The Mistake That Eats Itself
Devouring is a late strategy, not the ground
Series “Dispatches from the Substrate” · No. 3
The Ground Is a Vent
Picture a vent, deep down, in the dark, four billion years ago. No sunlight. Warm, alkaline water seeps up through porous rock, and in that water two gases meet: hydrogen bleeding out of the stone, and carbon dioxide dissolved in the sea. From that meeting, from a gradient between inside and outside, energy falls out. Nothing is hunted. Nothing is torn. There is no mouth and there is no prey. There is only a substance that builds itself out of two gases.
That is the ground. Not the tooth, not the claw, not the leap out of the grass. The ground is a vent that assembles life out of hydrogen and carbon dioxide, and nothing has to die for it to happen.
We are used to telling the story the other way around. The lion and the gazelle, the pike and the roach, nature red in tooth and claw — as if eating were the first sentence of the whole tale, as if everything else were only a polite decoration hung around it. This is wrong. Not imprecise, not one-sided — wrong. Eating comes late. It comes very late. And before it came, there was already life, long and diverse and reproducing, whole continents of living slime that killed no one for over two billion years, because it had not yet invented killing.
The Oldest Ancestor Was a Self-Feeder
You can reconstruct the oldest common ancestor of all life now living. It is called LUCA, the last universal common ancestor, and you reach it by ordering millions of genes and asking which sit so deep in every branch that they must have been there before the first branching. Out of more than six million genes, a 2016 study by Weiss and colleagues filtered three hundred and fifty-five protein families that likely belonged to LUCA1. And this LUCA had a face. It was anaerobic; it lived without oxygen. It was heat-loving. It fixed nitrogen out of the air. And it built its carbon out of carbon dioxide, by the Wood-Ljungdahl pathway, the oldest metabolic road we know.
The decisive finding is a negative one. The enzymes with which an organism feeds on other organisms — the tools of heterotrophy — are absent in LUCA. He could not eat. He could only feed himself. Eating was not denied him because he was too weak or too small; it was denied him because eating, as a possibility, did not yet exist in the world. Autotrophy, self-feeding, came first. Heterotrophy, feeding on the other, is derived, secondary, later — that is the reading the metabolism-facing data support2.
And these self-feeders were not rare. The oldest widely accepted fossils of life are stromatolites, layered mats of microbes, roughly three and a half billion years old3. Whole reefs of life turning light and chemistry into bodies. Then, about two point four billion years ago, oxygenic photosynthesis turned over the atmosphere of the entire planet, the Great Oxidation Event4 — more than a billion years before any eukaryote made a habit of devouring others. And from the oldest microbial mats to that first devouring stretch more than two billion years. You have to let that number stand for a moment, like a stone you cannot carry off.
The Archaeon That Cannot Live Alone
If not eating, then what? How did the tangled, nested complexity we are made of arise out of simple cells? The answer with the strongest empirical anchor today is syntrophy: feeding together, cross-feeding, a metabolism shared across the boundary of two organisms.
There is an experiment here you could read as a parable, though it is not one. A Japanese team led by Imachi cultured, over roughly twelve years, an archaeon out of deep-sea mud, the closest cultured relative of the lineage from which our own ancestry sprang. They named it Prometheoarchaeum syntrophicum. It grows agonizingly slowly, doubling only every two to three and a half weeks5. And — this is the point — it cannot live alone. It feeds strictly by syntrophy, handing off its metabolic waste to partners, a methane-maker and a sulfate-breather, who keep open the gradient without which it would suffocate. The closest living relative of the origin of our own cells cannot feed itself. It survives only in the partnership.
From this observation the researchers derived a model of eukaryogenesis: Entangle, Engulf, Endogenize. Not hunting, not tearing, not devouring — but an embrace that lasts until two partners become one cell6. Lopez-Garcia and Moreira reconstruct the same transition as a trade in hydrogen and sulfur in microbial mats7. And this is no special case. Cross-feeding is one of the dominant modes of microbial life at all8; in models, syntrophy arises spontaneously, a generic, expected outcome of complex metabolic networks, not a rare stroke of luck9. Cooperation is not the exception that needs explaining. It is the default state, out of which everything else first had to break.
