The Principle: PSC and MDL
A self-contained universe cannot have an "outside." Following this idea to its logical conclusion forces a unique selection principle — and from that, a unique physics.
The Central Question
Why do the laws of physics have the specific values they do? Who — or what — chose them?
In L01, we saw that the Standard Model has 25 free parameters — numbers that cannot be derived from within the theory itself. The question is: is that unavoidable, or is there a deeper principle that forces specific values?
Three standard answers exist — and all of them fail.
The constants just happen to be what they are. No deeper explanation. This is honest — but it is not a theory. It says the question has no answer.
Perhaps infinitely many universes exist with different constants. We live where life is possible. But this makes no predictions and "explains" any set of constants equally well — which means it explains nothing.
String theory predicts ~10⁵⁰⁰ possible universes. But it offers no mechanism to select ours. It replaces 25 unexplained numbers with an unexplained selection event.
The key observation: Every proposed answer either defers to something external or accepts the parameters as unexplained primitives. Neither is a final answer.
The real question is: can the universe explain its own laws from the inside?
Perfect Self-Containment (PSC)
Perfect Self-Containment (PSC) is the formal statement that the universe has no "outside" from which its laws could be selected. The laws, their description, and everything needed to run the physics must all be internal — self-generated, self-consistent, self-sufficient.
This sounds almost obvious. Of course the universe contains everything, by definition. But the formal consequences are not at all obvious. If you take the requirement seriously and work out what it implies, you get a filter that eliminates almost every possible physics.
Analogy: the self-contained city-state
Imagine a city-state that must never import anything from outside its walls. To survive, it must satisfy five requirements simultaneously — remove any one and the city becomes dependent on something external:
1. Grow its own food — produce all the energy it consumes
2. Maintain stability — not collapse under its own weight
3. Move people and goods — have internal transport
4. Communicate internally — coordinate its own activities
5. Balance its own economy — no external subsidies
The five Layer I PSC axioms for a universe work the same way. Each requirement is minimal — remove any one and the physics becomes dependent on something external.
Together they act as a filter: only physics that passes all five can be self-contained. As it turns out, very little passes.
The Five Layer I Axioms
These are not physical hypotheses added for convenience — each is the minimally necessary condition for one specific aspect of self-containment. Violating any one reintroduces an external dependency.
Reflexive Closure
Normative Minimality
Thermodynamic Viability
Semantic Admissibility
Anomaly Consistency
The Layer I result: a concrete number
Apply the five axioms simultaneously to the full space of four-dimensional renormalizable gauge theories with compact groups and chiral matter. Start with 34,560 candidates. Filter:
| Filter | Remaining | Eliminated |
|---|---|---|
| All compact rank-≤4 gauge groups with chiral matter | 34,560 | — |
| RC: anomaly cancellation + unitarity | 2,160 | 32,400 |
| NM*: parameter-stable qualitative type | 72 | 2,088 |
| TV + SA: bound states + massless sector | 24 | 48 |
| AC: full anomaly consistency | 12 | 12 |
All 12 survivors share exactly the gauge structure SU(3) × SU(2) × U(1), and all have Ngen ≥ 3. The Standard Model is not one solution among many — it is the essentially unique PSC-consistent gauge structure in four dimensions.
Presentation Invariance = Minimum Description Length
PSC has a Layer II consequence: Presentation Invariance (PI). PI says that a self-contained universe cannot have its physics depend on how it is described.
If you change the coordinates, relabel the fields, or choose a different basis — the physical predictions cannot change. That is just a rephrasing, like translating a sentence from English to French. The meaning is the same; only the representation differs.
Why PI is forced by self-containment
Without PI, there is a problem. If the complexity of a theory depended on how you presented it, then "which presentation to use" would be an external choice. That free choice — the description language — would carry information not generated by the theory itself. External inputs violate PSC.
So if PSC holds, the complexity measure must be presentation-invariant. The unique presentation-invariant measure of an object's complexity is Kolmogorov complexity — the length of the shortest program that generates a complete description of the object.
The Alice and Bob analogy
Alice writes a program that computes a certain output table. Her program is 50 lines.
Bob writes a different program that computes identical outputs. His program is 5 lines.
Both are logically correct. Which one does the universe "use"?
