IsThisOriginal

IsThisOriginal asks a narrower question than the AI-detection industry does. Not "did a machine write this" — it makes no claims about who or what wrote the words — but whether the argument inside a draft contains anything that isn't already lying around in public. Each analysis assembles its own baseline: what ordinary searches retrieve on the topic, and what ordinary prompts to a base model generate. The draft's claims, mechanisms, and syntheses are placed against that field, claim by claim, and the report shows where the writing follows familiar paths and where a claim genuinely departs from them. Originality here is contextual, not absolute — an argument can be common in one discourse and rare in another, and the report always states which corpus it measured against.

The design principle underneath is that rarity is never treated as truth. Following the creativity-research convention that originality and effectiveness are distinct properties, Idea Originality and Idea Quality are computed independently and neither feeds the other — an argument can be rare and bad, or common and sound, and the instrument is built to say so. The example report on the site runs one of my own published essays and prints the findings that do not flatter it: substantially beyond what search and models produced, and thinly supported in places all the same. Building a tool that grades its builder's writing honestly seemed like the only credible way to ship it.