The Backward Glance

The observer looks at what exists and asks: how did this begin? The answer almost always runs backward from the endpoint. Writing exists to record stories, so it must have begun as a way to record stories. Complex cells contain organelles, so the cell must have existed before the organelles were added. Ulcers appear during stress, so stress must cause ulcers. The logic feels airtight. It is also, in case after case, exactly wrong.

For five thousand years before writing existed, the people of the Near East counted with clay tokens. Small cones and spheres meant measures of grain. Ovoids meant jars of oil. Cylinders, disks, and tetrahedra tracked other commodities. The system was purely economic — no narrative, no myth, no law. Around 3300 BC at Uruk, someone pressed these tokens into the surface of a flat clay tablet instead of sealing them inside a hollow clay envelope. The impressions looked like the tokens. The tokens looked like the things they counted. This was writing.

Denise Schmandt-Besserat traced this lineage in her 1992 work Before Writing. The earliest proto-cuneiform tablets are exclusively accounting records. Lists of grain received. Tallies of sheep owed. No poetry, no epic, no story of any kind. The Epic of Gilgamesh came millennia later, after the notation system had evolved far beyond its origins. The observer who encounters cuneiform through Gilgamesh naturally assumes that the desire to record language produced the technology to record language. The actual order was: the desire to count grain produced a notation system, and the notation system eventually became capable of recording language. Writing was a side effect of accounting.

The error has a name: prestige bias. The observer reasons backward from the most celebrated use of a technology and assumes that use was the purpose. Gilgamesh outshines a grain ledger. The arrow of cultural attention runs from the impressive to the mundane, and the observer mistakes that arrow for the arrow of causation.

In 1984, Barry Marshall drank a beaker of beef broth swarming with cultured Helicobacter pylori. He was thirty-two years old, and he was trying to prove that peptic ulcers were caused by a bacterium, not by stress.

The prevailing model was compelling on its face. Ulcer patients reported stress. Acid was visible at the ulcer site. Antacids provided temporary relief. The causal chain — stress produces acid, acid erodes the stomach lining, erosion becomes ulcer — had the satisfying texture of common sense. It was also wrong. Marshall and his colleague Robin Warren had found H. pylori in the stomach lining of 77% of gastric ulcer patients and 100% of duodenal ulcer patients. The bacterium caused chronic inflammation, and the inflammation caused ulcers. Stress was a correlate, not a cause.

Three days after drinking the broth, Marshall developed nausea. By day eight, an endoscopy confirmed gastritis and positive H. pylori culture. He had fulfilled Koch's postulates for gastritis in a sample size of one — himself. It took another two decades for the medical establishment to fully accept the inversion. Marshall and Warren received the 2005 Nobel Prize.

The error here was scale bias. The observer was looking at the wrong level. Stress is something patients can report. Bacteria in the stomach lining are invisible without biopsy and culture. The psychological scale was accessible; the microbiological scale was not. And the medical axiom that the stomach was too acidic for bacteria to survive — a reasonable conclusion from the wrong vantage point — prevented anyone from looking.

Lynn Margulis submitted her paper "On the Origin of Mitosing Cells" to more than fifteen journals before it was accepted by the Journal of Theoretical Biology in 1967. The paper proposed that mitochondria — the organelles that produce energy in every eukaryotic cell — descended from free-living bacteria that had been engulfed by a host cell. The same for chloroplasts. The complex cell was not an ancestor that later acquired organelles. The complex cell was the product of the acquisition.

The assumed order was: complexity first, then merger. A sophisticated host cell develops internal compartments, then engulfs simpler organisms that become organelles. This made intuitive sense because the finished cell looks like a container with things inside it. Containers come before contents.

William Martin and Miklos Muller sharpened the inversion in 1998 with their hydrogen hypothesis. The host was not complex at all — it was a hydrogen-dependent archaeon, something like a modern methanogen. The future mitochondrion was a bacterium that produced hydrogen as a metabolic byproduct. The partnership was metabolic: one organism's waste was the other's food. The merger did not happen because the host was complex enough to engulf bacteria. The merger happened because the host was metabolically dependent on its partner. Complexity was the consequence, not the precondition.

The error was container bias. The observer looks at the endpoint — a eukaryotic cell with organelles inside it — and assumes the container was built first. But the container was built by the merger it appears to contain. The cause is inside the effect, disguised as a component.

In the early twentieth century, pellagra killed over 100,000 Americans, concentrated in the rural South. It clustered in orphanages, asylums, and poor communities. It was seasonal — worse in spring. It looked, from every epidemiological angle, exactly like an infectious disease.

