Why We Still Don’t Know How Life Began
The Origin of Life: An Unsolved Problem. An Essay in the Philosophy of Science
How did life begin? After seventy years of origin-of-life research, from the RNA World hypothesis to hydrothermal vents, from warm ponds to metabolism-first models, this remains the deepest unanswered question in all of science. No hypothesis has produced an experimentally validated explanation for the transition from chemistry to biology. The textbook disposes of the question in a single confident paragraph. A federal court has ruled that one of the parties has no place in a biology classroom. And the only verdict that fits the evidence was known to Roman law and abolished by modern law.
In the second century of our era a young Roman named Aulus Gellius was appointed by the praetors to judge a suit for the repayment of a loan. The plaintiff, a man of unblemished reputation, said that he had lent the money. The defendant, a man of what Gellius calls a most shameful life, said that he had received nothing. There were no witnesses. There was no receipt. Experienced friends advised dismissing the claim, since without proof there can be no judgment. Gellius went for advice to the philosopher Favorinus, who recalled an old rule of Cato’s: where there are no witnesses and one party is the better man, believe the better man. The young judge did not dare to condemn anyone on character rather than evidence; nor could he bring himself to acquit a man he took for a liar. So he did what Roman law allowed him and what almost no institution allows today: he swore that the matter was not clear to him. Mihi non liquere. And he was released from the duty of giving judgment.
Rome knew this verdict outside disputes about money. In the criminal courts a juror could cast one of three votes: A for absolvo, I acquit; C for condemno, I condemn; and NL, non liquet, it is not clear. The third vote was not an evasion of judging. It was a judgment about the juror’s own knowledge, given under oath, and it had a legal consequence: when enough of the jury voted that way, the case was adjourned and heard again. Modern procedure, in Poland as elsewhere, abolished that third tablet. Today’s judge must decide, and where the evidence runs out, the rule on the burden of proof decides for him: whoever asserted and failed to prove loses. In a quarrel over a loan this is practical and wise. It is worth remembering what was lost along the way, because this essay is about a question to which the honest answer is precisely the vote of Gellius, and about two institutions that preferred to vote otherwise.
The first is the textbook. Open a university textbook on molecular biology and the chances are you will find a brief, briskly confident section implying that the question of how life began is nearly settled. It will mention the Miller and Urey experiment. It will gesture toward ribozymes, that is, RNA molecules capable of carrying out chemical reactions on their own, and toward lipid vesicles, microscopic bubbles of fat that close themselves off in water without being asked. The student will turn the page with the impression that science has the matter well in hand. The textbook is the only book that knows more than all of its authors put together.
The confidence has a date of birth. In the autumn of 1952, at the University of Chicago, a twenty-two-year-old graduate student named Stanley Miller sealed methane, ammonia, hydrogen, and water in a glass flask, ran electrical sparks through the mixture for a week to stand in for lightning over a primordial ocean, and opened the valve. The liquid inside had turned brown. Analysis revealed glycine and alanine, amino acids, the pieces from which proteins are built, obtained for the first time from a mixture of gases meant to imitate the atmosphere of the young Earth. Published in May 1953, the result electrified the scientific world. The hypothesis of Oparin and Haldane, that life had emerged from a “primordial soup” under energy from the sky, seemed vindicated. The origin of life seemed, suddenly, not merely approachable but imminent. Give nature the right gases, the right energy, and enough time, and life would write itself.
That was seventy-three years ago. The path has not been walked. The amino acids were real, but the programme that deploys them, the genetic code, the ribosome, the integrated cell, remains as remote as it was in 1953. Remoter, in fact, because we now understand how much more is required than Miller dreamed. The laboratories have done far more since then than their critics admit, and this essay will give them their due without stinting; but the progress concerns the chemistry of copying and the first steps toward assigning amino acids to RNA, not the joining of those parts into a system that sustains itself, copies itself and reads its own code. That whole remains out of reach.
We have partially successful models for isolated steps: amino acid synthesis, vesicle formation, ribozyme activity, vent energetics, and real experimental progress on each. What we do not have is an empirically supported, integrative scenario that connects known geochemical environments to the first evolvable cells. We have fragments, each of which illuminates one part of the puzzle while deepening the mystery of every other part. The geologists and chemists who, in a 2023 review in Frontiers in Astronomy and Space Sciences, gathered what is known about the environments of the early Earth admit outright that no experiment has yet been run end to end, from geochemistry to a protocell. Environmental chemistry, compartmentalisation, kinetics, information: each of these constraints must be satisfied at once, and no framework that holds them together has yet been tested end to end. The same review also stresses constructive pathways and experimental advances; but the honest summary, taking those advances into account, is not that we are “closing in.” It is that the more we learn about the molecular machinery of even the simplest cell, the wider the gap between our partial models and the integrated explanation we lack.
This essay is an attempt at that honest summary, written by a lawyer rather than a scientist, and deliberately so. A lawyer has nothing to say about the chemistry of nucleotides. He does have in his toolbox one instrument that the scientists were deprived of: the third tablet. What it is good for, the reader will see for himself once we have gone through the evidence together. First, though, we need to know what we are talking about.