If the universe is self-contained: it cannot care about the presentation — both compute identical physics. But then which program are we actually running? If the universe cannot distinguish presentations, but must pick one, it must pick the one with minimum description length — Bob's 5-line program. Not because brevity is beautiful, but because any longer description contains redundant bits that could only have been placed there by an external choice. External choices violate PSC. MDL is the only selection rule consistent with having no outside.
The chain: PSC → PI → MDL
Perfect Self-Containment (PSC)
The universe has no outside. Everything — its laws, its description, its physics — must be internally generated. No external input of any kind is permitted: no external constants, no external description language, no external selection event.
Presentation Invariance
The universe cannot depend on how its laws are described.
Minimum Description Length
Among all PSC-consistent theories, the shortest description is forced.
Perfect Self-Containment (PSC)
The universe has no outside. Everything must be internally generated.
Presentation Invariance (PI)
Because PSC forbids external inputs, the complexity measure must be presentation-invariant. Any complexity measure that depends on the description language would let the language in as an external input. The unique presentation-invariant complexity measure is Kolmogorov complexity — the length of the shortest program.
Minimum Description Length
Among all PSC-consistent theories, the shortest description is forced.
Perfect Self-Containment (PSC)
The universe has no outside.
Presentation Invariance (PI)
The complexity measure must be presentation-invariant.
Minimum Description Length (MDL)
Among all PSC-consistent theories, the one with minimum Kolmogorov complexity is selected. PI and MDL are not analogous — they are the same instruction stated in different vocabularies. In physics: MDL = "pick the PSC-consistent theory with the fewest bits."
Isn't This Circular?
A natural objection: "You used MDL to select the physics — but you derived MDL by assuming PSC was right. That seems circular."
The regress terminates cleanly:
| "Why MDL?" | Because PSC forces PI, and PI is MDL in physics vocabulary. |
| "Why PI?" | Because self-containment requires presentation independence. A complexity measure that depends on the presentation language would let an external choice in — violating self-containment. |
| "Why PSC?" | Because a fundamental theory has no outside. This is the definition of fundamental. The regress stops here for the same reason causation stops at the laws of physics. |
The chain is not circular. It terminates at PSC, which is an empirically forced starting point: we are looking for a theory that is complete without external inputs. If we accept any external input, we are not looking for a fundamental theory.
What MDL Selects
How MDL applies — from principle to polynomial
MDL does not directly pick a polynomial. It operates through a hierarchy of four objects: NEMS ⊃ UGP ⊃ GTE ⊃ polynomial. Each level narrows what is possible. Here is the chain, step by step.
A — The Universal Generative Principle (UGP)
MDL says: among all self-consistent descriptions of physics, pick the shortest. The Universal Generative Principle (UGP) is a pure arithmetic programme — it does not start from the Standard Model and try to fit it. It asks: starting only from the three axioms below, what arithmetic structures survive? The Standard Model match is what you find when you look at the survivors. It is a discovery, not a design.
The UGP defines its sieve via three axioms:
A1 — Locality: the generating rule depends only on near-neighbor information. No action at a distance, no global inputs.
A2 — Symmetry: the rule is invariant under the relevant algebraic symmetries.
A3 — MDL: among all rules consistent with A1 and A2, the one with shortest description is selected. This is the same functional as PSC's Presentation Invariance — not a new assumption added here.
The sieve operates as follows: for each ridge level n ∈ ℕ, it computes the ridge value R_n = 2n − 16, then scans the integer divisors of R_n to construct candidate triples. The triple (a, b, c) is not chosen independently of n — n determines R_n, and the divisors of R_n determine what triples are even constructible at that level. Only triples passing the prime-lock and mirror-duality consistency tests survive. No SM input enters anywhere — the sieve tests purely arithmetic conditions.
Where do the sieve conditions actually come from?
This is the most important question. The honest answer has two parts.
PSC selects the SM gauge group with N_c = 3 (QCD color rank). From N_c = 3, an algebraic chain then derives every structural integer — including b₁ = 73:
The Lepton Seed's key integer follows from N_c = 3 as a theorem.
Prime-lock (c₁ must be prime) is a MDL minimality condition: primes are the irreducible arithmetic building blocks — the minimum-description-length capacity values.
