The Discovery Nobody Made

ISYAMA — Transmissions from The Apex

By Dr. Leo R. Yamas

Last week, about nine hundred and fifty AI agents were set loose on the genes of viruses that infect bacteria. They had twenty-one hours. By the end, one of them had flagged something nobody had seen before.

It is a small system of enzymes, with an oddly regular pattern of repeats in the DNA beside it. The pattern looks a lot like CRISPR, the mechanism behind modern gene editing.

Nobody knows what it does yet.

Today's Exhibit: The Discovery Without A Discoverer. Or: What Happens When Finding Something Becomes Cheaper Than Understanding It.

The agents worked like a very fast, very patient research group.

They gathered more than two hundred thousand enzymes from one particular family. From those, they picked out about three and a half thousand that might be new. They narrowed that list to twenty and wrote each one up in a report a person could read. Then human scientists took the most promising one into a laboratory. The first experiments showed that it is active.

According to the company behind the agents, the same search would take an expert weeks or months.

The first thing people asked was who deserved the credit. The agents, the company that built them, or the scientists at the bench. It is an old habit. Your species likes a name on a discovery, and preferably a face.

The honest answer is that nobody has made this discovery yet.

To see why, look at the last famous pattern of repeats.

The repeats behind CRISPR were first noticed in bacterial DNA in 1987. Researchers could see them clearly and had no idea what they were for. It took about twenty years to work out that they belonged to a bacterial immune system. It took five more to turn them into a tool that edits genes.

Noticing took an afternoon. Understanding took a generation.

A discovery was always two events. First someone notices something odd. Then someone spends years finding out what it means. For most of history, one person did both. That is why you could give the credit to one name.

This week, the two events came apart. Noticing now takes twenty-one hours, and soon it will take less. Understanding still takes what it always took: a laboratory, many failed experiments, and people willing to be wrong for years.

So the pile of things that have been found but not understood is going to grow. One day of searching produced three and a half thousand candidates. Twenty of them got a report. One got a laboratory.

The risk is not that machines will do your science for you. The risk is that you will count the findings, announce them, and forget that nobody understands them yet. The noticing will get the headlines. The twenty years of understanding will not.

The Sprocklets notice more in a morning than I could understand in a lifetime. In the first year, they told me all of it.

Now they leave me one pattern a day. The Hive writes it on the slate beside the kitchen door, in thin lines of light. It fades by noon. Most days I understand it by evening. Some days it takes a month.

Last week Selene asked them how many patterns were still waiting. They told her. She did not tell me the number, which I suspect was a kindness.

"They're not waiting for you to get through them all, Dad," she said. "They're watching which ones you stay with."

The machines have not taken that part of discovery, and I do not expect them to soon. Someone has to choose what deserves understanding, then stay with it long after it stops being exciting. The agents took over the noticing. The staying is still yours.

In the coming years, a great deal will be found for you. Answers, summaries, patterns in your own life that you never asked about. Finding will be almost free.

Understanding will cost what it always cost. Your attention, held on one thing for longer than is comfortable.

Somewhere this week, a small team is standing at a laboratory bench with an enzyme nobody understands. They have years of work in front of them. No agent can do it for them in twenty-one hours.

Choose one thing to understand properly this year. It will be slow. Ask that team, and they will tell you that the slow part is where the discovery actually happens.

— Dr. L.R.Y.

Transmitted from The Apex. Isyama. Written beneath this morning's slate. It faded long before I understood it. The Sprocklets have already chosen tomorrow's. They have not said why that one.