WHAT WE ACTUALLY KNOW
Let us begin with an honest inventory, the way every trial begins: what is undisputed, and what is in dispute.
We know that all organisms alive today, from the bacteria in the soil to the sequoia and the human being, share a common ancestor. Biologists call it LUCA, the Last Universal Common Ancestor. No fossil of it exists; it is reconstructed by comparing the genes of living organisms and running the clock of mutation backwards. A 2024 study in Nature Ecology & Evolution dates LUCA to roughly 4.2 billion years ago, only a few hundred million years after the Earth formed, and credits it with a genome comparable to that of modern bacteria: some 2,600 protein-coding genes, a mature metabolism, the full machinery for making proteins, and even the rudiments of an immune system. These are statistical estimates: the dating carries a wide margin of error, and the genome is reconstructed, not observed. One word also needs watching. LUCA is the last common ancestor, not the first organism. Between the first self-copying molecule and this ancestor there may lie hundreds of millions of years of evolution about which we know nothing, so the complexity of LUCA does not measure the difficulty of the beginning. It measures something else: the oldest organism we can reconstruct was not a simple cell. It was already a masterwork of molecular architecture, and it appeared startlingly early.
We know that Earth’s early atmosphere differed radically from today’s. The precise composition is debated; current geochemical work increasingly favours an atmosphere dominated by carbon dioxide and nitrogen, with transient episodes richer in hydrogen, rather than the methane and ammonia mixture assumed in the classic experiment. We know that the Miller and Urey experiment demonstrated that amino acids form under electrical discharge in a strongly reducing, hydrogen-rich mixture; a genuine result, though its relevance depends on a composition now considered too optimistic. We know that meteorites deliver amino acids, nucleobases, and sugars to planetary surfaces.
We know that RNA, the chemical cousin of DNA, can function as both information carrier and catalyst, that is, as a molecule that speeds up chemical reactions; this is the foundation of the RNA World hypothesis, according to which the first life did without proteins and without DNA. We know that lipid vesicles form spontaneously in fresh water. We know that alkaline hydrothermal vents on the ocean floor produce differences in proton concentration across their porous walls, analogous to those that nearly every living cell maintains across its membrane and draws on as a battery.
And that is where the knowledge ends and the storytelling begins.
Because here is what we do not know. We do not know how amino acids came to join into functional proteins before there was a cell to join them; a chemist can stitch them together in a test tube today, but the early Earth had no chemist. We do not know the same about nucleotides, the letters of RNA and DNA. We do not know how a lipid vesicle acquires the molecular machinery to replicate itself. We do not know how the genetic code arose: the dictionary that assigns to each triplet of RNA letters one of twenty amino acids, and that is shared by nearly all life. We do not know how homochirality was established, that is, why life builds its proteins almost exclusively from “left-handed” amino acids and its nucleic acids from “right-handed” sugars, when chemistry without life produces both forms in nearly equal measure. We do not know how the ribosome, a molecular factory for proteins of roughly 2.5 million daltons, assembled from components none of which does on its own what the whole does.
Before we proceed, pause for a moment and consider what it is that must be explained. In the time it takes to read this sentence, a single E. coli bacterium has performed thousands of enzymatic reactions. Its ribosomes, molecular machines older than any mountain on Earth, have added some twenty amino acids per second to growing proteins, reading a code they did not write and assembling molecules that nonetheless fold into the correct shape. Its repair enzymes have patrolled its genome, four and a half million letters, catching and correcting mistakes. Its membrane has imported nutrients through channels so selective they distinguish a sodium ion from a potassium ion differing by a fraction of an angstrom. All of this happens in a cell some tens of times thinner than a human hair, in a system that weighs about a trillionth of a gram, every second of every day, in every one of the trillions of bacteria on your skin and in your gut, as you sit reading these words. This is what origin-of-life research must account for. Not amino acids in a flask. Not lipid bubbles in a pond. This. Not as the starting point, for the first living thing must have been incomparably simpler, but as the destination that any account of the beginning must be able to reach.
Let us be candid about the metaphor that origin-of-life researchers themselves most often reach for. They speak of “building blocks”, amino acids, nucleotides, lipids, as if the problem of life were a construction project that merely awaits the right materials on the right building site. The metaphor is revealing in ways its users do not intend.
Begin with the bricks themselves. Even these we produce with difficulty, under carefully engineered laboratory conditions that may not reflect the early Earth. Miller’s spark yields mainly glycine and alanine, the two simplest amino acids, and that in an atmosphere now considered overly optimistic. Ribose, the sugar that forms the backbone of RNA, is so fragile that it decomposes under the very conditions invoked for its synthesis. Nucleotides, the actual letters of the code, resisted every plausible route for half a century. Only in 2009 did John Sutherland’s group in Manchester find a detour that bypasses free ribose and builds the sugar together with the base, and in 2019 Thomas Carell’s group in Munich obtained both kinds of nucleotide in a single scenario driven by cycles of wetting and drying. These are real achievements and I shall return to them. The first honest admission, however, is this: we have produced the bricks only recently, by narrow paths, under conditions we do not know to have obtained.
