Monad Cosmology: spacetime, gravity and the constants of nature from a jammed packing of structureless Planck-scale monads

Spacetime here is a jammed packing of identical spheres one Planck diameter across -- monads -- entangled with their neighbours and updated once per Planck time; isotropy forces the sphere, the disorder and the spin-1/2 Heisenberg interaction. Newton's constant is not fundamental but the vacuum's entanglement density, G=1/(4 eta_A), and counting entanglement across a surface returns the black-hole area law with a computable coefficient. The route on to the field equations is Jacobson's, inherited not re-derived. That makes the central identification -- monad diameter is Planck length -- testable, not a calibration. At every box size the full Hilbert space can reach, it comes out short -- and by more than this paper long thought. The entropy across a cut is not proportional to the bonds the cut crosses: it is S=a c + b ln(n_A n_B/N) with b -> 3/2, the logarithm of an ordered ground state with three Goldstone modes, and at every reachable size the logarithm is the larger term. The quantity the anchor needs is a; measured on a design calibrated against a state whose answer is known exactly it is a=0.0474 +/- 0.0013 against the 0.13635 required -- short by a factor of 2.87, which would put d/l_P at 0.590 and the black-hole coefficient at kappa_BH=11.5 against the exact 4, so the shortfall is not a parameter that can be absorbed. The estimator lets no property of the packing enter the slope; over 138 packings at N=12 to 24 the size trend is 2.0 sigma, no larger than the scatter. What is not established is either the level or its convergence: the boxes reach only 2.1 to 2.7 diameters, and the packings themselves have not reached bulk. So what this paper settles is which quantity the anchor constrains, not its value in the thermodynamic limit; reading it as a refutation would be as unearned as reading the number it replaces was as a confirmation. One substance carries the dark sector, split by a theorem and not by hand: dark energy is the medium counted globally, rho_Lambda/rho_P=(l_P/R)^2 with w=-1 exactly, and dark matter the same medium in local circulation, its shear variance read as mass, tracking no baryons and diluting as a^(-3). Its carrier is computed, not fitted -- two fewer free parameters than any particle account -- and direct detection ends it. Whether that carrier forms at all is now measured, and the classical answer is no: in a box of eight diameters a localised twist comes apart in ten ticks, indistinguishably from a gapless magnon, with under three per cent of its chirality left in place. If the object exists it is intrinsically quantum, and the question returns behind the exponential wall. Averaging closes by an identity: sigma_2 of a Jacobian is a null Lagrangian, so inhomogeneity does not back-react. Space is finite and closed, its curvature predicted into a two-sided window Lambda-CDM leaves free. Neither black holes nor the Big Bang has a singularity -- one mechanism, two thresholds -- and where the fabric tears, what it carried is destroyed: energy and information are not globally conserved. The area and first laws return the Hawking temperature exactly, yet evaporation is the tear, so a primordial hole emits no gamma signature. There is no mediating graviton; what radiates is a pattern in the entanglement. The one load-bearing assumption is supersymmetry below 10^9 GeV, unobserved; matter is a separate theory built on this one, deferred not missing. A companion Letter, eight pages, states the central measurement and the argument against it, for a reader who wants the result before the record.