Two independent readings
We read each amount from the document’s text and again from an image of the page. The two methods fail in different ways, which is what makes their agreement evidence. When they differ, the fee goes to review.
Every published fee record meets the four conditions below.
We read each amount from the document’s text and again from an image of the page. The two methods fail in different ways, which is what makes their agreement evidence. When they differ, the fee goes to review.
Each record keeps the fee’s exact wording, the source document and page, and the effective date and adopting resolution wherever the schedule or the jurisdiction’s record names them.
When the two readings show the same amount written two ways, such as $1,200 and $1,200.00, they count as agreeing. Every other difference is reviewed against an image of the page, and the schedule is published once every difference is resolved.
A schedule labelled draft or proposed is published only on evidence that it was adopted. We check each jurisdiction’s official sources for a newer schedule and read a new one by the same method.
We use AI models to read fee schedules. A model’s reading is published when it matches the reading from the document’s text, or when the difference is resolved in review.
These conditions apply to every FeeRegistry record and to every peer fee used in a FeeFoundry study.