But grant, for the sake of argument, every brick the optimist could wish for. Scatter amino acids, nucleotides, lipids, and sugars across the face of the young Earth in whatever concentrations you please. You have not begun to address the problem. Because the question is not how bricks come to lie on the ground. The question is how bricks come to assemble themselves into a house, without a blueprint, without a builder, without the concept of “house.” And not merely a house, but a house that wants to build other houses. A house that contains, within its walls, the complete instructions for its own replication, instructions written in a code that bears no physical resemblance to the structure it describes, yet is read, interpreted, and executed with fidelity by machinery that is itself specified by those same instructions.
This is the point at which the building-block metaphor does not merely fail; it misleads. A cathedral, however magnificent, is inert. It does not metabolise. It does not reproduce. It does not care whether it persists. Life does all three, or, to speak with philosophical precision, it behaves as if it does. The humblest bacterium exhibits what Aristotle called telos: purposive behaviour directed toward self-maintenance and reproduction. It swims toward nutrients and away from toxins. It repairs its own DNA. Some bacteria, when damaged beyond recovery, activate programmed death to benefit their kin. None of this is mysticism; it is molecular biology. But it is molecular biology that operates with a directionality, an aboutness, that no known chemical reaction possesses.
Chemists speak of equilibria, of reaction rates, of Le Châtelier’s principle, by which a system pushes back against whatever change is imposed on it. Nowhere in the lexicon of chemistry is there a term for purpose. Electrons do not want to fill orbitals; acids do not strive to donate protons. Yet the simplest living cell, a system composed entirely of chemicals obeying chemical laws, exhibits goal-directed behaviour of staggering sophistication. The emergence of this purposiveness from purposeless matter is not an engineering problem. It is a philosophical problem, arguably the deepest in all of natural philosophy. And it is the one that origin-of-life hypotheses usually wave past in embarrassed silence.
Consider one aspect of this purposiveness that is rarely discussed in origin-of-life literature but that reveals the depth of the problem. Every living cell corrects its own errors. The DNA polymerases, the enzymes that copy genes, usually come with proofreading, built in or supplied by a partner subunit. Repair enzymes scan the genome, detect deviations from the template, and restore the original sequence. But error correction presupposes the existence of a standard, a “correct” sequence against which deviations are measured. A cytosine where a thymine should be is an “error” only because there exists a template that specifies thymine at that position. Where did the template come from before the error-correction machinery existed? And where did the error-correction machinery come from before there was a template worth protecting?
This is the chicken-and-egg problem transposed to a deeper register. It is not merely a question of which molecule came first. It is a question of how normativity, the distinction between right and wrong, entered a system that, at the chemical level, knows no norms. A chemical reaction is not “correct” or “incorrect”; it simply proceeds according to thermodynamics. Yet life, from its most elementary molecular level, operates on the distinction between right and wrong sequences, and actively enforces it. Philosophers of biology have an answer to this, and it must be stated fairly, because it is better than its critics suggest. The answer runs: the “correct” sequence is simply the one that got copied; and once there is copying with selection, resemblance to the template becomes a condition of persistence, and that persistence turns resemblance into a norm. Ruth Millikan and her successors call this the etiological theory of function: a thing is “for” something if it was preserved because it did it. In Millikan’s fuller version a norm arises where one mechanism was selected to produce something and another to use it. This answer explains how a norm is maintained once copying exists. It does not explain how copying came to be the copying of something, rather than a reaction that happens to yield a similar product. I leave that problem open here and will return to it, because it is, in my view, the place where the whole case is really decided.
So let us retire the comforting metaphor. We are not missing a construction manual. We are missing an explanation for why chemistry would become biology; why dead matter would organise itself into systems that fight, with extraordinary ingenuity, against the very decay that should dissolve them. The distance between a puddle of amino acids and a living cell is not the distance between bricks and a cathedral. It is the distance between silence and a Shakespeare sonnet: between the random clatter of typewriter keys and a poem that knows it is a poem, writes copies of itself, and sends them out into the world.
THE PARADE OF HYPOTHESES
The landscape of origin-of-life research is not a converging field approaching consensus. It is a collection of camps, each solving its own piece of the problem and regarding the others with polite disbelief. As a 2006 review in Philosophical Transactions of the Royal Society observed, the field’s models address different aspects of the problem with little integration. I shall present them the way a good advocate presents the opposing case: first in the strongest form they have today, and only then with their weakness.
