Seeing the universe as a computational process, without going through physics ── a series that views the cosmos from computation theory.
"The universe is literally a computation" is stated explicitly as a hypothesis, then followed honestly.
The starting point is a naive intuition ── "the current description is too complicated. It must be writable more simply." Follow this without flinching, and it leads to "coordinates don't make it simpler," "what shrinks is when you find a redundancy," "dimension is not a number the world holds but a reading of the measuring method," and finally lands on a single blueprint: "how to compute the universe."
Coordinates don't make it simpler (reversible = information-preserving). What shrinks is when you find a redundancy. Dimension is not "a number it holds" but "a reading," and \(F=1/(Cn)^D\) is not a force law but a ruler for dimension. "The easiest dimension" is not one for the world but per phenomenon (= the upper critical dimension).
The question has 3 layers. Resources are firm ground (10⁹⁰ bits, 10¹²⁰ operations, RAM ∝ time² = c·t). The architecture is already screened (a lattice CA fails on Bell + Lorentz; the front-runner is quantum circuits ~ tensor networks). The remaining hole is the update rule, and if it's computationally irreducible you don't "solve" it but "run" it.
Continuous and discrete are of the same family (points are the lead). Lower the points and other representations appear (the algebra of observables = noncommutative geometry, category theory, p-adics, type theory). The zoo is the shadow of one invariant tied together by a web of dualities. Even the sole axis "info-side / physics-side" is a duality = a reading direction. The direction matters only at the holes = your bet.
Enough description. In the browser, actually generate causal graphs (chain / lattice 2D / 3D / 4D / tree), count N(r) by BFS from the center, and measure the slope = dimension on the spot. Dimension is not a declaration but a measured quantity of connectivity = an output. But a lattice bakes the dimension into the wiring = half a cheat ── growing it from a rule is the hole.
Many rules have no shortcut; to know the state n steps later you must actually run n steps (the system is its own fastest simulator). The root is universality plus the halting problem and Rice's undecidability. If the universe's rule is irreducible, you can't get ahead of it. But "unsolvable" isn't defeat ── it's the science of hunting for reducible islands, the source of novelty, the basis of the code.
Erasing 1 bit gives off at least kT ln2 of heat (experimentally confirmed). Computation itself, made reversible, is in principle free ── the cost is tied not to "computing" but to "erasing." Maxwell's demon is resolved here too. Episode 3's "information = physics" becomes one measurable number, kT ln2. Dissipation and the arrow of time = information you stopped tracking.
The topic "universe = computation" overlaps a common speculation and its map. The difference is one point: do you plant the flag of "solved," or the marker of "here is the edge"? With a verdict checker, watch the same topic flip between flag ⇄ marker. The ruler turns on the writer (the AI) as well. We drew the map; we planted no flag.
Since dimension D is a reading, it opens into the complex numbers. Put in \(D=a+i\beta\), and the real part a = decay (the ordinary dimension of size), the imaginary part β = a log-periodic oscillation = the tick of a favorite zoom factor \(\lambda=e^{2\pi/\beta}\). It's the mark of continuous scale symmetry broken down to discrete (the complex dimension of the Cantor set, log-periodic precursors). We are on the real side with zero imaginary part = no closing period → mass keeps running and falls to the geometric mean \(\sqrt{m_{\text{IR}}M_{\text{Pl}}}\) = meV.
The formulas of information "count" (\(S=k\ln W\), \(S\le A/4\), \(kT\ln2\); finite, algebraic, no derivatives); the formulas of physics "flow" (Einstein, Schrödinger; continuous, derivatives). The one operation that connects them: take the degrees of freedom to \(\infty\) and make it exact. Difference → derivative, sum → integral. Literally, impose area entropy + Unruh + Clausius exactly on every horizon and the Einstein equations drop out (Jacobson). But the continuous limit stands only at a critical point, and the area law is an assumption ── which shore is the foundation is undecided.
Apply the ruler \(D=-d\ln F/d\ln r\) to the four forces. All start at \(D=2\) at ultra-short range (the inverse-square of 3+1 dimensions). The only difference is how they run at long range ── gravity and electromagnetism stay at 2 (massless mediators), the weak force runs \(2\to\infty\) (the exponential cutoff of the heavy W/Z = short range), the strong force runs \(2\to0\) (confinement = doesn't diminish with distance). The answer to "what are the dimensions of the weak and strong forces?" is not one integer but a running function. \(C\) (coupling) and \(D\) (geometry) are separate axes.
The inversion of Episode 1's "D is a reading." Among the readings, there is exactly one load-bearing tick ── only a force's \(D=2\) (3-dimensional space) carves a valley in the effective potential (Ehrenfest), closes orbits (Bertrand), and makes atoms, planetary systems, and memory possible. The boundary is \(D=3\). And \(D\) is a measured quantity: the precision measurement of inverse-square (~50 µm, Eöt-Wash) is the experiment for "is \(D\) exactly 2?", and a deviation is the signature of a hidden dimension (\(1/r^{2+n}\)).
The connective episode that, around a reader's one line ── "force = a spring vibration passing to the next lattice site while diluting, so it can be computed" ── bundles Episodes 1–11 into a single correspondence table. Field = coupled oscillators, mediating particle = ripples, finite speed = \(c\) = the light cone, dilution = \(1/r^D\), mass term = an on-site spring (the weak force), nonlinear spring = the strong force, \(D=2\) = the blade that stores stably. Local + finite speed = it runs on a computer (lattice QCD). But "it runs" ≠ "it's easy to solve" (Episode 5), and whether the lattice is scaffolding or foundation is undecided (Episodes 2, 9).
Chase "are electrons and quarks little balls spinning around a spring lattice?" and not one ball remains. Matter is ripples in a spring web, one per kind. The electron is not a spinning ball but a standing wave ringing in the proton's bowl = the atom is a bell, the levels are pitches, the spectral lines are overtones (so it doesn't collapse = Episode 11). Inside the proton is a swirling sea of confined nonlinear field, and 99% of its mass is field energy. Everything is a field, so it runs on a computer (Episode 12). But whether the lattice is scaffolding or foundation, and the mass gap, are undecided.
The episode that turns "dimension is a reading" onto the universe itself. The container (space) is ~3 by inverse-square, but the contents' cosmic web has a fractal dimension of about 2 and is not 3 (approaching 3 at the homogenization scale). The quantum-gravity UV is 2 too, information is 2 too, and extra dimensions (string, hypothesis) give 9–10. The numbers don't converge on one ── the ruler and the scale decide. It closes Episode 1's "dimension is a reading" at cosmic scale.