Bills

“Gut and stuff” research tracks substantial replacements of a bill’s substantive text. Policy change is assessed separately: a rewrite can retain the same purpose. A proposed amendment is counted as adopted only when the official evidence supports adoption.

Two subscription subagents independently examine the same official evidence without seeing each other’s answers. An additional subagent can adjudicate disagreements against the sources. Unsupported conclusions, missing evidence, and unresolved disagreements remain uncertain. Agreement is not a guarantee of accuracy.

Session totals count each HB or SB once, including bills that did not become law. Resolutions and memorials are excluded. Counts describe the research inventory, independent of your search, interests, following, and status filters. Missing or unassessed bills are not negative findings.

The replacement watch is a rule, not an AI assessment: a bill whose introduced text is a short study bill (a study word, at most three sections, under 400 words) or a blank form (a run of blanks, 300 words or fewer) is listed once a proposed amendment is posted for it. In the 2021 to 2026 training sessions, 271 of 374 such bills were later replaced; the rate is historical and says nothing about a particular bill’s intent.

Assessments are published research snapshots. Opening this page reads saved research and does not run an AI review. Refreshes require another evidence export, independent assessment, and validated import; reports do not continuously refresh. Each bill’s evidence and reviewer provenance are available on its detail page.