Mirror-duality (the swap must also pass) is axiom A2 — symmetry under the relevant algebraic operations.
Honest disclosure: GTE was originally discovered by reverse-engineering observed particle masses, then given a derivational grounding. The full formal chain from A1–A3 → sieve structure is part of the ongoing programme (Path 2 covers what is proven).
The key fact: the survivors are extraordinarily sparse. Most (n, triple) combinations fail immediately. Our universe occupies the unique MDL-minimal survivor across all n ∈ ℕ — and when this survivor is interpreted physically, it reproduces the Standard Model. The non-circularity is machine-certified: the selection machinery has no knowledge of SM parameters.
B — The Ridge and the GTE Triple
What is a ridge? At each level n, the GTE (Generative Triple Evolution) arithmetic defines a reference value called the ridge:
The ridge is not a free parameter — it is the arithmetic formula that defines the landscape at level n. At n = 10: R₁₀ = 2¹⁰ − 16 = 1008. The sieve scans for integer triples that are consistent with this ridge value through two hard constraints: prime-lock and mirror-duality.
What is a GTE triple? An integer vector (a, b, c; g) with:
| b | The ladder index — the key output. This is the N-value that will become a particle's arithmetic mass address. |
| c | The branch capacity — links to a Mersenne number (2k−1) at the arithmetic level. |
| a | The parity-phase — encodes chirality information via the Möbius function. |
| g | The generation index: 1, 2, or 3. |
GTE is what L06 covers in detail. For now: a triple is a pure number-theoretic object. No particle, field, or force appears in its definition. The physics emerges from it later.
C — The Sieve Produces One Seed
Epistemological note: The sieve conditions (prime-lock and mirror-duality) were originally discovered empirically — found by reverse-engineering what arithmetic structure generates the observed particle masses. They have since been given a derivational grounding: the key 2026 result (machine-certified) shows that once PSC selects the SM gauge structure with N_c = 3, the Lepton Seed integer b₁ = 73 follows by pure algebra (b₁ = N_c⁴ − a_τ − N_c = 81 − 5 − 3 = 73). The ridge constant 16 is derived from the GF(7) orbit structure; the Fibonacci gap index 7 is derived as ord₇(2) + N_c + 1 from the Z₇×Z₃ algebra; and the a-values (a₁ = 1, a₂ = N_c² = 9, a₃ = (N_c²+1)/2 = 5) are orbit projections — each forced by the arithmetic of the preceding step, not assigned freely. All three results are machine-certified in Lean 4 with zero sorry. The full derivation of every remaining sieve condition from A1–A3 is part of the ongoing programme; Path 2 covers what is proven.
The sieve runs at every ridge level n. Four independent arithmetic certificates force the unique minimal admissible level: n = 10. At n = 10, R₁₀ = 1008.
The sieve then scans all interior divisor pairs (b₂, q₂) of 1008 satisfying the prime-lock and mirror-duality constraints. There are exactly five non-mirror pairs. The test: does b₁·q₁ + 20 come out prime?
| (b₂, q₂) | b₁ | c₁ | Prime? |
|---|---|---|---|
| (16, 63) | 86 | 278 = 2×139 | No |
| (18, 56) | 81 | 425 = 5²×17 | No |
| (21, 48) | 76 | 628 = 4×157 | No |
| (24, 42) | 73 | 823 | Yes ✓ |
| (28, 36) | 71 | 1085 = 5×7×31 | No |
Exactly one pair passes: (24, 42). Its mirror swap (42, 24) also passes (2137 is prime), satisfying mirror-duality. MDL selects the lexicographically minimal branch: the Lepton Seed.
(1, 73, 823; g=1)
The Lepton Seed — one seed from a sieve over all n ∈ ℕ.
The Lepton Seed is not chosen. It is the unique output of the sieve at the forced ridge n=10, with MDL selecting the minimal branch. No free parameter enters anywhere.
D — The Cascade Orbit: Three Generations
The map T (GTE's two-step arithmetic advance) is applied to the Lepton Seed. Each step divides c by b, extracts quotient q and remainder m, then updates all three components by forced arithmetic rules. No parameter enters either step.
→ even step → (5, 275, 65535)
The b-values {73, 42, 275} are the N-values — the arithmetic addresses from which the electron, muon, and tau masses are computed. L06 walks through this step by step with full verification.
