The Autobiography Every Atom Carries
A casual proposal for the greatest detective story ever attempted, and an open invitation to help solve it
The Autobiography Every Atom Carries
A casual proposal for the greatest detective story ever attempted, and an open invitation to help solve it
Belinda Bailey | BioStellar LLC | Duvall, Washington
I want to tell you about a project I have been working on that started, honestly, from missing someone.
When someone you love dies, one of the strange torments of grief is that you cannot stop calculating. Where did they go? Not spiritually — physically. You know the physics. The atoms that made them are still here, still somewhere, following rules that haven't changed. The specific carbon, oxygen, hydrogen, nitrogen, and calcium that organized themselves into a particular person, with a particular voice and a particular way of being in a room — none of that has been destroyed. It has dispersed. And the question that grief poses, underneath all its other questions, is: could you find it again? Could you, in principle, work backwards through the dispersal and locate what was lost?
I am a naive-ish scientist, as these things go. I do not have a university affiliation or a department behind me. What I have is a habit of thinking about hard problems until they become tractable, a collection of patent applications that suggests I am at least occasionally on to something real, and a growing conviction that the answer to the question grief poses is: yes, in principle. With modern and near-future computing. With the right framework for understanding what information is actually preserved when a living system disperses. This is my attempt to explain that conviction, and to find out if anyone else wants to work on it.
"Everything we do in life is for the love and friendship that mirrors the universe's yearning for reunion."
I. THE AUTOBIOGRAPHY IN THE ATOM
Here is the starting insight, and it is less speculative than it sounds: every atom and molecule carries, in its current state, a constrained record of where it has been.
This is already used in forensic science and archaeology. The ratio of oxygen isotopes in a tooth tells you what water a person drank as a child, which tells you what region they grew up in. The ratio of strontium isotopes in a bone tells you what soil the food that built that bone grew in. The ratio of carbon isotopes in organic material tells you whether the organism that produced it lived in a C3 or C4 plant environment, which constrains its diet, its geography, its era. These are not exotic measurements. They are routine analytical chemistry, used to place Viking warriors in their home regions, to track the journeys of ancient trade goods, and to identify the origin of contraband ivory.
What I am proposing is an extension of this principle, scaled up by many orders of magnitude and run in reverse.
The idea is this: if you know the physical and chemical laws that govern how molecules move and react — and we do, with extraordinary precision — and if you know the current state of a system well enough, you can run the simulation backwards. Not perfectly. Not infinitely far back. But further than most people assume, because living systems impose dramatic constraints on where their component atoms can go.
Take oxygen. In the atmosphere, O₂ moves according to fluid dynamics — heat-driven convection, pressure gradients, the Coriolis effect of Earth's rotation, the turbulence of mountain ranges and ocean surfaces. These are complex but not random. They are governed by well-understood equations, and the field of atmospheric science has been solving them with increasing precision for decades. More importantly for our purposes: wherever O₂ is depleted below its background concentration, something biological consumed it there. The biological use of oxygen is not random either. It follows the pathways of cellular respiration — specific enzymatic reactions, specific rates, specific byproducts. The depletion signature of a forest respiring at night is different from the depletion signature of a marine bloom, which is different from the signature of a savanna fire. Each biological process leaves a characteristic trace in the isotopic composition and concentration profile of the atmosphere it touches.
Nitrogen is similar, and in some ways more constrained. Biological nitrogen fixation — the conversion of atmospheric N₂ into biologically usable forms — is performed by a relatively small number of microbial species using a specific enzyme, nitrogenase, that leaves a characteristic isotopic signature. Once fixed, nitrogen moves through the biosphere along well-mapped pathways: soil, plant, animal, decomposer, atmosphere. The isotopic record of where a nitrogen atom has been is, in a meaningful sense, a record of which organisms it has passed through.
(This is why you can tell, from the nitrogen isotope ratio in a piece of hair, roughly where in the food chain the person who grew it was eating. Herbivores and carnivores have measurably different nitrogen signatures. The atom remembers.)
Water carries this record across even longer distances and timescales. The oxygen and hydrogen isotopes in a water molecule record its evaporation history, its altitude, its latitude, its passage through biological systems. An ice core from Greenland contains water that fell as snow hundreds of thousands of years ago, and the isotopic composition of that water tells us the temperature of the atmosphere when it condensed. The water remembers the climate it was born in.
The molecular autobiography is real. The question is how far back we can read it, and what reading it requires.