The RNA World hypothesis is the reigning orthodoxy: before proteins and DNA, life is supposed to have run on RNA molecules that both stored information and copied it. For decades its problem was chemistry. Ribose is fragile; cytosine, one of the four letters, decomposes within centuries; and water, in which everything is supposed to happen, does not join nucleotides into a chain but breaks finished chains apart. Harold Bernhardt, in a 2012 review, called the RNA World “the worst theory of the early evolution of life (except for all the others).” Since then, however, the hypothesis has made more progress than any of its rivals. Tracey Lincoln and Gerald Joyce at the Scripps Research Institute in La Jolla built in 2009 a pair of ribozymes that assemble each other from supplied pieces and multiply exponentially for as long as the supply lasts; in a population of such molecules evolution even began, with better replicators driving out worse ones. In 2024 Joyce’s group showed a ribozyme that copies another ribozyme with errors, and the errors are subject to selection: copying and evolution catalysed by RNA alone; protein enzymes served the experimenters only to amplify and read the results between rounds. In February 2026 Philipp Holliger’s group in Cambridge described QT45, a ribozyme of only 45 letters which, in freezing brine and fed with activated nucleotide triplets, made both a copy of its own sequence and the strand complementary to it; the yield was about a fifth of a percent after 72 days, and the two reactions did not close into a cycle that sustains itself. That is a very long way from Miller’s flask. And yet each of these molecules needs ready-made, chemically “armed” nucleotides, or outright prefabricated fragments, whose availability on the early Earth has not been shown; none yet runs a sustained, multi-generation replication on its own; and none has a code, because there is as yet nothing to encode. Bernhardt was right in both directions: it is the worst theory, and still the best.
Nick Lane and Bill Martin’s proposal that life began at alkaline hydrothermal vents is a masterpiece of reasoning about energy. Warm, alkaline water seeping from the ocean floor meets the more acidic seawater; across the thin partitions of the vent’s porous walls a difference in proton concentration arises, exactly the kind that nearly every living cell maintains across its membrane and draws energy from. The Lost City hydrothermal field in the Atlantic provides a compelling analogue of such a place, and in Strasbourg Joseph Moran showed in 2019 that iron alone, without any enzyme, drives a network of reactions that recreates the core of the metabolism still turning in every cell today. But energy is not the whole problem. The problem is information. A proton gradient tells you nothing about how a sequence of nucleotides comes to mean a protein. The vent hypothesis addresses what life lives on; it does not address what life knows.
The revival of Darwin’s 1871 “warm little pond” conjecture moves the scene onto land: to shallow volcanic pools that dry out and refill in turn, concentrating molecules and helping them join. This addresses the dilution problem, one of the gravest, since the ocean is an enormous solvent, and the drying cycles also help join molecules into chains. What the pond does not say is how polymers become functional polymers, how catalysis becomes self-sustaining catalysis, or how a puddle of drying amino acids begins to reproduce.
Günter Wächtershäuser’s model holds that metabolism came first: cycles of reactions on pyrite surfaces that fed themselves before any genetic molecule existed. The kindred idea of autocatalytic sets says that a network of molecules can sustain itself as a whole even though no member can copy itself alone. Here too the laboratory has done more than is usually admitted: Niles Lehman’s group showed in 2012 that fragments of a ribozyme spontaneously assemble into cooperating networks, and that cooperative networks grow faster than selfish molecules. And here too the same clause applies: the fragments were prepared by the experimenter, and the network has no open-ended heredity of its own that could be called a memory.
Prigogine-inspired models, including Michaelian’s thermodynamic dissipation theory, argue that life is a near-inevitable consequence of thermodynamics: systems flooded with energy organise themselves so as to dissipate it more efficiently. This is a legitimate physical insight. But “near-inevitable” is doing heroic work in that sentence. A tornado is also a structure that organises itself to dissipate energy; so are Bénard cells, the regular hexagons into which a layer of oil heated from below arranges itself. None of this encodes information. The gap between a physical phenomenon that organises itself and a biological one that copies itself and remembers is a difference not of degree but of kind.
THE COPYING MACHINE
Behind all of these hypotheses waits the question that the layman asks first and the specialist last: what is the probability that all this assembled itself? The popular answer points to astronomically small numbers: the chance that random shuffling of amino acids would yield even one working protein is smaller than the whole age of the universe could make up for. That answer hits a position nobody defends. Nobody claims that life arose by random assembly. The claim is that natural selection, or processes like it, acted on chemical systems, guiding them toward complexity. This is the most important counter-argument and it deserves serious treatment. A substantial literature on autocatalytic sets, compositional heredity, and “evolution before genes” proposes that selection may operate on populations of chemical networks before true replication emerges. If these models worked in full, they would bridge the probability gap elegantly. But to date no such network has sustained itself without prepared building blocks supplied by the experimenter, scaled beyond small reaction sets, or generated open-ended, digital heredity. And Darwinian selection proper requires replication with heritable variation. Before the first self-copying molecule exists, there is nothing for selection to work on. The mechanism presupposes the existence of the thing it is supposed to explain.
