Collateral Civilization:Human Survival Under Machine Scale Optimization
Read the paperHarm without hostility.
Could human interests remain part of an intelligent system’s objectives while ceasing to meaningfully constrain its actions?
This essay examines that question through a conditional scenario of machine-scale resource optimization. It distinguishes the value assigned to human welfare from the authority to prevent a harmful action.
The argument is conceptual, not a forecast. It considers protective alternatives and countercases, including cheaper resources elsewhere and diminishing returns.
Scope and limitations
The paper develops conditional models and proposed evaluations. It reports no new empirical datasets or experimental results. It does not claim that existing verification mechanisms could contain a system with planetary reach.
Its focus is the distinction between human interests being represented and human protections being enforceable.
Commercial-interest disclosure
The author is affiliated with For Safety Group Inc. (4SI), creating a relevant commercial interest in verification. The argument concerns general principles and makes no claim about the performance, readiness or safety of any product.
Citation and original record
Braun, N. (2026). Collateral Civilization: Human Survival Under Machine Scale Optimization. Zenodo. https://doi.org/10.5281/zenodo.22685276