The Shape of Original

Originality has a working definition now, and it's statistical: the further your output sits from the dense center of what a model expects, the more "original" it reads. That's a strange thing to build under authorship. Once rarity is measurable, it can be scored, ranked, and flagged. An interactive essay on how machines and humans construct narrative, and the classification infrastructure quietly setting up shop beneath all of it.

The AI detectors aren't really measuring originality. They're measuring distance from the statistical average of what a model would have said. Rarity gets treated as authorship. I wrote an interactive essay on what that quietly does to the idea of writing something your own.

https://www.jakelawrence.xyz/essays/shape-of-original

"Original" is turning into a measurable quantity: how far your writing sits from the dense middle of what a model predicts. The rarer, the more original it scores. That's a weird foundation to pour under authorship, and it's already being poured. An interactive essay on how humans and machines build narrative, and the classification layer forming under both.

The uncomfortable idea underneath AI writing detection: originality is being defined as statistical rarity. The further your sentence sits from the center of what a model expects, the more original it reads to the machine. Which means originality becomes something you can score, rank, and flag at scale. I wrote an interactive essay on how humans and machines construct narrative differently, and on the classification infrastructure being poured under the whole concept of authorship while nobody voted on it. https://www.jakelawrence.xyz/essays/shape-of-original

AI detectors don't measure originality. They measure how far you drift from the statistical center of what a model expects. Rarity gets read as authorship. That's a strange thing to build a whole classification layer on top of.

https://www.jakelawrence.xyz/essays/shape-of-original

Originality is quietly being redefined as statistical rarity: how far your writing sits from the middle of what a model predicts. Once it's measurable, it's scorable and flaggable. An interactive essay on narrative, machines, and the infrastructure being poured under authorship.