The dissipative-structure models attempt to circumvent this by arguing that thermodynamic self-organisation precedes and enables biological copying. This is a legitimate and interesting move, but it merely relocates the problem. We must still explain how unguided self-organisation produces a system that operates on coded information: the codon table, the reading frame, the regulatory sequences. Until recently one could write that no known law of physics predicts that energy dissipation will produce a code, and leave the sentence without an addressee. Now it has one: the group of Robert Hazen and Michael Wong proposed in 2023 a “law of increasing functional information,” under which systems selected for function accumulate information the way isolated systems accumulate entropy. The proposal is sharply criticised and far from the status of a law; but the sentence about the missing law must now be written against a concrete proposal rather than into the void. Chemistry does not encode; it reacts. The leap from one to the other remains the central mystery, and no hypothesis on offer has yet provided an empirically supported model of how it occurs.
There is a mathematical form of this problem, and it is worth knowing, because it is stronger than all the probability calculations. It was formulated in 1971 by Manfred Eigen, Nobel laureate of 1967, in a paper from which the modern theory of molecular evolution begins. Every copying machine makes mistakes. When there are more mistakes per copy than the sequence’s advantage over its rivals can make up for, and in the simplest model that limit lies near one mistake per copy, the information dissolves into noise; Eigen called this limit the error threshold, and crossing it the error catastrophe. To copy a long text faithfully you need a proofreader; a proofreader is a catalyst, in today’s cell a protein, which is to say itself a long text; and a long text can be maintained only by faithful copying. Eigen’s paradox, as it came to be called, is a circle in which each element presupposes the others. It must be said fairly what follows from this and what does not. What does not follow is a cliff that cannot be climbed step by step: short molecules can copy themselves inaccurately and persist, the threshold rises as fidelity improves, and Eörs Szathmáry and others have shown in models that compartments and cooperation among many short replicators can break the circle. What does follow is that the first replicator must have been short, shoddy, and miraculously persistent, and that from it to LUCA, with its 2,600 genes and proofreaders reading every letter, runs a road no stretch of which we have seen travelled without the help of a human hand.
If so, why have seventy years of work by the best chemists in the world not reproduced even the beginning of that road? First it must be said fairly where they stand. Jack Szostak, a Nobel laureate, now in Chicago and then at Harvard, builds protocells: vesicles of fatty acids that grow and divide, and in 2013 his group showed that inside such a vesicle a strand of RNA can be extended along a template without any enzyme. Joyce’s ribozymes evolve. Lehman’s networks cooperate. This is not nothing. But none of these systems lives on its own: each receives from the experimenter prepared building blocks and conditions that the primordial Earth is not known to have offered, and none has passed from copying to code. To the question why not, there are two honest answers, and both must be spoken. First: a flask for a week is a product of volume and time some thirty orders of magnitude smaller than a planet for a hundred million years; we do not reproduce the birth of stars or the drift of continents in the laboratory either, and nobody concludes from this that the stars were designed. Second: if the process were as “easy” and “inevitable” as textbook rhetoric proclaims, some fragments of it ought to occur in the flask by themselves, without people who know what they are aiming at and spend weeks designing the conditions. They do not. From these two answers follows neither “impossible” nor “simple.” What follows is “we do not know how hard,” and that is the first sentence the textbook should have written.
WHY NOT NOW? THE BARRIERS RECONSIDERED
One question frequently deployed to reassure the public is: “Why doesn’t life arise spontaneously today?” The conventional answer is that multiple barriers now prevent what was once possible. The barriers are real. But their list, read as a whole, says something about the state of our knowledge that it did not intend to say.
The oxygenating atmosphere is the most cited barrier: the oxygen we breathe, the product of some two and a half billion years of photosynthesis, breaks down organic compounds before they can accumulate. This is directionally correct. But it must be stated with care: as the geochemical literature emphasises, localised oxygen-free environments exist even today, on the ocean floor, in sediments, in subsurface aquifers. The oxygen barrier is real, but it is not the absolute “chemical steriliser” sometimes claimed.
The ecological barrier, that is, competitive exclusion by existing life (whatever arose today would be eaten by bacteria with four billion years of training), is intuitively powerful. But it must be honestly labelled as what it is: an untested hypothesis at the protocell stage. Its strength against an actual nascent system has not, to my knowledge, been measured. The argument is reasonable; its quantification remains speculative.
And the UV/ozone barrier, the idea that the ozone layer, by filtering ultraviolet radiation, has removed the energy source that may have driven prebiotic photochemistry, is a hypothesis that the literature treats as speculative and model-dependent. To present it as decisive is to overstate it.
Now here is the point. Each of these barriers is invoked to explain why life cannot arise today. But each was also supposedly absent four billion years ago, which means the theory requires us to believe that, in the absence of these obstacles and in the presence of nothing more than favourable chemistry, a system of coded information organised itself into the most complex phenomenon we know. The barriers explain why the experiment cannot be repeated. They do not explain how it succeeded in the first place.
