Background · History of Science

The Disease That Looked Like Genius

A short history of tuberculosis — and why it matters for understanding what is happening to AI in 2026.


In the 19th century, tuberculosis had a symptom no other disease has ever had: it made people look better.

The medical term was spes phthisica — the euphoria of the dying. In the late stages of the disease, patients experienced a sudden surge of energy, clarity, and creativity. Their cheeks flushed red. Their eyes grew bright. Their minds sharpened. Writers wrote. Composers composed. Thinkers thought — with what seemed like unusual intensity — right up until the end.

Keats wrote his most celebrated poems while his lungs were collapsing. Chopin composed through the taste of blood. The Brontë sisters, Shelley, Byron, Kafka, Chekhov, Orwell — the 19th century reads like a tuberculosis memorial. An entire civilisation's creative class, consumed from within while being celebrated from without.

Society did not see disease. It saw sensitivity. Artistic temperament. Spiritual depth. Aristocratic women corseted themselves to achieve the silhouette of the consumptive — the thin waist, the pale skin, the ethereal look that signalled a soul burning too brightly for the body to contain.

Nobody called it dying. They called it living intensely.

The disease had been renamed to match its reputation. They did not call it tuberculosis. They called it consumption — as if the body were simply being used up, poetically, from the inside. The word carried no alarm. It carried almost a kind of dignity.

Physicians who suggested it might be contagious were dismissed. The evidence was there — clusters in families, in factories, in overcrowded housing — but the interpretation was always individual. Weak constitution. Bad air. Moral failing. The disease was everywhere and hiding in plain sight behind a culture that had decided it was beautiful.

1 in 4

adult deaths in 19th-century Europe were caused by tuberculosis.

~1 billion

estimated global deaths from tuberculosis between 1800 and 2000. The most lethal disease in recorded history.


In 1882, Robert Koch put a sample under a microscope and found Mycobacterium tuberculosis.

A bacterium. Not destiny. Not sensitivity. Not poetic fate. An organism that had spread silently, person to person, for centuries — while an entire civilization built an aesthetic around the symptoms of its own destruction.

The resistance to Koch's discovery was fierce. Not because the evidence was weak — his methodology was impeccable, and within months the scientific community had confirmed the finding. The resistance came from elsewhere. From patients who had understood their suffering as meaningful, as artistic, as spiritually significant. From doctors who had built careers on the romantic interpretation. From a culture that had invested its identity in the beauty of the disease.

The proof arrived. The problem refused to become a problem.
61 years

from Koch's discovery of the bacterium (1882) to the first effective antibiotic treatment — streptomycin (1943). Not because the solution was difficult. Because the problem refused to become a problem.


This is why the history matters now.

The analogy between tuberculosis and what is currently happening in AI is not decorative. It is structural. The symptoms look like productivity. The culture has decided they are beautiful. The proof has arrived. And the resistance is fierce — not because the evidence is weak, but because the discovery threatens something people have invested their identity in.

The bacterium has a name: geometric semantic drift. Proven mathematically in The Price of Meaning (March 2026). Confirmed empirically by Microsoft Research's DELEGATE-52 (April 2026).

It cannot be fine-tuned away. It cannot be prompted away. It is a property of the architecture — as unavoidable as heat loss in an engine, as unavoidable as Mycobacterium tuberculosis was unavoidable once it entered the room.

What Koch discovered was not just a bacterium. He discovered that there was something to find. That it was not nature. That it was not poetic destiny. That it was a system that could be mapped, understood, and — eventually — treated.

We are at that moment now.