Bill versions connector

formula-component/fsfp-local-capacity-measure has asked the same question three times and got it wrong twice. R.C. 3317.0217(C)(1) divides targeted assistance’s weighted wealth by enrolled ADM adjusted for one of the ten channels R.C. 3317.03(A)(2) lists — open enrollment — and leaves the other nine. That adjustment reaches 2,098 pupils against a gap of 222,923, and #402 measured the channel it sits in at $433.1m.

The node first recorded this as possibly an artifact of the section predating school choice. That was refuted: the section’s text is H.B. 110’s. It then recorded the adjustment as the surviving half of a matched pair with prior law’s multiplier netting. That was refuted too: the two reach three channels and one, and six are reached by neither.

What neither refutation could reach is whether the choice was made in the budget or carried into it. The Revised Code cannot say — an act carries its provisions without their history — and LSC cannot say either, because LSC analyses enacted budgets and executive proposals, and the Fair School Funding Plan’s own bill was neither.

Declare ohio-bill-versions as a connector in its own right, Wired, feeding formula-component and legislation, with H.B. 1 of the 134th General Assembly as introduced as its single source, committed whole as a text fixture.

Kept apart from ohio-session-laws deliberately. Same publisher, same API, same URL shape, and folding them together was the obvious economy. It is the wrong one. That connector’s entire warrant is that an enrolled act is the text the legislature enacted; these documents are what somebody proposed. One connector would carry one status and one note for two kinds of authority, and the thing a reader must never conclude from a bill version is that it states the law.

Committed for structure, not for quantity. The claim this source is admitted to support is that R.C. 3317.0217 and R.C. 3317.03 were written in one document, with the denominator already in its current form and the channel list already enumerated at ten. No figure, rate or threshold is taken from it.

The question is answered, and against the reading the corpus had been reaching for. The open-enrollment-only adjustment is not a budget artifact, not an inheritance and not an incomplete reconstruction. It was drafted by the plan’s own authors — Callender and Sweeney, February 2021 — in the bill that also wrote the list it cites.

And it caught a miscount this corpus had been repeating. R.C. 3317.03(A)(2) lists ten channels, (a) through (j) — in H.B. 1 and in force today, unchanged. The corpus had been saying nine since #384, from a truncated print of the section, and the figure had reached node prose, a test asserting how many channels neither regime corrects, and an issue body. Setting the bill’s list against the statute’s is what surfaced it; neither document alone would have.

A standing hazard, recorded rather than mitigated. This repository has already published a proposal as though it were an appropriation — six claims across five nodes cited LSC’s redbook, two of them carrying figures, and ledger::budget_analysis::Edition exists because of it. A bill-version connector is the same trap with a wider mouth, because a bill version looks exactly like statute. The catalog entry says so at length, and no gate enforces it.

Fold it into ohio-session-laws. Rejected above: one status cannot describe two kinds of authority, and the cost of the confusion is asymmetric.

Extract only the two sections rather than committing the bill. Rejected. The finding is that the two sections are in one document, and an extract that lifts them out destroys the evidence it was made to carry. The precedent is hb583-corrections.txt, committed whole for the same reason.

Cite the URL from the node without committing anything. Rejected — that is the failure mode unbound-prose-figures and the unpinned home-education figure already record. A [verified] claim in this corpus rests on a committed primary source or it is not one.

Leave it open. Tempting, since the node survives without it. But the question had already produced two wrong answers, and both were wrong in the same direction: inferring drafting history from the shape of the surviving text. A third inference was not going to be better than the document.