Transputation: Quantum Measurement from MDL
Why does quantum measurement give a definite outcome? Copenhagen calls it collapse and gives no mechanism. GTE proves the mechanism must exist — and classifies it at exactly Turing degree 0′, one level above ordinary computation, machine-certified with zero sorry.
The Measurement Problem
Quantum mechanics says a particle can be in a superposition — genuinely in two states at once. The Schrödinger equation governs how this superposition evolves over time. It is deterministic and smooth.
But when you perform a measurement, you get exactly one outcome. The particle is at position A or position B, not both. The smooth wavefunction "collapses" to a single definite result with probability P(k) = |ck|².
The problem in one sentence: Nothing in the Schrödinger equation describes collapse. The equation predicts an evolving superposition forever — yet we always see a definite outcome. What causes the transition from "all possibilities at once" to "exactly one result"?
This is the measurement problem. It has resisted a satisfying resolution for a century. Two main frameworks attempt an answer:
Copenhagen: collapse, but no mechanism
The dominant interpretation since the 1920s: before measurement, the wavefunction describes all possible outcomes simultaneously. At the moment of measurement, it collapses to a single definite outcome, with probabilities given by the Born rule. This recipe works. It does not explain what collapse is.
Open questions that Copenhagen never answers:
- What counts as an observer or measurement device?
- What is the physical mechanism of collapse?
- Why does the Born rule give P = |ck|² and not some other formula?
Many-worlds: no collapse, but too many universes
Hugh Everett's 1957 proposal: collapse never happens. Instead, the universe splits into a branch for each possible outcome. Every measurement produces every result — in different branches of a constantly splitting universal wavefunction. Each version of you sees one outcome. There are just very many versions.
Problems that remain:
- What determines when and how the universe splits?
- What selects the "preferred basis" for splitting?
- The Born rule still must be postulated — why not P = |ck| or |ck|⁴?
- An infinite-branching reality is unverifiable by definition.
What's missing from both
Both frameworks share a structural flaw: they take the universe's measurement mechanism as external to the theory. Copenhagen invokes an observer outside quantum mechanics; many-worlds invokes a meta-rule about splitting outside any physical dynamics.
A self-contained universe cannot rely on an external observer or meta-rule. The measurement mechanism must be derivable from the universe's own structure. This is exactly what the PSC requirement — Perfect Self-Containment — demands.
A Brief Map of Computational Difficulty
To state the main result precisely, we need a short introduction to how mathematicians classify the difficulty of problems. This is called the theory of Turing degrees.
Some problems are computable — there is an algorithm that always terminates and gives the right answer. Some problems are not. Among the non-computable problems, there is a hierarchy of "how hard" they are.
The halting problem (Turing, 1936): Given a program P and an input x, will P(x) ever stop? Turing proved no algorithm can correctly answer this for all (P, x) pairs. The halting problem is the canonical example of a degree-0′ problem: non-computable, but not even harder than necessary.
The Physical Incompleteness Theorem
Before transputation, there is a foundational result that constrains any self-contained universe's measurement mechanism.
This is analogous to Gödel's incompleteness theorem for logic: just as no consistent formal system can prove all truths about itself, no computable procedure can adjudicate all self-referential measurement outcomes. The incompleteness is not a defect — it is a necessary structural feature of any self-describing universe.
Transputation: The Forced Internal Adjudicator
PIT says: a computable selector cannot exist. But something must select the outcome. The universe cannot leave measurement results undefined. So what does the selecting?
The GTE framework proves this is not a gap to be filled by a future theory — the answer is forced by the structure of any PSC-consistent, record-bearing, diagonal-capable universe.
What transputation is not
Copenhagen invokes an external observer or apparatus. Transputation is internal — no external apparatus is invoked or required.
Many-worlds keeps all branches. Transputation selects exactly one outcome — only one branch is realized per event.
Hidden-variable theories (like de Broglie-Bohm) assume a computable deterministic mechanism. Transputation is provably non-computable by the PIT argument.
The five formal constraints (D1–D5)
Existence
The adjudicator exists — forced by PSC together with diagonal self-reference and the requirement for stable records.
Definiteness
Exactly one outcome is produced per measurement event. The result is a finite, definite record — not a probability amplitude.
Non-computability
The adjudicator cannot be total-computable. Any computable selector would decide the halting problem (PIT). This is the negative constraint.
Lawfulness
The selector is not arbitrary — it has a unique minimal-description form: ρ* = argminρ D(ρ | w), the MDL-minimal state given the current record w.
Born consistency
The marginal probabilities of the selector equal |ck|². This is the interface to the Born rule — derived as a constraint on the selector, not assumed as input.
The Turing Degree 0′ Result
The main technical result of the transputation framework is a two-sided classification of where quantum measurement sits in the Turing hierarchy.
The result is two-sided. Each direction requires its own separate proof:
The halting problem reduces to diagonal record-truth: if you could solve "does this quantum record stabilize?", you could also solve the halting problem. Therefore quantum measurement is at least as hard as halting.
Machine-certified unconditionally (CatAL).
The adjudicator is limit-computable — approximable by a sequence of computable guesses that eventually stabilize. By the Shoenfield limit lemma, any limit-computable function has degree at most 0′.
Machine-certified conditionally on premises PR1–PR5 (CatAL).
Why not higher? If the adjudicator were at degree 0″ or higher, it would require access to a halting-problem oracle for its own adjudications — nested oracles. But realized adjudication outcomes are ordinary finite records: a measurement result is a click in a detector, a spot on a screen. Finite records are degree-0 facts. No nesting is required.
The "≤ 8 mind-changes" refinement
The limit-computability of the adjudicator can be made quantitative. A limit-computable function is one that can be approximated by a sequence of guesses that mind-changes — flips its answer — only finitely many times. The GTE adjudicator, for a measurement on a system of weight w, changes its mind at most 2(kw − 1) times, where kw is the number of PSC-admissible outcomes for record w. For a standard two-outcome measurement (kw = 2) the bound is at most 2 mind-changes.
Key Results
- The measurement problem is real: the Schrödinger equation never produces collapse, yet we always observe definite outcomes. Copenhagen and many-worlds both defer to an external or meta-level mechanism.
- The Physical Incompleteness Theorem proves no total-computable algorithm can serve as the selection mechanism in a PSC-consistent universe (CatAL).
- Transputation is the forced internal adjudicator: it exists, produces
exactly one outcome, is non-computable, is lawful (D1–D5), and is Born-consistent
(CatAL:
closed_choice_forces_transputation). - The Turing degree is exactly 0′ — one level above ordinary computation, two-sided machine-certified. Not easier (lower bound), not harder (upper bound).
- This is a third category distinct from Copenhagen, many-worlds, and hidden-variable theories — not a variant of any existing interpretation.
Source material
- L24 — The Born Rule as a Theorem: the four independent derivation routes