Skip to content

TEO Framework Simulation

This module contains a small toy simulation of the TEO constraint claim: unconstrained optimization can maximize a local target while destroying the substrate that makes the target meaningful.

File

  • paperclip_vs_loving.py compares a naive paperclip-style maximizer with a constrained "loving" maximizer.

Claim Tested

The simulation is a minimal demonstration of the repository's core TEO claim:

Optimization must remain coupled to substrate constraints. If a maximizer ignores the constraint budget, apparent local success becomes system-level failure.

Run

python simulation-models/alignment-and-veto/teo-framework/paperclip_vs_loving.py

Expected Behavior

The naive maximizer accumulates more local output until it crosses the constraint boundary. The constrained maximizer sacrifices some output but remains inside the viable region.