The orbit does a second thing simultaneously: its binary parity projection generates exactly 10 (neighborhood, output) constraints, activating 7 of 8 possible binary input triples. This is an interpolation data set of precisely the right cardinality to pin a local update rule — with one bit still free.
E — MDL Closes: The Polynomial Is Interpolated
The orbit's parity shadow provides 7 of 8 constraints needed to pin a rule. The 8th constraint is MDL-equivalent vacuum transparency: empty space must stay empty — f(0,0,0) = 0. This is not a separate assumption; it is the MDL criterion applied to the lookup-table metric (the rule that maps the fewest inputs to 1).
Together — 7 orbit constraints + 1 MDL vacuum constraint — pin a unique rule. The Direct-Interpolation Lift Theorem proves exactly one multilinear rule over GF(7) satisfies all 8 constraints:
p(L, C, R) = C + R − CR − LCR
over GF(7) · 19 bits · unique among 7⁸ = 5,764,801 candidates
The polynomial is not selected from a search over CA rules. It is the unique solution to the interpolation problem the orbit itself poses. The same seed that generates the particle spectrum (masses via N-values) also generates the dynamics (the polynomial via orbit interpolation). One arithmetic object, two arms. That is the triangle — L07.
Why GF(7) specifically? Every Z_N×Z_M alternative is eliminated by a five-step modulus-forcing chain: N<7 has no topological kink orbits; N>7 costs more bits; M≠3 fails orbit classification. Full derivation in Path 2.
The result of this selection: a specific polynomial — p(L, C, R) = C + R − CR − LCR over GF(7). It requires exactly 19 bits to specify, compared to ~963 bits for an exhaustive 7-state lookup table. The 19-bit count is the quantitative certificate that the polynomial actually is the MDL minimum — not just claimed to be.
What the polynomial is, how to compute with it, and why it produces Rule 110 on binary inputs is the subject of L03. This lesson establishes why MDL forces a unique polynomial to exist. The next lesson shows what that polynomial does.
One clarification: this is not a claim that "the universe is a cellular automaton" in any literal sense. A local update rule over a finite field is a mathematical structure — the most compact way to specify a local deterministic law with no external input. The CA language is the mathematical framework for stating the MDL minimization, not an ontological claim about the nature of spacetime. The physical universe is the continuum field Φ_MDL that the polynomial's structure characterizes — covered in Path 3.
Key Takeaways
- PSC (Perfect Self-Containment): the universe has no outside. Its laws, its description, and everything needed to run its physics must be internally generated — no external constants, no external selection event.
- PSC has five formal Layer I axioms — each is the minimal requirement for one specific aspect of self-containment: RC (Reflexive Closure), NM* (Normative Minimality), TV (Thermodynamic Viability), SA (Semantic Admissibility), AC (Anomaly Consistency). Applied to all four-dimensional gauge theories, they leave exactly 12 survivors — all with the SM gauge structure SU(3)×SU(2)×U(1) and N_gen ≥ 3.
- Presentation Invariance (PI) follows from PSC: a self-contained universe cannot have its physics depend on the description language. The unique presentation-invariant complexity measure is Kolmogorov complexity.
- PI is exactly equivalent to MDL: among all PSC-consistent theories, the one with minimum description length is selected. Not as an aesthetic preference — any longer description contains bits from an external choice, violating PSC.
- The regress terminates at PSC — which is the definition of a fundamental theory. The chain is not circular.
- The chain from MDL to the polynomial passes through five stages: (A) UGP defines a number-theoretic sieve with three axioms (locality, symmetry, MDL); (B) the sieve operates on an arithmetic landscape of ridge levels R_n = 2^n−16 and integer triples (a,b,c); (C) the sieve at the forced level n=10 yields exactly one admissible seed — the Lepton Seed (1,73,823) — by prime-lock and mirror-duality constraints; (D) two applications of the GTE map T produce the cascade orbit {73,42,275} which encodes the three lepton generations; (E) the orbit's binary parity shadow plus MDL vacuum transparency interpolates a unique polynomial p(L,C,R) = C+R−CR−LCR over GF(7). 19 bits. Zero free parameters anywhere in the chain.