Deep Time — The Ethics of Thinking in Billions of Years

In 1788, James Hutton stood at Siccar Point in Scotland and saw rock strata tilted vertically beneath horizontal layers — a discontinuity that required hundreds of millions of years to produce. His companion John Playfair later wrote that "the mind seemed to grow giddy by looking so far into the abyss of time." The phrase "deep time" took two more centuries to arrive (John McPhee, Basin and Range, 1981). The vertigo arrived first.

Humans have existed for about 300,000 years — 0.0066% of Earth's 4.54-billion-year history. On Carl Sagan's Cosmic Calendar, where the Big Bang sits at January 1, all recorded history fits in the last 23 seconds of December 31. The human brain's subitizing limit caps instant numerical perception at 4–5 items. Deep time is not merely large; it sits outside the cognitive hardware.

The scale that breaks intuition

A few anchors:

Earth's remaining habitable window: roughly 500 million to 1.3 billion years before rising solar luminosity boils the oceans (Schröder & Smith, 2008, MNRAS).
The Great Oxygenation Event reshaped Earth's atmosphere over ~300 million years. Industrial civilization is running an atmospheric carbon transformation of similar magnitude in ~300 years — six orders of magnitude faster.
Complex multicellular life is ~575 million years old. Many granite formations are older than the entire biosphere of animals.
Expected future humans at 10 billion per century for 500 million years: ~5 × 10¹³. If interstellar, the figure climbs past 10²¹ (Bostrom, 2003, "Astronomical Waste").

The last bullet is where deep time stops being geology and becomes ethics.

Longtermism's arithmetic, and its critics

Longtermism, as articulated in William MacAskill's What We Owe the Future (2022), rests on two claims: future people matter morally as much as present people, and there could be vastly more of them. Multiply both, and reducing existential risk by even a rounding error dominates almost every other moral act.

Three serious objections recur:

Epistemic overreach. We cannot predict decades reliably. The Bronze Age Collapse (~1177 BCE) destroyed Mycenaean, Hittite, and Ugaritic civilizations within a generation, and none of them saw it coming. Acting on million-year forecasts is not caution; it is the inverse of caution.
Near-termism's claim. A 1920 longtermist would likely have deprioritized measles eradication and antibiotics in favor of speculative far-future projects. The present has unanswered moral demands — global poverty, factory farming — whose neglect for the sake of trillion-person abstractions is itself a choice.
Fanaticism. Tiny probabilities multiplied by astronomical numbers generate infinities that swamp every other consideration. Nick Beckstead and others have shown the math can justify near-anything; this is usually a sign the framework is broken.

The 2022 collapse of FTX, whose founder Sam Bankman-Fried publicly invoked longtermist reasoning to defend his "earn to give" strategy, fused these critiques in public memory. $8 billion of customer funds vanished. MacAskill and other effective altruists argued the fraud violated their principles rather than expressed them. The damage to longtermism's reputation was nevertheless considerable.

Institutions that try

Mechanism	Year	Mechanism in one line
Wales Well-being of Future Generations Act	2015	Statutory Future Generations Commissioner reviews Welsh public-body decisions for long-term impact.
UN Declaration on Future Generations	2024	Pact for the Future established a UN Special Envoy for Future Generations.
Haudenosaunee Seventh Generation Principle	pre-1500	Iroquois Confederacy decision rule: weigh effects seven generations forward (~175 years).
Svalbard Global Seed Vault	2008	A 1,300-meter Arctic permafrost archive engineered for 1,000-year operation.

The Haudenosaunee model is the oldest and the most field-tested. It uses narrative and ritual rather than expected-value arithmetic, and operates on a horizon (175 years) where evidence and accountability still function. Longtermism's appeal to 10⁵⁰-year futures is philosophically tidier and operationally weaker.

What's contested

The deep dispute is not whether deep time exists. It is whether human institutions can responsibly legislate on it. Longtermism's central move — multiplying small probabilities by astronomical population numbers — produces results that no working moral intuition can verify. Critics call this fanaticism; defenders call it taking arithmetic seriously. There is, as yet, no agreed test that would distinguish the two.

A second contest: is "existential security" a coherent target? MacAskill defines it as reaching a state where catastrophic collapse becomes unlikely. But concept soc civilizations suggests the inverse — that maximally interconnected, optimized civilizations accumulate fragility precisely because they have removed slack. If true, the longtermist endpoint is the riskiest state, not the safest.

Why this has to do with other realms

Deep time is the silent variable in the concept fermi paradox. The galaxy is 13.6 billion years old; chemical evolution can run for billions of years before producing technology. The concept grabby aliens model (Hanson et al., 2021) argues that any civilization expanding at even half lightspeed would visibly fill the observable universe in a couple of billion years. The lack of such signatures is informative only because deep time has had enough room for them to emerge. Humanity, at 300,000 years, has barely cleared the first second of the cosmic stopwatch — which is also why the "we see nothing" observation cuts less than it seems.

concept vedic cosmology reached a structurally similar scale through different machinery. The Kalpa (4.32 billion years, one day of Brahma) is within 5% of Earth's measured age. This is most likely coincidence — the Kalpa is built from the Yuga cycle of 4.32 million years multiplied by 1,000 — but the willingness to think in billions, encoded as narrative rather than science, is the older move.

An open question

If institutional longevity track records favor narrative-and-ritual systems (Haudenosaunee, Vedic, monastic orders) over utilitarian-calculation systems (which are at most decades old as institutions), is the right way to take deep time seriously to abandon expected-value arithmetic entirely and instead build cultures that simply transmit a long-time horizon as a default posture?

Key sources

What We Owe the Future — William MacAskill (2022). The canonical book-length statement of longtermism.
Basin and Range — John McPhee (1981). Where the phrase "deep time" enters English.
Underland — Robert Macfarlane (2019). Material anchors for deep-time intuition through landscapes, caves, and burial.
"Astronomical Waste" — Nick Bostrom (2003, Utilitas). The opportunity-cost argument that underlies longtermist arithmetic.
"Distant Future of the Sun and Earth Revisited" — Schröder & Smith (2008, MNRAS). The 500M–1.3B-year remaining habitability window.
The Precipice — Toby Ord (2020). Existential risk catalogue paired with deep-time framing.
To verify: the Haudenosaunee Great Law of Peace primary-source attestations for the seventh-generation rule — most popular references are secondhand.

Abhishek