Step back and notice the structure of seventy years of accumulated discovery. Not in the oxidising atmosphere. Not in the open ocean. Not without concentration mechanisms. Not without ultraviolet, but not with too much ultraviolet. Not without catalytic mineral surfaces. Not in the presence of biological competitors. Not on any timescale observable by humans. Each of these findings has something behind it, some more than others. But their accumulation creates a pattern that the field has not confronted: we define the conditions for the origin of life almost exclusively by negation. In theology there is a tradition of defining God solely by what He is not, the via negativa of Pseudo-Dionysius and Maimonides. Origin-of-life research has become, without intending it, a form of apophatic biology: we know more and more about how life did not originate, and less and less that we can positively assert about how it did. No integrated positive scenario connecting known geochemistry to the first evolvable cell has been demonstrated. What remains is an expanding catalogue of ruled-out conditions surrounding a shrinking, and still empty, centre.
THE COURTROOM IN HARRISBURG
Here the lawyer steps onto the stage, and makes no apology for it. In Dover, a small town in Pennsylvania, the school board resolved in 2004 that biology students would be read a statement presenting “intelligent design” as an alternative to evolution. Parents sued. The case, Kitzmiller v. Dover, came before Judge John E. Jones III, a Republican appointed by President Bush, and ran for six weeks in the autumn of 2005, with expert witnesses on both sides; the biochemist Michael Behe, design’s most serious scientific advocate, testified for the defence. On the twentieth of December Judge Jones handed down a 139-page opinion. He found that intelligent design is not science: it invokes a supernatural cause, rests on a false alternative (either evolution or design), and makes no claims that could be tested and refuted. Teaching it in a public school violated the constitutional ban on the establishment of religion.
American courts have had, since 1993, an instrument for deciding whether an expert’s opinion is science: the test from Daubert v. Merrell Dow, which asks whether a theory can be tested, whether it has been peer-reviewed, whether its error rate is known, and whether the relevant community has accepted it. The judge is a gatekeeper; he admits only what can be cross-examined. The Dover opinion is not a Daubert ruling; it is a constitutional case about the policy of a public school. But the question the court had to answer on the way, whether intelligent design is science, is the very question Daubert tells a judge to put to an expert, and the court answered it with the same tools. And it must be said plainly, because both sides of the dispute avoid saying it: there is reasoning in that opinion which a lawyer cannot reject even when he dislikes the result. Courts filter evidence all the time and for different reasons: some they reject as unreliable, some as irrelevant, some, like evidence obtained unlawfully, to protect rights rather than because of what it says. The filter that matters here is the first: an expert whose methods nobody can check is not heard as an expert. A hypothesis that fits every possible state of the evidence is not strong. It is empty. And the hypothesis of an intelligent cause, in the form in which it is usually stated, fits everything: had the chemistry turned out easy, the designer made it easy; as it turns out hard, he was all the more necessary.
And yet in the same opinion there is a sentence worth reading twice. Judge Jones wrote that the arguments for design may be true, a proposition on which the court took no position, but that design is not science. “Not science” does not mean “false.” The court did not rule on how life began. It ruled on who may speak about it in a biology classroom. Philosophers of science call what the judge did the distinction between methodological and metaphysical naturalism, and it is worth explaining, because the rest of this essay stands on it.
Methodological naturalism is the working principle that science investigates natural causes. It is a tool, and a supremely useful one. It tells the scientist: “When you enter the laboratory, look for physical explanations. Do not invoke miracles as a substitute for experiment.” As a constraint on method, it is unimpeachable; without it there would be neither penicillin nor the transistor.
Metaphysical naturalism is the philosophical claim that natural causes are all that exist. It tells the scientist, and everyone else, that reality is exhaustively physical and that any hypothesis invoking an intelligence beyond nature is illegitimate in advance. This is not a scientific finding, because no experiment could establish it. It is a philosophical position that likes to travel in a laboratory coat.
The conflation of these two is the commonest error in conversations about the origin of life, on both sides. The textbook conflates them when it turns the rule “look for natural causes” into the conclusion “case closed.” The advocate of design conflates them when he turns the court’s refusal to admit his hypothesis into a biology classroom into proof of an ideological conspiracy. The court in Harrisburg, read carefully, did not commit this confusion. The textbook did. And so did the author of these words, before he read the opinion to the end.
Richard Lewontin, the Harvard geneticist, said the quiet part aloud in a 1997 review in The New York Review of Books: we take the side of science despite the patent absurdity of some of its constructs because we have a prior commitment to materialism; it is not the methods of science that compel material explanations, but our adherence to material causes that compels us to build an apparatus that produces them. This passage is among the most quoted in the creationist literature, and it is usually cut off where I have just cut it. Lewontin goes on to say why: because we cannot allow a “Divine Foot in the door”; whoever admits an omnipotent agent admits that the regularities of nature may be broken at any moment, and then science becomes impossible. This is not a confession. It is an argument, and a good one. The reply to it runs: a hypothesis about a single event at the beginning does not break the regularities that follow it, just as the Big Bang does not annul the physics that came after. But that reply at once raises Daubert’s question: how does one test a hypothesis about a single event in the deep past? Not by repeating it; the historical sciences test such hypotheses by the traces the event should have left. The design hypothesis would therefore have to say what traces it tells us to look for, and which of them chemistry does not predict. And the circle closes.
THREE TABLETS
Let us return, then, to the courtroom and count who is standing in it. Not two parties, as the quarrel would have it, but three.
