What your habits look like.
Paste something you wrote. It looks for four habits that turn up more often in machine drafts than in published human writing — and it tells you when it can't tell, which is often. Nothing is uploaded.
Everything happens in this tab. Nothing is sent anywhere, and there is no key to enter.
Questions
Does this tell me if a machine wrote it?
No, and it is built so it can't pretend to. It measures habits, not authors. Careful human writing sets it off all the time — that is the finding, not a flaw in it.
What are the four things it looks at?
Paragraphs built alike — whether every paragraph opens and moves the same way. Repeated phrasing — how often the same runs of words come back. Contrast setups — phrases shaped like "not this, but that". Asides in brackets — how often you break off into parentheses.
That is the whole list. Each one is counted, compared against two small samples of writing, and reported on its own so you can see when they disagree.
Then what is the number?
A count, not a verdict. It comes from comparing your habits against
20 machine passages and 16 human ones — a small, uneven sample, most of the
human writing from before 1950. The model file marks itself discovery-only
for that reason. When there isn't enough to go on, it says so instead of guessing.
Why does it so often say nothing?
Because a few sentences can't carry this much counting. On samples this small, one habit will line up perfectly by pure luck about one time in seven — so saying nothing is the honest answer, not a failure.
Where's the model?
There is a closer reading that runs a small language model on your own device and watches which words it finds surprising. The model file isn't installed yet, so that button is switched off and this page counts habits only. Either way your text goes nowhere.
How do I know any of this is real?
You mostly don't yet, and we'd rather say so. The whole record is written down in the measurements — including the part where the check that guards all of this was caught approving writing that had nothing in it at all, and what it took to catch that.