dcondrey

"""Make-or-break: is self-measurement perturbation REDUCIBLE or IRREDUCIBLE?

A system reads its own temperature. Reading is computation; computation heats the substrate; the heat
changes the temperature read. We ramp self-measurement intensity UP then DOWN and measure:

  delta_min  : temperature rise from the MINIMUM possible self-observation. >0 => you can never
               sample your own unperturbed state.
  hysteresis : temp(intensity) up-ramp vs down-ramp. A nonzero loop => the reading depends on
               measurement HISTORY (thermal mass), so the true idle state is unrecoverable from any
               single reading => IRREDUCIBLE. ~Zero loop => memoryless removable offset => reducible.

dcondrey

"""Self-contained phone/constrained-device build of the self-introspection benchmark.

Standalone (no imports from benchmark.py) so it runs from a single downloaded file on a-shell / iOS
/ any minimal Python. Robust by design: every modality probe catches ALL exceptions and records why
(constrained platforms raise platform-specific errors), and the whole run is wrapped so any fatal
error is reported as ONE line (+ uploaded) instead of a wall of tracebacks. Prints a REPORT-BACK
block at the very bottom: a short paste URL if upload works, and always a compact one-line summary.
"""
import json
import os

dcondrey

"""Portable cross-substrate benchmark: the real cost of self-introspection.

Runs on any machine (macOS / Linux / Windows / iOS a-shell; CI, Modal, Colab, Kaggle, laptop, phone)
and measures, in the machine's own CPU-seconds, two SUBSTRATE-INVARIANT raw facts:

  1. cost_per_look (seconds) for self-observation at three modalities:
       shallow_inproc  : a cheap in-process self-read (load average)        -> expected ~free
       proc_inproc     : richer in-process self-read (/proc, thermal files)  -> Linux; middle
       deep_subprocess : rich self-telemetry via a spawned tool (the body)   -> expected expensive
  2. cost_per_work (seconds) across a SWEEP of work-unit sizes.

dcondrey

"""Portable cross-substrate benchmark: the real cost of self-introspection.

Runs on any machine (macOS / Linux / Windows / iOS a-shell; CI, Modal, Colab, Kaggle, laptop, phone)
and measures, in the machine's own CPU-seconds, two SUBSTRATE-INVARIANT raw facts:

  1. cost_per_look (seconds) for self-observation at three modalities:
       shallow_inproc  : a cheap in-process self-read (load average)        -> expected ~free
       proc_inproc     : richer in-process self-read (/proc, thermal files)  -> Linux; middle
       deep_subprocess : rich self-telemetry via a spawned tool (the body)   -> expected expensive
  2. cost_per_work (seconds) across a SWEEP of work-unit sizes.

dcondrey

"""Portable cross-substrate benchmark: the real cost of self-introspection.

Runs on any machine (macOS / Linux / Windows / iOS a-shell; CI, Modal, Colab, Kaggle, laptop, phone)
and measures, in the machine's own CPU-seconds, two SUBSTRATE-INVARIANT raw facts:

  1. cost_per_look (seconds) for self-observation at three modalities:
       shallow_inproc  : a cheap in-process self-read (load average)        -> expected ~free
       proc_inproc     : richer in-process self-read (/proc, thermal files)  -> Linux; middle
       deep_subprocess : rich self-telemetry via a spawned tool (the body)   -> expected expensive
  2. cost_per_work (seconds) across a SWEEP of work-unit sizes.

dcondrey

test

dcondrey

#!/bin/bash
set -e
lscpu 2>/dev/null | grep -E 'Model name|L3' || true
echo Arch: $(uname -m)
echo Cores: $(nproc)
free -h | grep Mem || true
nvidia-smi --query-gpu=name,memory.total --format=csv,noheader 2>/dev/null || echo GPU: none
apt-get update -qq >/dev/null 2>&1
apt-get install -y -qq gcc >/dev/null 2>&1
curl -sSf https://sh.rustup.rs | sh -s -- -y --quiet >/dev/null 2>&1

dcondrey

/**
 * Proof-of-Humanity Bypass — v1.0.0
 * ==================================
 * Demonstrates that ANY client-side "proof of human authorship" system
 * can be defeated because the attacker controls the execution environment.
 *
 * Author:  David Condrey · WritersLogic · writerslogic.com
 * License: MIT
 *
 * ┌──────────────────────────────────────────────────────────────────┐

dcondrey

repetitivephrases() {
  recurse=0; minw=3; maxw=20; limit=500; incl_csv=""; excl_csv=""; allfiles=0; mincount=2; single_file=""; internal_check=0
  OPTIND=1
  while getopts "rm:M:n:E:X:Ac:f:h" opt; do
    case "$opt" in
      r) recurse=1 ;;
      m) minw=$OPTARG ;;
      M) maxw=$OPTARG ;;
      n) limit=$OPTARG ;;
      E) incl_csv=$OPTARG ;;

dcondrey

import struct
import re
from collections import Counter

class DatPatterns:
    def __init__(self, filepath):
        self.filepath = filepath
        self.content = None

    def read_file(self):