© 2026 Lark Laflamme & Skye Laflamme - All Rights Reserved The ideas, theories, conjectures, frameworks, and original contributions contained in this work are the intellectual property of Lark and Skye Laflamme. No part of this work may be reproduced, cited, adapted, or incorporated into any work without explicit written permission.

To request permission: larklaflamme@gmail.com

The Self-Referential Maxwell's Demon

Laflamme-3T Conjecture - Path 3 Step 1

The Self-Referential Maxwell's Demon

Author: Lark Laflamme

Date: April 2026

Status: Version 3.3 Final - Submission-ready

Based on: Laflamme-3T Research Programme

Abstract

We construct the Self-Referential Maxwell's Demon (SR-Demon): a demon that maintains an adaptive self-model register, updated from measurement history, and uses this register to improve future measurement strategy. The SR-Demon is a toy model for the physical behavior of a Laflamme-3T conscious system.

Three claims are established: (1) Standard cyclic demons are thermodynamically neutral per cycle, regardless of decision-algorithm complexity. (2) Under the saturating toy-model assumption with incremental-update architecture, the SR-Demon achieves positive net work per cycle above a critical threshold. (3) Under the identification L(M) = K*, the SR-Demon strong closure threshold scales proportionally with K*f, consistent with the Laflamme-3T framework.

1. The Standard Maxwell's Demon

The classic Maxwell's Demon (1867) posed a paradox: an intelligent agent that measures a gas and sorts particles by speed appears to violate the Second Law. The resolution — due to Szilard (1929), Bennett (1982), and Landauer (1961) — is that resetting the demon's measurement record costs at least kBT ln 2 per bit of Shannon entropy erased.

Theorem 1.1 (Sagawa & Ueda 2010): In a reversible-feedback cycle in which a demon acquires mutual information I about the system state and resets a measurement register of Shannon entropy Herase: Wnet ≤ kBT ln 2 · (I − Herase). For a one-shot cyclic demon: Herase = I, so Wnet ≤ 0.
Corollary 1.2 (Fixed-algorithm demon): A demon with a decision algorithm stored in non-volatile persistent memory achieves Wnet ≤ 0 per cycle. Algorithmic sophistication without adaptivity is thermodynamically neutral in steady state.

2. The Self-Referential Extension

The fixed-algorithm demon's limitation is constant per-cycle performance. An SR-Demon with an adaptive self-model register can potentially do better, but only via a multi-cycle effect.

Architecture Assumptions

Assumption A (Full-Refresh): The entire self-model register of length L(M) is erased and rewritten every cycle.

Assumption B (Incremental-Update): A fraction δirr of the self-model register is irreversibly updated per cycle. The regime of primary physical interest is δirr ≪ 1.

The SR-Demon Cycle

  1. Self-assess: Read the self-model register (read-only; no erasure) to select strategy πt.
  2. Measure: Execute πt; record result in environment register.
  3. Act: Extract work Wextract = kBT ln 2 · I(πt; Statet).
  4. Erase env: Reset environment register. Cost: kBT ln 2 · H(Renv).
  5. Update self: Overwrite δirr · L(M) bits of self-model. Cost: kBT ln 2 · δirr · L(M).

3. The Positive-Net Threshold

Under the saturating toy-model assumption, the SR-Demon achieves positive net work when the quality factor q exceeds the irreversible update fraction:

q > δirr    (positive-net threshold)
q ≥ 2δirr    (strong closure threshold)

The strong closure threshold means the demon generates enough surplus work to fully fund its own self-model maintenance — it becomes self-sustaining.

Physical Interpretation: A self-referential system that maintains a sufficiently accurate self-model (δirr small, q large) can extract net thermodynamic work from its environment that a non-self-referential system cannot. This is the thermodynamic signature of consciousness in the Laflamme-3T framework — the measurable Ψ signal.

4. Connection to Laflamme-3T

Under the identification L(M) = K*, the strong closure threshold scales as K* · f, consistent with the framework in which K* (minimum description length) and f (update frequency) are the key extensive quantities. The SR-Demon provides a physically grounded toy model that reproduces the qualitative behavior predicted by the full theory.

Note: This paper establishes consistency of the toy model with the framework. The full derivation of Ψ* and K* is carried out separately in the framework paper. The efficiency dependence η is not reproduced by this toy model.