It imputes the drafters' goals for you to confirm or correct, and surfaces the likely effects whether or not they serve them.
What this is. A prototype of the analysis flow, run on the official discussion-draft text (linked above) on and spot-verified against the source. The analysis is precomputed, not generated live in this page. It is not a cost estimate, not legal advice, and it covers two titles of a 902-page draft — the report notes its own source limits, including a truncated tail of Title II.
Countersign the imputed goals
These are what the drafters appear to be trying to achieve, phrased as staff would own them. Confirm, edit, or strike each. Nothing below is held back — the full analysis is open from the start. In the working product, your decisions here steer the next pass: the goal-conditioned questions (does this provision serve your confirmed goals, and what would you trade?) are generated from them. Here they're recorded to show the flow.
Shown regardless: likely effects appear whether or not they serve the confirmed goals, adverse mechanisms included. Confirming goals steers what the tool asks next; the full analysis stays open either way.
Where this meets Thesis
Each provision's analysis ends in two artifacts built for Thesis, the scored forecasting lab this tool belongs to: an outcome metric restricted to officially published, first-print-resolvable series, and a conditional forecast — the metric's path if this is enacted versus not. Registered in Thesis, those become pre-registered, externally witnessed predictions graded against the government's own first prints, with the misses kept on the page. That is the point of the pairing: "this carve-out will suppress take-up" is unfalsifiable commentary on its own; as a Thesis cell it is a claim with a scoreboard. None of the candidate cells below are registered yet — the badges say honestly which series Thesis can already resolve, which it doesn't track yet, and where no official series exists at all.