Epiplexity
Information theory for real observers with finite compute.
The Problem
Classical information theory (Shannon entropy, Kolmogorov complexity) assumes unbounded computational observers. Real systems have limits.
The Decomposition
For bounded observers, information splits into:
- Epiplexity (S_T): Structural, learnable content
- Time-bounded Entropy (H_T): Residual unpredictability
Application to Resonant Transformer
The flux ratio |Im(flux)| / |Re(flux)| maps to local epiplexity.
- High ratio = high computational cost to maintain without underlying generator
- This may be the signature of hallucination: generating complexity that looks like information but has no structure
Hallucination Detection Hypothesis
| Content Type | Re(flux) | Im(flux) | Interpretation |
|---|---|---|---|
| Fact | High | Low | Clear structure |
| Pun/Poetry | High | High | Valid interference |
| Hallucination | Low | High | Pseudorandomness |
The "wobble" in flux represents the energy cost of maintaining fiction.
Reference: Finzi et al. (2026), "From Entropy to Epiplexity" arXiv:2601.03220