The Species That Became Too Powerful
Status: draft.
Opening Question
A single human being, in 2026, can — with the right access — end a city. The same species, ten thousand years ago, could barely reshape a valley over generations. What changed was not human nature. What changed was leverage. The opening question of this book is simple to state and uncomfortable to sit with: what happens when a species whose instincts were shaped for small, face-to-face groups acquires the power to affect the entire planet?
Historical Perspective
For the overwhelming majority of the roughly 300,000 years Homo sapiens has existed, the largest unit any person needed to cooperate, compete, or negotiate with was a band of a few dozen, occasionally a tribe of a few hundred, related by kinship, proximity, or long-standing alliance. Robin Dunbar's work on primate group sizes and cognitive limits proposed roughly 150 as the ceiling on stable social relationships a human brain can maintain — a number that shows up with striking consistency across hunter-gatherer bands, historical military units, and even modern organizational subunits. Whether or not the exact number holds, the underlying claim is well supported: human social cognition evolved to handle small, repeated, reputation-tracked interactions, not anonymous coordination among millions.
Then, in a period that is vanishingly brief on an evolutionary timescale — roughly the last 12,000 years, and overwhelmingly the last 300 — the scale of what humans could build, organize, and destroy expanded by many orders of magnitude. Agriculture allowed populations to concentrate. Writing allowed knowledge to outlive its originator. States allowed coordination among strangers. Fossil fuels allowed one person's labor to be multiplied by machines. Nuclear physics allowed one weapon to do what previously required armies. Computing and biotechnology are, in this same lineage, the latest multipliers — and they are not slowing down.
The gap this opens is the subject of this book: capability has scaled by orders of magnitude; the wiring for regulating that capability has scaled hardly at all.
What Modern Evidence Suggests
This is not merely a rhetorical observation. Several independent lines of evidence point at the same mismatch:
- Evolutionary psychology documents cognitive biases — in-group
favoritism, discounting of statistically distant harms, short time horizons for reward — that were adaptive for small-group survival but produce systematically bad judgment at planetary scale (see 03_science_of_flourishing/04_cognitive_bias.md).
- Complexity science and network theory show that as systems grow in
scale and interconnection, failure modes change qualitatively, not just quantitatively — small local errors can cascade globally in ways that never occurred, and could not occur, in small isolated bands. Financial contagion, pandemic spread, and cascading infrastructure failures are modern instances of a structural risk that simply did not exist for a band of forty foragers.
- Existential risk research (Toby Ord's The Precipice, Nick Bostrom's
work on global catastrophic risk) argues that humanity's cumulative risk of a civilization-ending event has risen sharply in the last century specifically because destructive capability has outpaced the coordination mechanisms needed to govern it responsibly.
Where the Principle Fails
It would be a mistake to read this chapter as declinist nostalgia for a simpler past. Small-scale societies were not peaceable by nature — the archaeological and ethnographic record of pre-state violence (discussed further in 04_violence_and_the_escalation_problem.md) shows homicide and intergroup conflict rates often far higher, per capita, than in modern states. The claim here is narrower and more specific: it is not that ancient humans were more virtuous, but that ancient consequences were bounded by ancient scale. A band that destroyed itself destroyed itself; it could not take the biosphere with it. That ceiling no longer holds, and nostalgia for a lower-capability past is neither available nor, given the reductions in violence, disease, and poverty technology has also enabled, obviously desirable.
Civilization Design Principle
> Institutions must scale their capacity for foresight and restraint at > least as fast as they scale their capacity for consequential action.
Any capability that increases what a single actor — a person, a corporation, a state — can affect should trigger a proportional increase in the oversight, transparency, and reversibility built around that capability. Capability without correspondingly scaled governance is not neutral; it is a standing invitation to catastrophe, whether the catastrophe arrives through malice, error, or simple negligence.
Institutional Translation
- Technology governance that treats scale of potential harm, not just
intent, as the trigger for regulatory scrutiny (see 10_technology/08_existential_risk.md).
- Dual-use research oversight in biotechnology and AI proportional to
the ceiling of possible harm, not merely the likelihood of misuse by any one actor.
- International coordination mechanisms built specifically for risks
that no single nation's institutions were designed to handle alone (see 05_power_and_governance/08_global_governance.md).
Metrics
- Ratio of a technology's plausible worst-case impact radius to the
governance and oversight resources allocated to it.
- Time lag between a capability becoming available and a corresponding
regulatory or normative framework being established.
- Number of independently redundant safeguards (technical, legal,
institutional) protecting against catastrophic misuse of high-leverage technologies.
Questions Still Unresolved
Who gets to decide how much foresight is "enough" before a capability is released, and by what legitimate authority? Overly cautious gatekeeping can itself concentrate power in whoever controls the gate — a tension this book returns to directly in 10_technology/01_technology_as_force_multiplier.md and in Part 14's failure-mode simulations. This chapter states the problem; it does not claim to have solved the governance design question that follows from it.