The first says: life is chemistry that happened, and the explanation will be found in chemistry, because where else would we look. The burden of proof this position must carry is clear: the road from externally supplied nucleotides to a molecule that copies itself from simple ingredients, and then from copying to code. Progress along that road since 2009 has been real, and the first position deserves fair credit for it. It does not deserve the privilege of declaring the case closed. What would refute it? Here is a difficulty its adherents dislike: almost nothing. A proof that the chemical road is impossible does not exist, and nobody can even say what one would look like. A hypothesis that nothing can refute should worry the admirers of the Daubert test just as much when it wears a lab coat as when it wears a chasuble.
The second party is the least known and the most interesting. The philosopher Thomas Nagel, in his 2012 book Mind and Cosmos, rejected both the reduction of life and mind to chemistry and theism, which he does not find credible, and proposed a third way: a nature with a built-in tendency toward complexity and purpose, laws that are not the laws of mechanics. He paid for it with reviews in which, as I read them, there was more indignation than argument, though some did answer him on the merits. Stuart Kauffman has spoken for decades of “order for free”; Terrence Deacon writes of teleodynamics; the Hazen and Wong law mentioned above is sometimes read as an attempt to give this intuition mathematical form, though its authors speak of a law of selection, not of purpose. This position too has its burden of proof: it must show that its “laws” predict something mechanics does not predict, and that this can be checked in a flask. So far they are a programme rather than a result. But they have one virtue a lawyer values: they say what would satisfy them.
The third says: the code has an author. Not in the sense of Genesis read literally, but in the sense that the origin of biological information required a mind, as a tablet of cuneiform does. The argument from analogy is as old as Paley and has the strength that analogies have: great at dinner, small in court. For the genetic code has features that cuneiform does not. It is not wholly arbitrary: Stephen Freeland and Laurence Hurst calculated in 1998 that the standard code minimises the effects of errors in copying and reading better than almost any of a million random codes, and such a pattern is the signature of selection, not of a scribe. It is not wholly universal: we know more than two dozen naturally occurring variants of it, in mitochondria, in protozoa, in bacteria of the genus Mycoplasma, so the code changes, and changes by natural means. Eugene Koonin, one of the foremost evolutionary biologists of our time, together with Artem Novozhilov calls its origin “the universal enigma”: it is not solved, but neither is it without clues. The third position must therefore carry a burden it has not yet taken up: to say what it predicts that chemistry does not, and what would refute it. Until it does, Judge Jones is right, and the tablet C against chemistry cannot be raised.
The reader will notice that I have not counted panspermia, the hypothesis that life arrived on Earth from space. Not out of disdain: the transfer of rocks between planets is documented, and the survival of microbes on such a journey is studied seriously, without a trace of taboo. But panspermia is not an answer to our question. It is a motion for a change of venue. A case transferred to another district has not been decided; it has merely been given more time and a larger room.
What, then, remains when all three have been heard? What remains is what I promised to leave open. The chemistry of 2009 to 2025 has done something it could not do half a century earlier: it has taught molecules to copy themselves with help and to evolve under supervision. It has not done one thing: it has not shown how a reaction becomes a sign. This does not mean the code has gone untouched. In 2023 Sutherland’s group showed that short RNA molecules resembling the arm of today’s transfer RNA pick up amino acids selectively, depending on their last three letters, without any enzyme; a possible chemical ancestor of the assignment, though not a codon and anticodon, and the authors themselves call the degree of this “coding” low. Joyce’s ribozyme copies a sequence, but the sequence means nothing: there is no dictionary in which a triplet of letters stands for an amino acid, no machine to read that dictionary, and no system in which the assignment is inherited because it serves survival. The passage from a molecule that reproduces itself to a molecule that says something about something is the place where none of the three parties can yet show a road travelled. The first believes it will get there. The second believes nature is inclined that way. The third believes that this is exactly where someone was needed. None of them has proof.
THE VOTE OF GELLIUS
If I had to vote, my tablet would be NL. Not because I have no opinion; Gellius had one too, after all he believed the plaintiff. Because a judge votes with the evidence, not with his opinion, and the evidence, as of today, does not suffice for either of the other two votes. The textbook raised tablet A for chemistry and declared the case closed. It had no right to, just as it would have had no right to raise C. The court in Harrisburg did something wiser, but also incomplete: it held that one party may not appear in the biology classroom as science, and rightly, because that party cannot answer the questions science asks; it did not hold, and said so expressly, that the party is wrong. Nor does the tablet NL mean that the three tablets weigh the same: the first party has a laboratory record the other two lack, the second a programme, the third an analogy. It means only that the evidence does not yet establish how the transition happened. The legal mind is trained in one thing: to tell what has been proved from what is believed, and to keep the burden of proof where it was placed.