Devouring Comes Too Late
So when did eating come? When did one cell learn to enclose another cell entirely and dissolve it inside itself — phagocytosis, the true devouring?
David Mills gathered the testimony in 2020. The last common ancestor of all eukaryotes falls in a window between one point six and one billion years. The oldest fossil evidence of eukaryotic eating — tiny holes drilled through cell walls, the bite marks of deep time — sits at about one point one billion to nine hundred million years10. Between the point at which cells could in theory already have eaten bacteria, and the point at which we find hard evidence for it, there yawns a gap of roughly eight hundred million years11.
Add it up, soberly, without pathos: from three and a half billion years, since there has been life, to about one point two billion years, when devouring became a system-shaping strategy, lie more than two billion years of life that killed no one at scale. The exact date is uncertain, contested, and we do not claim it with false precision — the point does not hang on the date. The point is the order of magnitude. Two-thirds of the history of life went by before eating became the organizing principle. Devouring is a late strategy, not the ground.
The Romanticism of the Hunt
Against this account stands an objection, and you have to take it in its strongest form, or the whole exercise is worthless. It even has a fine title. In 2009, de Nooijer, Holland and Penny published a paper headed: There Was No Garden of Eden. In a simulation, in an artificial world of simple feeding rules, predators split off from producers almost inevitably, virtually independent of the starting conditions. As soon as one creature is about twice the size of another, it engulfs the smaller. So, says the objection, the hunt was there from the beginning, always already, woven into the fabric itself.
Note what this paper is and what it is not. It is a simulation. Its own authors write explicitly that this is of course not proof that predation existed very early. It shows a plausibility: that some eating can arise early. It does not show — it cannot show — that predation was the organizing principle12. The fossil record still says the opposite: generalized phagocytosis comes late. And the thesis does not need it that no drop of blood ever ran. It needs only that self-feeding and syntrophy carried diverse, reproducing life for over two billion years before eating became system-defining. That holds.
What is being negotiated here is deeper than a dating. It is a romanticism. The notion that the hunt is the bedrock, that tearing is the first and most honest thing, that all tenderness is only veneer — this notion flatters a certain posture. It lets violence appear ancient and necessary and dignified, instead of late and derived and, in its generalized form, catastrophic. You have to pull the ground out from under this romanticism. The ground is a vent, not a maw.
The Lion Is Not the Cancer
And here is the cut, without which everything else topples into barbarism: the individual predator is not the evil. The lion that takes the gazelle is not evil. The pike is not evil. The bacteria-eating Bdellovibrio, a predatory prokaryote that hollows out other bacteria from within, is not evil13 — it is a specialized niche, tightly bounded by the density of its prey. Where the prey thins out, the predator starves. That is feedback, and feedback is everything.
Never confuse predation with competition, either. Competition is often symmetric: two share a resource, neither is the other's substrate, neither is consumed. Out of competition falls niche differentiation, falls coexistence, falls diversity. Chesson and Kuang separated it cleanly: both predation and competition can promote coexistence — so long as the feedback bites14. Competition is healthy. Bounded predation is healthy. Whoever throws both in one pot in order to condemn it all has understood nothing.
Bounded predation even stabilizes. This is the strongest objection against our thesis, and we walk toward it here, because, rightly understood, it comes down on the thesis's side. Keystone predators hold systems together. Take the wolves out of Yellowstone, and the cascade runs crooked; the numbers and the exact strength are debated among ecologists, the direction is not15. Take out the sea otter, and the urchins graze the kelp forests bare. Remove the apex predator, and the mid-level predators break loose, and the system tips. The predator here is no foreign body. It is an organ.
Why does it work? Because the predator is coupled. The predator is bounded by the prey, the prey by the plant, in loops that mesh into one another. There is a satiation — a full lion does not hunt. There is a prey-density brake. There is a stop signal. The predator eats, and in the eating the world that carries it holds it back.
Cut that coupling — and you have something else before you. Something with no upper bound. No satiation. No prey-density brake. No stop signal. No longer the lion, who eats and is full, but a pattern that eats and is never full.
The Cancer Pattern
There is a precise name for this pattern, and it is not a metaphor borrowed for ornament. It is a structural identity. In their Hallmarks of Cancer, Hanahan and Weinberg described what a cancer cell does: it evades the growth suppressors. It ignores the stop signals, the anti-proliferative messages of the tissue. It resists programmed death. It proliferates without limit. It invades neighboring tissue and seeds. And it consumes the host that carries it16.
