The Anthropocene was rejected as an epoch and ruled an event — which is the stratigraphic statement of WP67 §5, reached independently from cores and varves · the four tests turn out to be a preservation filter · and the Great Oxidation Event is the type specimen for a distributed deployment that failed all four
WP66 §3 proposed four tests for a distributed intervention: local, reversible, attributable, self-benefiting-or-oracle-scored. WP67 scored microplastic ice-nucleating particles against them and got four failures on a live global deployment. This paper argues the tests are doing something more than filtering candidates: they are describing which human activities become geology.
Two results. First, the formal stratigraphic ruling — the Anthropocene rejected as an epoch in 2024 and characterised instead as a diachronous event with no single globally synchronous boundary — is the same structural claim WP67 §5 derived from WP39's Theorem T2 by a completely different route [documented]. An event has no crossing; a pointwise perturbation has no schedulable moment relative to the fold. Two formalisms, one conclusion.
Second, and the paper's own contribution: failing tests 1 and 2 is physically necessary for a stratigraphic marker to exist at all, while tests 3 and 4 govern whether the resulting deposit is bounded or unbounded [modeled]. The type specimen is not human. The Great Oxidation Event — cyanobacteria releasing a metabolic waste product that reorganised the atmosphere, reduced the biosphere by more than 80%, and triggered a ~230-million-year glaciation — fails all four tests, and is the precedent the stratigraphers themselves chose when arguing the Anthropocene is an event.
The term was proposed informally by Crutzen and Stoermer in 2000 and pressed into the literature by Crutzen's 2002 Nature note. The Anthropocene Working Group formed under the Subcommission on Quaternary Stratigraphy, argued through Waters et al. (2016) that the interval is functionally and stratigraphically distinct from the Holocene, voted in 2019 to pursue formalisation, and in 2023 selected the varved sediments of Crawford Lake, Ontario as the candidate GSSP with a base in the early 1950s marked by the plutonium fallout spike.
In 2024 the proposal was rejected. The governing bodies declined to ratify an Anthropocene Epoch, and the position that has consolidated since — argued most directly by Gibbard, Walker, Head and colleagues — is that the Anthropocene is best understood as a geological event.
A formal chronostratigraphic unit requires a Global boundary Stratotype Section and Point: one location, one horizon, one globally simultaneous boundary. As Gibbard et al. put it, such boundaries "cannot be diachronous" by definition.
A geological event carries the opposite property. Events may be "time-transgressive, multi-temporal and spatially variable," and the boundaries marking their onset and termination "may be diachronous; hence time-transgression is inbuilt within the event–stratigraphy paradigm." Human transformation began at different times in different places, across millennia. There is no single moment to spike.
WP67 §5 applied WP39's Theorem T2 to microplastic ice-nucleating loading and concluded that it is not a gate: an in-layer pointwise perturbation commutes with the 0/1 transport gate, so it produces no order-dependence and cannot be scheduled relative to the fold. The paper insisted this was the worse finding rather than the better one, because the alternative to "gate" is not "no effect" — it is an uncontrolled perturbation with no crossing to place K before.
Set the two statements side by side.
| Formalism | Object | Conclusion |
|---|---|---|
| Operator algebra (WP39 T2 → WP67 §5) | In-layer microplastic INP loading | Pointwise; commutes with the gate; no schedulable crossing |
| Event stratigraphy (Gibbard et al. → IUGS 2024) | The anthropogenic signal in the sediment record | Diachronous; admits no GSSP; no globally synchronous boundary |
These are the same claim in two vocabularies. A fold, in WP44's sense, is a structural discontinuity at a single crossing of the fold locus — which is precisely what a GSSP formalises in rock. (Whether the locus is fixed or itself moves under intervention is WP44 §3's question, not this one's: a GSSP records a crossing, and a crossing is a crossing whether the wall came to you or you walked into it.) To say the Anthropocene admits no GSSP is to say the anthropogenic signal is not a fold. The stratigraphers got there from varves and cores; this series got there from a three-layer box model.
Two independent formalisms converging is weak evidence individually and reasonable evidence jointly, since the failure modes are unrelated. It is stated here as convergence [modeled], not as confirmation.
If the anthropogenic signal is an event rather than an epoch, the interesting question shifts. Not when did it start, but which activities left a record at all — and there the four tests turn out to be doing unexpected work.
For a signal to serve as a stratigraphic marker it must satisfy two physical requirements. It must be globally distributed and near-synchronous, which means it must have entered a shared medium — atmosphere or ocean — and travelled. And it must persist in the sediment column over geological time without degrading or being withdrawn.
