THE MDL AUDIT

Minimum Description Length

1. The scientific wager

Science searches for the shortest generative description that remains answerable to Reality. “Short” does not mean few printed symbols. It means that after the ontology, dynamics, constants, state data, algorithms and errors are all transmitted, the receiver can reconstruct the observed relations.

\[L_{\rm total}=L(\text{ontology+dynamics+symmetries+algorithms})+L(\text{parameters})+L(\text{boundary/state data})+L(\text{data}\mid\text{model}).\]

The last term matters. A tiny theory that predicts badly requires a huge correction file. A larger theory can be more economical if it predicts the data precisely. A theory also gains compression when the same cause generates results that otherwise require separate rules.

WSM’s serious claim is not “one sentence beats modern physics.” It is: one real-wave generator may be paid for once and reused across many domains.

2. What a fair audit must count

If a continuous parameter $\theta_i$ is transmitted over an allowed range $\Delta\theta_i$ to precision $\delta\theta_i$, a simple coding estimate is

\[L(\theta_i)\sim \log_2\!\left(\frac{\Delta\theta_i}{\delta\theta_i}\right).\]

Therefore “twenty parameters” is not yet an MDL result. Priors, precision, encoding grammar, symmetries and predictive residuals all matter. This page uses parameter counts as a transparent diagnostic, not as a substitute for the full code length.

3. Mainstream physics — profound compression with a divided foundation

The Standard Model, general relativity and modern cosmology are not a random list of facts. Gauge symmetry, representation theory, local dynamics and statistical inference compress enormous bodies of data. Their precision is part of their simplicity because it keeps the residual code short.

Particle names are not simply added to fitted constants. Many states are generated economically by one symmetry rule. Nor does use of a field variable prove that Nature contains a new material substance for every symbol. WSM may reuse successful equations as response mathematics, but it must explicitly translate each symbol into motion, displacement, timing, recurrence or stress in one Space.

Mainstream achievement. It presently wins the mature prediction contest. Its open foundational question is whether several highly successful mathematical structures and measured constants are expressions of a still shorter common physical cause.

4. WSM — the proposed compressed generator

WSM proposes one infinite, eternal, continuous elastic Space. Its activity is real longitudinal wave motion. Matter is a finite open recurrence: directional plane waves arrive from every direction, converge, cross a centre and continue outward while continually rebuilding an e-sphere. The normalised One Law is

\[\frac{c'(\mathbf x,\hat{\mathbf n},t)}{c_0}=\frac{E_d(\mathbf x,\hat{\mathbf n},t)}{E_{d0}}.\]

Directional wave activity changes the speed of a real plane wave. Changed speed changes its travel time through the e-sphere. Changed time writes phase and a real curved displacement onto the departing front. The surrounding Huygens wave sea carries that changed history onward, and other e-spheres reconstruct from what arrives.

A fair WSM budget cannot simply be advertised as “two inputs.” Its strongest target ledger is:

The WSM opportunity is enormous but precise: derive those reads from one frozen action, then the same paid-for engine can replace many separately specified mechanisms.

5. What the symbols mean in real-wave language

6. Existing bridges — where one structure already does several jobs

These results do not by themselves complete WSM. They show why the programme is worth the decisive solve: the same small mathematical structures repeatedly connect different physical reads.

7. Sector-by-sector compression test

8. Equal input counts would still not mean equal theories

Suppose two theories each required twenty transmitted numbers. They would not thereby be equally simple. One might use twenty unrelated patches; the other might use one recurrent engine whose twenty numbers specify only a particular state. The questions are:

• How much universal structure is paid once?
• How accurately does it predict held-out data?
• How many domains reuse the same mechanism?
• How many new kernels appear after each failure?
• Can the decoder reconstruct the observations without verbal interpretation?

\[\text{causal compression gain}=L(\text{separate sector rules})-L(\text{shared generator+derived projections}).\]

WSM’s distinctive advantage is common descent of explanation. The same directional activity changes crossing time, the same crossing-time difference writes phase, and the same altered fronts rebuild centres, moving eggs, transition trains and long-range timing relations. If a single blind solution returns the numbers, the compression is genuine. If every sector needs an independently chosen read rule, the apparent unity disappears.

9. Free functions, cosmology and hidden description length

A placeholder kernel is not “one parameter.” Its code length depends on how much independent shape information it contains. This is particularly important in cosmology, where a redshift kernel, angular-blur kernel, thermalisation kernel, structure kernel and distance kernel could silently become separate patches.

WSM therefore uses a rigid control before a flexible fit. The arrival-map statement is

\[1+z=\frac{dt_o}{dt_e},\qquad C_D(\tau)=C_0\!\left(\frac{\tau}{1+z}\right).\]

The same generator must dilate an optical carrier near $10^{-15}\,\mathrm s$ and a supernova envelope near $10^6\,\mathrm s$ by the same factor—about twenty-one decades—while preserving angular information. The current logarithmic distance control

\[D_L=R_z(1+z)\ln(1+z)\]

has $q_0=j_0=0$ in its convention and satisfies internal dilation and distance-duality identities, but it does not adequately fit the supernova distance relation. That is useful information: the next kernel must improve the distance data without buying freedom by sacrificing whole-train dilation, sharp images, action accounting or the microwave-background temperature history.

A free kernel earns compression only when the same derived function passes several independent observations. Otherwise it is a disguised library of answers.

10. The fifty-question map — breadth without false arithmetic

The original page added subjective grades across fifty questions. That looked quantitative but had no justified metric: “full,” “partial” and “conditional” are ordinal judgements, not numbers that may be summed. The breadth map remains valuable when used for routing and omission control.

11. Seven bounded calculations that would turn compression into physics

1. Living matter: one frozen directional action produces a stable finite open e-sphere, its conserved currents and its moving family.
2. Interaction chain: one write → propagate → read → stress kernel produces charge and gravity with their observed signs, ranges and universality.
3. Quantum completion: the same transition dynamics yields normalised outcomes, Bell correlations and exactly two transverse optical helicities.
4. Electron response: the finite projection gives Dirac dynamics, $g$, $\alpha$, $F_1$, $F_2$ and anomalous moments blind.
5. Higher matter: a nonlinear $C_3$ solve yields proton/neutron structure and held-out spectrum and scattering data.
6. Cosmic transport: one whole-train kernel jointly fits supernovae, CMB spectrum and $T(z)$, redshift drift, images and thermodynamics.
7. Prediction: freeze all conventions before comparing at least one result that was not used to construct the model.

These are difficult calculations, but they are no longer an unbounded appeal to future explanation. Each has defined inputs, outputs and observations capable of ending a branch.

12. What would break the compressed WSM programme

The programme loses its central claim if any of the following becomes necessary:

• a different constitutive law or free kernel for each successful sector;
• a singular hidden particle, independent field-substance or non-wave agency inserted to rescue the recurrence;
• fitted coefficients presented as geometry after comparison with the target;
• a stable e-sphere that cannot move with the Lorentz/de Broglie relations or cannot carry the required conserved currents;
• a local detector-race model asked to violate the Bell bound it mathematically obeys;
• a cosmological transport law that broadens spectra or images, fails whole-train dilation, or cannot account for $T(z)$;
• precision residuals whose code is longer than the structure WSM claims to remove.

These are not gloomy caveats. They are what make the proposal scientific. A simple theory becomes powerful when a small number of consequences can expose the whole engine.

13. Present MDL verdict

Pay once for one cause. Count every independent choice. Preserve every successful observation. Let Reality choose the shorter true description.