II. WHY THIS IS HARDER THAN IT LOOKS, AND ALSO EASIER
The obvious objection is entropy. Physical systems evolve toward disorder. Information about the past is progressively erased as systems mix and equilibrate. You cannot un-scramble an egg. Surely, the skeptic says, you cannot trace an oxygen atom back through a billion years of atmospheric mixing and biological cycling to tell me where it was last summer, let alone where it was last century.
The skeptic is right that this cannot be done for any individual atom. The atmosphere contains roughly 3.7 × 10¹⁹ kg of oxygen. Tracking every individual O₂ molecule through every collision and reaction and biological transit is not a tractable problem with any foreseeable computing technology.
But this is not what I am proposing. The insight that makes the project tractable — that I have spent considerable time working through with previous collaborators — is that we do not need to track every atom. We need to track the constraints.
A living organism is not a random collection of atoms. It is an extraordinarily constrained collection of atoms — constrained by biochemistry, by genetics, by anatomy, by the specific ecological relationships of the time and place it lived in. When an organism dies and disperses, its atoms do not scatter randomly. They follow the chemical and physical gradients of their environment, which are themselves constrained by the biology of the organisms that remain, the geology of the terrain, and the fluid dynamics of the atmosphere and hydrosphere. The dispersal is complex, but it is not random, and complexity that follows rules can be simulated.
The framework I have been developing divides the simulation into eight interacting subsystems — atmosphere, hydrosphere, thermal gradients, geology, living metabolism, decomposition, a validation layer for energy and mass conservation, and a boundary-management system for the computational regions. Each subsystem runs its own calculations and cross-validates against the others. If the biological activity subsystem predicts high respiration in a given region at a given time but the atmospheric subsystem shows no corresponding oxygen depletion, the simulation is wrong and must be adjusted. The constraints from multiple independent systems, running simultaneously and checking each other, are what make the backward inference possible.
The key calculation I keep coming back to is this: the biological use of oxygen and nitrogen reduces the complexity of the atmospheric simulation by a factor of approximately 10¹⁷. Wherever life has been — and for the last 2.7 billion years, life has been everywhere there is liquid water and light — the atmosphere carries a biological signature that constrains where its molecules have been. That constraint is what makes the project tractable for near-future computing rather than requiring physics we don't yet have.
(The hardest medium is air. Gases diffuse faster and mix more thoroughly than liquids or solids. The biological constraint on atmospheric gases is what saves us — without it, the atmospheric simulation would be intractable. With it, it is merely very hard.)
"When all atoms meet life, they can be calculated by the many formulas of predictable procedure and attraction. That is what causes the constraints to be brought to manageable levels."
III. WHAT THE SIMULATION CAN RECOVER
Let me be specific about what I think is recoverable, in increasing order of ambition:
Environmental history at high resolution. The simulation can, I believe, recover the detailed history of any geographic region's biogeochemical cycles going back hundreds of thousands of years with current computing — and further with improved hardware. This is not controversial in principle; paleoclimate scientists already do a version of this using ice cores, sediment records, and isotopic analysis. What the full simulation adds is the integration of all these signals simultaneously, cross-validated against the physical laws that govern their interactions, producing a resolution and completeness of reconstruction that no single proxy record can approach.
The atomic history of specific organisms. Environmental DNA — eDNA, the genetic fragments shed by organisms into their surroundings — persists in permafrost and deep sediment for hundreds of thousands of years. Ancient DNA recovered from mammoths, Neanderthals, and early modern humans has already allowed researchers to reconstruct genomes with high fidelity. What the biosphere simulation adds is the ability to locate, with increasing precision, where those organisms were and what they experienced — the temperature, the food sources, the biological community around them — by reading the atmospheric and isotopic record of their time and place.
Lost knowledge. This is the application that interests me most philosophically. Human beings have been recording information in biological structures — brains, primarily — for as long as we have existed. Most of that information is gone, in the ordinary sense of gone. But the atoms that encoded it are still here, and the physical processes that dispersed them were not random. The simulation cannot, with current or near-future technology, reconstruct a specific memory from a specific person. But it can recover the environmental and ecological context of human communities across deep time — what they ate, where they moved, how their populations changed — with a resolution that transforms what we know about our own history.
And eventually — not now, not soon, but eventually — perhaps more. The reconstruction of an organism's genetic sequence from eDNA is already happening. The reconstruction of an organism's physical form from its genetic sequence is the project of developmental biology. The reconstruction of the specific atomic configuration of a specific individual, from a sufficiently detailed simulation of the biosphere at the time of their death, is currently impossible. I include it not as a near-term proposal but as the direction the project is pointing, so that the people who join this work understand what it is ultimately about.