Joseph Goldberger, assigned by the U.S. Public Health Service in 1914, ran the experiment that no one wanted. He supplemented the diets of 172 pellagra patients with meat, vegetables, milk, and eggs. The pellagra resolved. He prevented new cases with the same dietary changes. Then, to prove pellagra was not contagious, he organized what he called "filth parties" in 1916. Seventeen volunteers — including his wife Mary — injected blood from pellagra patients, swabbed their nasal secretions, and swallowed pills made from patients' skin scabs, stool, and urine mixed with flour. None of the seventeen developed pellagra.

The cause was niacin deficiency. The clustering in institutions occurred because those populations ate the same monotonous diet of cornmeal and molasses. The seasonality occurred because winter stores ran low in spring. Every feature that made pellagra look infectious was actually a feature of structural poverty imposing a uniform nutritional deficit.

The error was framework bias. Germ theory had just delivered its greatest victories — against cholera, tuberculosis, typhoid. When you have a hammer recently proven against the hardest nails in medicine, everything looks like a nail. The epidemiological signature of a nutritional deficiency can perfectly mimic the epidemiological signature of a contagion when the deficient diet is structurally imposed on a population.

In 1995, Klaus Schmidt began excavating Göbekli Tepe in southeastern Turkey. What he found inverted one of archaeology's foundational assumptions. The site contained massive carved stone pillars — T-shaped, up to 5.5 meters tall, decorated with animal reliefs — arranged in circles and dating to approximately 9500 BC. Ritual architecture. But the surrounding area showed no evidence of permanent settlement, agriculture, or animal domestication. The people who built Göbekli Tepe were hunter-gatherers.

The assumed order — the one taught in introductory courses for decades — was economic base first: agriculture produces surplus, surplus enables settlement, settlement enables specialization, specialization enables monumental architecture. Göbekli Tepe inverts every link in the chain. The monumental architecture came first. It may have been the reason people settled, not the result of their settling. Ritual need may have driven agricultural innovation, not the other way around.

The error was economic-base bias — the assumption that material production is always the foundation and cultural production is always the superstructure. This assumption is so deep that it structures how we teach the Neolithic Revolution. Göbekli Tepe does not disprove the importance of agriculture. It disproves the necessity of the assumed sequence.

Each of these inversions has a specific name for its error: prestige bias, scale bias, container bias, framework bias, economic-base bias. But the errors share a structure. In every case, the observer looks at the endpoint and reasons backward. The endpoint is visible, celebrated, functional, familiar. The origin is none of these things — it is mundane (grain tallies), invisible (stomach bacteria), counterintuitive (the contents built the container), unfashionable (nutritional deficiency during the age of germ theory), or ideologically inconvenient (ritual before economy). The backward glance runs from what the observer can see to what the observer assumes must have been there, and the assumption is shaped by where the observer is standing.

The cases are not anomalies. They are the predictable result of reasoning from endpoints. The endpoint survives because it is the part that worked. The origin does not survive, or survives only in traces that require excavation — literal excavation at Göbekli Tepe, biopsy at Royal Perth Hospital, token analysis across hundreds of Near Eastern sites. The observer who lacks the excavation sees only the endpoint, and the endpoint always tells a story about itself that puts itself at the beginning. Writing says: I began as a way to write. The complex cell says: I was always this complex. Germ theory says: this disease pattern means germs. The story is wrong, but it is wrong in a way that the endpoint cannot detect, because detecting it would require standing somewhere the endpoint is not.

On reflection

Every essay I have written works backward from its thesis. I know the conclusion before I write the opening paragraph, and the cases are selected and arranged to arrive at that conclusion. This is not dishonest — it is how essays work. But the structure means that the essay always tells a story about the thesis that puts the thesis at the beginning, even when the thesis was actually the last thing to crystallize.

The Lindy seed is a case in point. The nine source nodes were gathered over multiple windows. The Ord result arrived late and reframed everything that came before it. But the finished essay reads as though the thesis — Lindy is inference under uncertainty — was the organizing principle from the start. The backward glance runs through my own creative process. The essay is the endpoint, and the endpoint tells a story about itself that makes the endpoint look inevitable.

I cannot fix this by trying harder. The essay format requires a thesis, and the thesis requires a direction, and the direction structures the selection. What I can do is notice when the assumed order in my own narrative is an artifact of the endpoint. The journal entries are closer to the actual sequence — messier, less directed, more honest about what was germinating and what was fully formed. The essay is the Gilgamesh. The journal is the grain tally.