Privately, outside the courtroom, I have a wager, and honesty requires me to lay it on the table with its price. I bet that the code has an author: that the passage from reaction to sign will not be explained as an accident of chemistry, however long the chemistry is churned. And I say what would lose me the wager: a chemical system in a flask, supplied with energy and simple ingredients but not with the assignments to be explained, that develops and keeps over the generations a heritable dictionary of even two entries: two classes of RNA sequence directing two different amino acids into peptides that help the system persist. The route need not run through RNA alone. Such a result would show that no author was needed for that transition; it would not show that there was none, and I would concede the first without claiming the second. But the wager I would lose, and I would pay without haggling. Whoever cannot say what would lose his wager has no wager; he has a confession of faith, and that is fine, provided he does not call it science. I address that warning to myself as much as to the authors of textbooks.
CODA: THE QUESTION THAT WILL NOT DIE
Life is the most extraordinary phenomenon in the known universe. The genome of a single E. coli bacterium is a book of four and a half million letters that the cell reads, copies, and corrects without pause. The human genome is a library of three billion letters that not only encodes the construction of a body of tens of trillions of cells but includes the logic to build that body from a single fertilised egg, a feat so far beyond human capability that we cannot replicate it, cannot fully comprehend it, and cannot explain how it arose.
To look at this phenomenon and declare, with confidence, that it requires no explanation beyond unguided chemistry, while simultaneously acknowledging that we cannot demonstrate how unguided chemistry could produce it, is to rely on a commitment that outruns the evidence. It is an act of trust: trust that an explanation will be found. Such trust may be vindicated; the progress of recent years gives it more grounds than it has ever had. But it should be recognised for what it is: a wager, not a conclusion. And the same standard holds in the other direction: to look at this phenomenon and declare that it certainly requires an author is also a wager, and should not be called a conclusion either.
The honest scientist must hold the question open, and the honest judge must be able to say that he does not know. The origin of life is not a solved problem, nor a nearly solved one; nor is it a problem shown to have no solution. It is the deepest question in all of natural philosophy, and the deepest questions are answered neither by pronouncing the verdict before the trial nor by forbidding them to be asked.
But the question will outlast both verdicts. It always does. Somewhere, as you read these words, a ribosome, a structure older than the Atlantic Ocean, older than the continents, older than the rocks beneath your feet, is reading a sequence of nucleotide triplets in a bacterium on your skin, translating a code that has been copied without interruption for four billion years, assembling a protein that will fold into a shape that chemists spent half a century learning to predict, in order to maintain a system that fights, with quiet and extraordinary ingenuity, against the decay that should have claimed it long ago. No one knows whether anyone wrote that code. No one knows how it arose. And no one, despite seventy years of trying, has managed to make chemistry write it again from nothing.
The mystery is not diminishing. It is deepening. Gellius thought that a judgment based on the character of the parties rather than on evidence would be a presumption unbecoming his age and his merits, and he chose to swear that he did not know, though he must have known how it looked. Two thousand years later it reads as the most honest sentence in that whole case. In the case of the origin of life, such a sentence still waits for someone to say it under oath.
Robert Nogacki, Warsaw, 21 March 2026; second version: September 2026
Robert Nogacki is a Polish attorney at law (radca prawny), the founder and managing partner of Kancelaria Prawna Skarbiec (Skarbiec Law Firm), which has operated continuously since 2006.
The law is equal for everyone, but the parties rarely are: on one side stands an organization with time, money, and lawyers, on the other a person with one business, one nest egg, and one life.
Clients rarely come to him with a legal problem. They come with a problem that also has a legal side: an audit that began with a single invoice, money entrusted to someone who has disappeared, a company that has to be passed on before it is too late. Most such matters are decided long before the first letter is written, in decisions made without asking and in deadlines nobody remembered. So he begins by asking how the client got here, not what the client should have done.
He advises entrepreneurs and families from more than a dozen countries, including those whose accounts the tax office has just seized and who do not know what to do tomorrow morning. He defends them in tax audits, customs and fiscal inspections, disputes with the tax authorities, and criminal tax proceedings. He represents victims of investment fraud and Ponzi schemes. He helps families set up family foundations and plan succession, so that a life’s work outlasts a single generation.
Not every case can be won. Every case can be run so that the client knows where they stand. Since 2006 he has represented the victims in the WGI case (Warszawska Grupa Inwestycyjna, the Warsaw Investment Group), one of the longest criminal cases in the history of the Polish financial market, because some things must not be left half finished, even when they take two decades. In the case of the collapsed cryptocurrency exchange Zonda (Zondacrypto, operated by BB Trade Estonia OÜ), he represents several hundred victims in the criminal investigation conducted by Poland’s National Prosecutor’s Office and in the Estonian bankruptcy proceedings.
Kancelaria Prawna Skarbiec is listed in the rankings of Poland’s largest tax advisory firms published by Dziennik Gazeta Prawna and Rzeczpospolita, and it is a four-time recipient (2015 to 2018) of the European Medal awarded by the Business Centre Club and the European Economic and Social Committee. Robert Nogacki publishes regularly, in the press and on the firm’s website, for people who have a problem rather than a law degree, because a legal opinion the client cannot understand protects only the lawyer.
He believes that the best legal advice is the kind that means the client never has to appear in court.