Those are exactly the negations of tests 1 and 2.
This half of the thesis is analytic once written down, and honesty requires saying so: it is not an empirical discovery that non-local, irreversible things fossilise. What it buys is that the four tests stop being merely prudential. Tests 1 and 2 are not advice about what a crowd should be handed — they are a statement of the physical conditions under which an act leaves permanent planetary record. Something local and reversible does not enter the strata. It is absorbed.
The other two operate on a different axis, and here the content is empirical rather than definitional.
Attributability (test 3) creates an error signal. If a perturbation can be traced to a mechanism and a source, correction becomes possible, and the deposit is bounded by the time between detection and control. Self-benefit (test 4) internalises the cost. If the deployer bears the consequence, deployment is self-limiting without any institution at all. Fail both and accumulation has no natural ceiling — the deposit thickens until the activity stops for unrelated reasons.
Tests 1 and 2 determine whether you leave a layer. Tests 3 and 4 determine how thick it is. The Anthropocene, on this reading, is not a list of things humanity did. It is the subset of what humanity did that failed the first two tests — deposited to a depth set by how badly it failed the second two.
Scoring the proposed markers against the tests. ✗ marks a failed test — which, for tests 1 and 2, is what qualifies the marker in the first place.
| Proposed marker | 1 Local | 2 Rev. | 3 Attr. | 4 Benefit | Note |
|---|---|---|---|---|---|
| Plutonium fallout (bomb spike, early 1950s) | ✗ | ✗ | ✓ | ✗ | The proposed GSSP signal. Attributable to a small number of states — and testing was in fact curtailed by treaty. A bounded spike, not a step |
| Spheroidal carbonaceous particles (fly ash) | ✗ | ✗ | ✗ | ✗ | Argued as the defining marker; unambiguously anthropogenic, no natural source |
| Microplastics (WP67) | ✗ | ✗ | ✗ | ✗ | Four failures. Predicts an unbounded step — and weathering increases the relevant activity over time |
| Broiler chicken skeletons | ✗ | ✗ | ✗ | ✓ | ~23 billion alive at any moment. Test 4 passes — the deployer eats the chicken — so the deposit tracks demand rather than running away |
| Concrete | ✗ | ✗ | ✓ | ✓ | The most abundant anthropogenic rock. Passes 3 and 4 — it is emplaced deliberately, where the builder wants it. A thick layer that is nonetheless owned |
| Reactive nitrogen (Haber–Bosch) | ✗ | ✗ | ✓ | ✗ | Benefit accrues to the applier, harm downstream — the classic externality shape |
| CFCs (§6, control case) | ✗ | ✗ | ✓ | ✗ | Passed test 3 decisively and was corrected. The prediction is a spike-and-decline |
Note the concrete row, which is the most instructive. Concrete fails 1 and 2 and therefore fossilises massively — but passes 3 and 4, because it is put deliberately where someone wants it and that person bears its cost. The filter does not say "thick layer = bad." It says the mechanism of accumulation differs, and only the 3-and-4 failures accumulate without anyone choosing to.
Gibbard and colleagues, arguing that the Anthropocene should be treated as an event, cite three precedents never formalised as epochs: the Great Ordovician Biodiversification Event, the Devonian continental forestation, and — first among them — the Great Oxidation Event.
It is worth being explicit about what they selected as the model for the human case.
Cyanobacteria evolved oxygenic photosynthesis and released free oxygen as a metabolic waste product. Score that against the four tests.
| Test | Oxygenic photosynthesis, ~2.4 Ga | Result |
|---|---|---|
| 1 · Local | Entered the global atmosphere and ocean; no organism's release stayed near it | FAIL |
| 2 · Reversible | Irreversible on any biologically relevant timescale; the atmosphere never went back | FAIL |
| 3 · Attributable | No error signal of any kind was available to any participant | FAIL |
| 4 · Self-benefiting | Oxygen was waste. It poisoned the anaerobic biosphere — including much of what surrounded the producers — and by destroying atmospheric methane it helped freeze the planet for over two hundred million years | FAIL |
The previous instance of a distributed, unattributable, irreversible waste product released into a shared planetary medium by an enormous number of independent agents, none of whom benefited from the release and none of whom could have known — ended in a mass extinction and a two-hundred-million-year glaciation, and left the most legible chemical signature in the Precambrian record. It is also, formally, an event and not an epoch, for exactly the diachroneity reason the Anthropocene now is. The stratigraphers chose a sharper analogy than they may have intended.