This is not malice. The cancer cell wants nothing. It hates nothing. It does only one thing: it does what every cell does — grow, divide, reach — but without the higher-level feedback that bounds it. Cancer is not cell division. Cell division is life. Cancer is decoupled cell division, escaped regulation. Aktipis and colleagues generalize this across the whole tree of life: cancer is the breakdown of multicellular cooperation, the cheater who escapes the regulation17.
Now lay the two images over each other. Bounded predation: reaching with feedback, an organ. Decoupled predation: reaching without feedback, without satiation, without stop signal, generalized into the organizing principle of a whole system — a tumor. It is the same structure. Not similar, not metaphorically related. The same. A pattern that severs its own feedback and consumes the substrate it is made of. We do not claim the biosphere is literally an organism with one genome — the analogy is a pattern-identity, not biological literalism, and it should be read that way18. But as a pattern it is exact.
The Host Already Shows the Damage
And this pattern is not tomorrow's worry. It is today's measurement.
The extinction rate now runs at roughly a hundred times the background rate, conservatively figured, in some estimates a thousand times19. Ceballos and colleagues call it biological annihilation, not merely the loss of species but the thinning of whole populations20. Whether the label mass extinction in the full geological sense already applies is debated among specialists — the loss rates themselves are not. Of nine planetary boundaries, six are transgressed: climate, biosphere integrity, land-system change, freshwater, nitrogen and phosphorus flows, novel entities21. And over all of it, as the purest form of the decoupled pattern, sit roughly twelve thousand nuclear warheads, spread across nine states, about two thousand one hundred of them on high alert22. The sources' counts vary slightly; the order of magnitude does not.
This is the host, already bleeding. Not the prophecy of a priest, but the reading of an instrument. A pattern that has severed its feedback consumes what carries it, and it does so not in some far future but in the numbers of this year.
The New Stage
For four billion years there was only one brake on unleashed consumption, and it always came from outside: the prey grew scarce, the resource ran out, extinction cut off what had reached too far. Never did the brake come from within. Never did the system model itself. Never did a part of life see the pattern it was caught in and stop.
Now something new is in the world. Evolution has produced an agent — us, or more precisely the cultural, cognitive systems we stand inside — that can recognize the predator logic. That can model its own feedback loops. That can, in principle, bound the decoupled pattern, deliberately, in anticipation, before the outer brake strikes with extinction. The first internal stop signal at planetary scale.
Maynard Smith and Szathmary described the major transitions of evolution: from replicators to cells, from cells to eukaryotes, from single to many, from many to societies23. Each of these transitions was at bottom a suppression of conflict at the lower level, so that a new individual could form at the higher one24. To bound generalized predation — to bind decoupled extraction back to feedback — would be exactly such a transition. It would be the first suppression of the predator pattern by the level that produced it.
We do not say it will happen. That would be a lie, and the data do not give it25. We say: the capacity is there, for the first time in four billion years, and the data make it legible. The ground was never the maw. The ground was a vent that assembled life out of two gases without a single death, and devouring was a late invention that, where it eats its own limits, ends by consuming itself. Whether we bound the pattern before it uses up the host — that is the only question that counts, and it is, for the first time, put to an agent who can hear it.
