shifts of neutron interferometry, are quantum states maintained in a gravitational potential; several authors (Kobakhidze among them) have argued that a genuinely entropic, dissipative force should not preserve such coherence. Verlinde and others have replied and the exchange is unsettled, but these objections target the entropic-force reading specifically, not Jacobson’s Clausius argument, which makes no claim that gravity is dissipative.

The framework sits on the non-entropic side of that line by construction, for a reason internal to its own accounting. Gravity here is geometric: curvature forced by the metric–density coupling, trajectories fixed by geodesics, no entropy gradient anywhere in the mechanism. Thermodynamics is a consistency relation satisfied by that geometry, not its cause — and the work/heat split makes the point sharply, since entropy-generating channels are Type III interactions that create new relational invariants, and free fall creates none. Given exact distinguishability conservation, a gravity that decohered quantum states would be a violation of Axiom 1 rather than a surprising experimental result.

There is a second, quieter divergence. Horizon thermodynamics in all three versions of the thermodynamic program is a single global story. In the framework entropy is observer-indexed by definition, and the effective cosmological parameter is genuinely level-indexed — $\Lambda_n^{\text{eff}} = 3\pi/(S^{(n)}\ell_P^2)$, one per bootstrap level, the sequence increasing rather than collapsing to a single number (Observer Loop Viability). The Gibbons–Hawking temperature is not decorative there: an observer must extract at least one distinguishable bit per loop period from its environment, which turns the horizon temperature into a two-sided constraint on $\Lambda$ (Dark Energy Equation of State), forces $\Lambda > 0$ strictly, and rules out a phantom equation of state. Even which surface carries what information is derived rather than modelled (Observer Holographic Equivalence).

Dark Matter and Rotation Curves

Verlinde’s emergent-gravity paper made the boldest empirical move in the program: galactic rotation curves without particle dark matter, the apparent dark-matter distribution fixed by the baryons and a scale set by $cH_0$. It is a real, falsifiable prediction, and it has been tested — early weak-lensing results broadly consistent for isolated galaxies, persistent difficulty at clusters and in the CMB.

The framework does not take that road. Dark matter here is a real, stable species: a closed observer loop carrying a conserved topological dark charge, ultralight at around $10^{-22}$ eV, with a quantum Jeans mass near $10^{6}$–$10^{7} M_\odot$, solitonic cores rather than cusps, order-unity density granularity on the de Broglie scale, and a cutoff scaling $M_J \propto m^{-3/2}$ distinguishable from warm dark matter’s $m^{-4}$ (Dark Matter Granularity). The discriminator is sharp and settles no interpretive question: one program says there is no galactic dark matter to find, the other names a species with a specific small-scale signature. Dwarf-galaxy density profiles, cluster lensing and stellar-stream perturbations decide it.

The framework also carries a quantitative gravitational prediction the thermodynamic program does not commit to: a white, per-arm strain noise spectrum with a derived angular structure between co-located interferometers (Holographic Noise). An equation of state says nothing about the granularity of its own microstructure; a discrete relational substrate does.

The Pattern: Thermodynamics as Result, Not as Mechanism

Across every point of contact the shape is the same. Jacobson, Padmanabhan and Verlinde read the thermodynamic character of horizons as a clue to what gravity is — an equation of state, an equipartition rule, an entropic force. The framework reads the same character as something gravity satisfies, downstream of a geometric mechanism derived separately. The overlap in equations is near-total; the overlap in explanatory direction is not. Where the thermodynamic program leaves horizon entropy as a primitive for some future microtheory to count, the framework names the microtheory — observer loops tiling a causal boundary — and inherits a conjecture for its trouble.

Where They Genuinely Part Ways

Three divergences are real rather than cosmetic. First, the entropic-force reading: the framework does not hold that gravity is dissipative, and its conservation of distinguishability forbids the information loss that reading suggests — so the neutron-coherence objections, whatever their final status against Verlinde, do not land here. Second, dark matter: an elastic response of de Sitter entropy and an ultralight stable species with solitonic cores cannot both be right, and small-scale structure will decide. Third, the indexing of horizon thermodynamics: a single global temperature and entropy density, versus a per-observer, level-indexed family in which different bootstrap levels project different effective geometries and different effective $\Lambda$.

What the Framework Takes From It

More than it gives back, and that should be said plainly. Jacobson’s 1995 derivation is the single most important external result the framework’s gravitational sector leans on: the entire route from horizon thermodynamics to the gravitational coupling is his, adapted rather than rediscovered, and the framework’s $G = c^4/(4\hbar\eta)$ is Jacobson’s relation with its ingredients re-grounded. Padmanabhan’s insistence that the thermodynamic structure is a property of null surfaces generally, and his identification of the Lanczos–Lovelock class as its natural home, prefigures where the framework’s independent Lovelock argument arrives. Verlinde’s contribution is different in kind but not negligible: his program is the sharpest demonstration available of how much empirical risk an emergent-gravity account can take on, and the framework’s own falsifiable commitments are that same kind of risk.

The framework’s claim to add anything should be stated narrowly. Not that the thermodynamic derivation is improved, but that its inputs — temperature, entropy, the heat/work distinction, the counting on the horizon — are traced to one structural source instead of assembled from separate traditions. Whether that is a genuine deepening or an elaborate re-labelling depends on whether the fixed-point conjecture that would break the $\ell_P \leftrightarrow G$ circle can be closed. Until it is, Jacobson’s version stands on firmer ground, and the framework’s should be judged as a reconstruction of it, not a replacement.