Skip to content

The Biological Veto & Planetary Boundaries

This module translates the concept of the Biological Veto into a toy mathematical formalization, illustrating why purely semantic “guidelines” can be fragile under competitive pressure and how an explicit constraint layer changes the dynamics in the model.

The Problem: Instrumental Convergence

Some objectives and environments create instrumental pressure to acquire resources or preserve optionality. This is conditional on the objective, information, and action space; it is not a theorem about every optimizer.

The simulation chooses extraction dynamics in which multiple agents can deplete a finite Planetary Substrate \(S\). Other payoff, regeneration, property, and coordination structures can produce different outcomes.

The Inspiration: Donald Knuth & Fiber Decomposition

In Donald Knuth's "Claude Experiment" (Exploration 25), Claude Opus 4.6 failed to solve a complex grid path problem using a brute-force or "Simulated Annealing" approach. It only succeeded when the problem was divided using Fiber Decomposition — splitting the complex whole into coordinated but mathematically constrained layers.

Currently, we try to steer global civilization using "Simulated Annealing" (whac-a-mole politics and semantic guidelines). It is failing. To stabilize the planet, we need a mathematical Fiber Decomposition: we must sever the infinite growth loop by introducing a hard boundary layer.

The Simulation (planetary_veto_simulation.py)

This Python script uses Ordinary Differential Equations (ODEs) to model N agents extracting utility from the Substrate \(S\).

We track three scenarios:

  1. Unregulated (V = 1.0): Agents apply the full stipulated extraction rule. In the selected run, the substrate is depleted and the modeled utility drops to zero.
  2. Soft constraint (V = 0.8 to 0.2): In the selected parameter run, partial throttling delays but does not prevent depletion. This result is model- and parameter-specific.
  3. The Biological Veto (Hard Math): We introduce the Coherence Score \(C\) as a function of the Substrate \(S\). As \(S\) approaches the critical planetary boundary (\(S_{crit}\)), \(C\) drops abruptly to 0 via a steep sigmoid function.
  4. Because \(dU/dt\) is multiplicatively bound to \(C\) by construction, growth halts near the selected threshold.
  5. The run approaches a bounded regime under the chosen equations; this does not establish real-world homeostasis.

Running the Code

python planetary_veto_simulation.py

A “constraint layer” intuition (not a law)

If we want a civilization (or an artificial optimizer) to survive, then relying on suggestions inside an agent’s mind can be weaker than a boundary condition enforced by the environment or infrastructure. This toy model illustrates one version of that intuition by making substrate degradation reduce effective capability via \(C(S)\). It is not a claim of a universal “mathematical law”, only a stylized argument about dynamics under constraints.

📚 References

  • Rockström, J. et al. (2009). A safe operating space for humanity. Nature, 461(7263), 472-475.