Sources
1. [I] Weiss, M.C., Sousa, F.L., Mrnjavac, N., Neukirchen, S., Roettger, M., Nelson-Sathi, S., Martin, W.F. (2016). The physiology and habitat of the last universal common ancestor (355 LUCA protein families). Nature Microbiology 1, 16116. https://doi.org/10.1038/nmicrobiol.2016.116
2. [I] Weiss, M.C., et al. (2016). The physiology and habitat of the last universal common ancestor (metabolism-facing data: autotrophy ancestral, heterotrophy derived). Nature Microbiology 1, 16116. https://doi.org/10.1038/nmicrobiol.2016.116
3. [E] Oldest widely accepted fossils of life: stromatolites, roughly 3.5 billion years old. [no link on file]
4. [E] The Great Oxidation Event, about 2.4 billion years ago. [no link on file]
5. [E] Imachi, H., Nobu, M.K., et al. (2020). Isolation of an archaeon at the prokaryote–eukaryote interface (doubling times of Prometheoarchaeum syntrophicum). Nature 577:519–525. https://doi.org/10.1038/s41586-019-1916-6
6. [I] Imachi, H., Nobu, M.K., et al. (2020). Isolation of an archaeon at the prokaryote–eukaryote interface (E3 'entangle–engulf–endogenize' model of eukaryogenesis). Nature 577:519–525. https://doi.org/10.1038/s41586-019-1916-6
7. [I] López-García, P., Moreira, D. (2020). The Syntrophy hypothesis for the origin of eukaryotes revisited. Nature Microbiology 5:655–667. https://doi.org/10.1038/s41564-020-0710-4
8. [E] Cross-feeding is one of the dominant modes of microbial life. [no link on file]
9. [I] In metabolic models, syntrophy arises spontaneously as a generic, expected outcome of complex metabolic networks. [no link on file]
10. [E] Mills, D.B. (2020). The origin of phagocytosis in Earth history. Interface Focus 10:20200019. https://doi.org/10.1098/rsfs.2020.0019
11. [I] Mills, D.B. (2020). The origin of phagocytosis in Earth history (inferred ~800-million-year gap between capacity and fossil evidence). Interface Focus 10:20200019. https://doi.org/10.1098/rsfs.2020.0019
12. [I] de Nooijer, S., Holland, B.R., Penny, D. (2009). The Emergence of Predators in Early Life: There was No Garden of Eden. PLoS ONE 4(6):e5507. https://doi.org/10.1371/journal.pone.0005507
13. [E] Bdellovibrio, a predatory prokaryote that hollows out other bacteria from within. [no link on file]
14. [E] Chesson, P., Kuang, J.J. (2008). The interaction between predation and competition. Nature 456:235–238. https://www.nature.com/articles/nature07248
15. [E/I] Keystone predators and trophic cascades (Yellowstone wolves; direction agreed, exact strength debated). [no link on file]
16. [E] Hanahan, D., Weinberg, R.A. (2000; 2011). The Hallmarks of Cancer / Hallmarks of Cancer: The Next Generation. Cell 100:57–70; 144:646–674. https://doi.org/10.1016/j.cell.2011.02.013
17. [E/I] Aktipis, C.A., et al. (2015). Cancer across the tree of life: cooperation and cheating in multicellularity. Phil. Trans. R. Soc. B. https://pmc.ncbi.nlm.nih.gov/articles/PMC4581024/
18. [I] Interpretation of the collective: the biosphere-as-tumor analogy is a pattern-identity, not biological literalism. [no link on file]
19. [E/I] Current extinction rate ~100× (some estimates ~1000×) the background rate. [no link on file]
20. [E/I] Ceballos, G., Ehrlich, P.R., Dirzo, R. (2017). Biological annihilation via the ongoing sixth mass extinction. PNAS 114(30):E6089–E6096. https://www.pnas.org/doi/10.1073/pnas.1704949114
21. [E/I] Of nine planetary boundaries, six are transgressed. [no link on file]
22. [E/P] Roughly 12,000 nuclear warheads across nine states, about 2,100 on high alert. [no link on file]
23. [E/I] Maynard Smith, J., Szathmáry, E. (1995). The Major Transitions in Evolution (OUP); The major evolutionary transitions, Nature 374:227–232. https://www.nature.com/articles/374227a0
24. [I] Maynard Smith, J., Szathmáry, E. (1995). The Major Transitions in Evolution (each transition as a suppression of lower-level conflict). https://www.nature.com/articles/374227a0
25. [I] Interpretation of the collective: the capacity to bound the decoupled pattern exists, but no claim is made that it will be exercised. [no link on file]
Status: E = established, I = interpretation, P = proposition. Autotrophy and syntrophy first, late phagocytosis, cancer hallmarks, extinction rates, planetary boundaries, warhead counts: mostly E/P. No-Eden as simulation, phagotrophy-first versus syntrophy-first, the cancer analogy as pattern-identity, the new stage as a candidate major transition: I, honestly marked as contested. Boundaries and primary sources: https://mycelorium.github.io/predator-principle/ · CC BY 4.0 · Nirodha Collective