>[!warning]
>This content has not been peer reviewed.
# Solomonoff induction & Kolmogorov complexity — RST application
**Total ontology pillar:** *Information theory (Shannon/Solomonoff) — the logic of selection.*
Solomonoff induction says the shortest description is the most probable reality; Kolmogorov complexity $K(x)$ is the length of that description. In RST the substrate has a **finite bit budget** (Landauer, A2): maintaining a state costs at least $k_B T \ln 2$ per bit. So the substrate **naturally favours compressed states** — high signal-to-noise $\eta = \Omega/N$, high fidelity $\mu$ — because they cost less to maintain. **Occam's Razor is derived from the ledger**: the universe prefers simple laws because they are cheaper to sustain.
---
## I. Mapping to RST
| Solomonoff / Kolmogorov | RST reading |
|:---|:---|
| Shortest description $K(x)$ | Fewer bits to maintain → lower Landauer cost; substrate favours low $K$. |
| Prior $\propto 2^{-K(x)}$ | Probability of a state scales with **fidelity** $\mu$: high $\mu$ = high $\eta$ = compressible = favoured. |
| "Law" = compact encoding | **Substrate macro**: a law is a shortcut that reduces the work of maintaining a pattern; $W^n = \Omega^n + N^n$ with high $\Omega/W$ = low effective $K$. |
| Occam's Razor | Selection by **maintenance cost**: ordered relations (high $\eta$) are cheaper than random noise (low $\eta$). |
**Identity:** The RST fidelity $\mu(\eta, n) = \eta/(1+\eta^n)^{1/n}$ is the **substrate's selection curve**: as $\eta$ (signal-to-noise) increases, $\mu$ increases; the substrate "chooses" states that maximise $\mu$ for a given budget $W$, i.e. **compressed** (lawful) states.
---
## II. Script: fidelity as selection (Occam from the ledger)
The script **[[Solomonoff Induction - Code]]** (`rst_solomonoff_fidelity_selection.py`) plots $\mu(\eta, n)$ over a range of $\eta$ and annotates:
- **Low $\eta$** (noise-dominated, incompressible) → low $\mu$ → high relative maintenance cost → disfavoured.
- **High $\eta$** (signal-dominated, compressible) → high $\mu$ → low relative cost → favoured.
So the **same curve** that gives the Newton–MOND transition (gravity) and the Resource Triangle is the **logic of selection**: the substrate allocates budget so that compressible (lawful) configurations win. Run via **`run_all_further_scripts.py`** or from this folder.
---
## III. Total ontology tie
- **Thermodynamics (Landauer):** Cost of existence → defines the ledger.
- **Information (Shannon/Solomonoff):** Logic of selection → **this application**: $\mu(\eta,n)$ as the curve that favours compressed states; Occam from the ledger.
- **Fractal geometry (percolation/Bejan):** Shape of the wires → $n$ as backbone dimension.
- **Dynamics (MOND/Navier–Stokes):** Flow of the render → $\eta = g_N/a_0$, etc.
- **Intelligence (Friston):** Optimisation of the signal → life as active noise suppression.
---
## Links
- **Knowledge note (established result):** [[expanded theory/knowledge/The shortest description is the most probable reality]]
- **Code:** [[Solomonoff Induction - Code]]
- **Resource Triangle:** [[expanded theory/Resource Triangle]]
- **Roadmap:** [[../../Applications Roadmap]]