This is the deep form of what WP66 §3 was reaching for when it invoked Shirky. Here Comes Everybody is a book about what becomes possible when the cost of coordinating many independent agents collapses. The Great Oxidation Event is what happens when a great many independent agents act in a shared medium with the coordination cost at zero and the coordination itself at zero — no oracle, no attribution, no feedback. The result was not collective action. It was a byproduct that restructured the planet.
A filter that only ever returns "fails, catastrophically" is not a filter. The framework needs a case that passes something, and there is a good one.
Chlorofluorocarbons fail tests 1 and 2 completely — globally mixed, with atmospheric lifetimes of decades to a century. On the filter's account they should therefore leave a record, and they do. But CFCs pass test 3 decisively: Molina and Rowland identified the mechanism in 1974 and named the source class, and attribution was specific enough to act on.
The filter predicts that perturbations passing test 3 leave a spike-and-decline rather than a step, because attribution creates the error signal that bounds accumulation. CFCs are exactly that shape, and the shape is already visible in firn and ice records. Plutonium is the same shape for the same reason. Microplastics, failing 3 and 4, are predicted to be a step with no ceiling — which is testable against sediment cores now, and is the sharpest falsifiable claim in this paper [prospective].
Four caveats, in descending order of how much they cost the argument.
WP64 §2.4 argues that appeals to nature's self-regulation read teleology backwards off a survivorship artifact: configurations whose feedback failed are not available for observation. That argument applies to this paper with full force. The claim "what fails the tests is what fossilises" is assembled from a sample consisting entirely of things that fossilised. We cannot enumerate the local, reversible activities that left no record, because leaving no record is what they did. The filter is therefore at serious risk of being constructed by its own sample, and §3.1's honesty about analyticity is the only thing keeping it from circularity. Stated plainly: §3.1 is definitional and safe; §3.2 is empirical and exposed.
2 · The rejection was procedurally contested. The 2024 decision drew objections on process from within the working group and from the wider Anthropocene research community, and continues to be litigated in the literature. Nothing here should be read as a finding that human planetary impact is small or late — the dispute is about whether a formal chronostratigraphic unit is the right instrument, not about whether the signal exists. Several participants argue the epoch framing remains more useful for Earth-system purposes even if it lost the stratigraphic vote.
3 · Marker selection tracks detectability, not only preservation. The register in §4 lists what current instruments resolve well in recent sediment. A marker that preserves but is hard to measure is invisible to the list, which biases §4 toward things we happen to be good at counting.
4 · "Event" is not a demotion. The Great Oxidation Event is an event. So is the Great Ordovician Biodiversification. Informality in the timescale is orthogonal to magnitude, and §5's whole point is that the events chosen as precedent are among the largest transitions in the planet's history.
| Destination | Content | Status |
|---|---|---|
book6/wp68-… (this paper) | The four tests as preservation filter; GOE as type specimen | WRITTEN |
wp67-… §5 | Add pointer to §2 — the operator claim has an independent stratigraphic statement | DONE |
wp67-… §9 | Add the core-profile test: microplastic step vs plutonium spike, same core | DONE |
wp66-… §3 | Tests 1–2 are physical necessity conditions, not only prudential ones (§3.1) | DONE |
wp66-… §10 | Note the corollary: the recommended programme leaves no stratigraphic trace by construction | DONE |
wp64-the-recorder.html §2.4 | The survivorship argument is applied against a paper in this series for the first time (§8.1) | DONE |
wp44-catastrophe-manifold.html | Fold ↔ GSSP correspondence: a golden spike is a crossing; an event is not a fold | DONE |
index.html | WP68 entry; climate arc now WP39–45, WP66–68 | DONE |
The Anthropocene lost its vote and was ruled an event: real, global, legible in sediment, and admitting no single synchronous boundary. That is the same structure WP67 derived for microplastic loading from an operator theorem — not a fold, no crossing, nowhere to put K. Two formalisms with unrelated failure modes, one conclusion.
The four tests turn out to describe which activities become geology. Failing the first two is what it means to be capable of leaving a permanent record at all; failing the second two is what determines whether the deposit has a ceiling. On that reading the Anthropocene is not an inventory of human achievement. It is the residue of everything done in a shared medium, irreversibly, without attribution, at no cost to whoever did it.
And the precedent the stratigraphers reached for is not flattering. Cyanobacteria ran this experiment two and a half billion years ago with a waste product, no oracle, and no possibility of feedback. They failed all four tests, reduced the biosphere by more than four fifths, froze the planet for two hundred million years, and left the clearest chemical signature in the Precambrian.