Observer Thresholds for Objectivity: Local Windows on Cellular Automata and Finite Constraint Models

Read full text Summarise with VIRAT
Authors: Xinlei Wang
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Journal manuscript (v2) superseding the v1 preprint. The v1 spectral bound on the single-window residual entropy does not hold and is withdrawn here; the threshold law is proved on the domains stated in the evidence table. When do embedded observers—subsystems of the world they describe, each seeing only a local projection of it—come to agree about that world? Objectivity is taken operationally: a global configuration is objective for K observers when their pooled records determine it exactly; beyond this the word is an interpretation. We model the world as a finite constraint model with one global configuration and observers as cumulative local projections, with two order parameters: the pooled objectivity, the information that the records of K observers fix jointly, and the pairwise parameter mean O_t, the statistical dependence between two records plus a structural term. (i) For elementary cellular automata the minimum number K^* of evenly spaced width-w observers whose pooled records resolve every initial state is given, on the proved domains of the evidence table, by a case split on a local gap-resolution time T^*(f) ∈ {1,2,3,∈fty}, computed for all 256 rules, and a local witness pattern: K^* = 1 for the 28 rules with T^* = 1, ⌈ N/(w+1) ⌉ for 14 rules with T^* = 2, and ⌈ N/w ⌉ for the 178 gap-quenching rules and the two centre-only rules, each under explicit side conditions on N and w; a closed form is thus proved for 222 rules, and the other 34 are resolved only in part. The threshold is not a function of Wolfram class. (ii) For Rule 110 one width-3 observer fails to resolve a positive fraction of initial states at every computed size, rising from 0.254 to 0.969 for even N = 8--24 at horizon 2N under the uniform law (exact enumeration; positivity proved for all N ≥ 6). Both order parameters are non-decreasing in time under the uniform law with noiseless cumulative records, by the data-processing inequality, a consistency property distinct from the threshold. The models have no measurement settings and are not Bell scenarios. Code and data supplement: doi:10.5281/zenodo.23646191.

Fields of Study & Subjects

Cellular Automata and Applications Computability Logic AI Algorithms Complex Systems and Dynamics

Publication details

Journal / source
Zenodo (CERN European Organization for Nuclear Research)
Publication date
25 March 2026
Language
EN
Document type
Preprint