H.B. 643 (136th G.A., as introduced)

Introduced in the 136th General Assembly by Representatives Pizzulli and Somani. It amends a single section, R.C. 3310.032, to make a student ineligible for an educational choice expansion scholarship if the family’s federal adjusted gross income in the preceding tax year was at or above $500,000, beginning with the 2026-2027 school year and indexed annually to the consumer price index. verified

It is the first income ceiling proposed since H.B. 33 made the expansion scholarship universal, and it is small: an eleven-kilobyte bill against an operating budget that amends more than two thousand sections. That contrast is why it is here.

What it would change, and through which mechanism # ↑ Contents

The expansion scholarship is an award paid to a family, not a deduction against a resident district’s foundation payment — the Fair School Funding Plan changed that mechanism, after which community and STEM students are funded directly rather than deducted. verified So a narrowing of eligibility reduces the number of awards paid and the appropriation they are paid from, and reaches a traditional district’s own funding only through whatever enrollment consequence follows a family’s choice, which no source here measures. inference

Its status is a fact about the bill and not about its merits # ↑ Contents

As introduced and nothing further: the version index carries one entry. verified A bill that has been introduced and not reported out of committee is the ordinary state of most bills, and it is why this class exists separately from legislation — there is no signing date, no effective date and no veto list, and there may never be any.

The channel it acts on is not the channel the formula runs through # ↑ Contents

Ohio funds traditional districts through the foundation formula and funds expansion scholarships from a separate appropriation, and since the Fair School Funding Plan the two no longer meet in a deduction against the resident district. A district’s foundation payment can therefore rise while its net position worsens, and a change to who may receive a scholarship moves the second of those and not the first. verified

What this repository computed # ↑ Contents

Not what Ohio publishes. Figures derived here from committed fixtures, each one citing the test that reproduces it.

Nothing. One provision, none priced, and the run says so rather than saying zero.

$ edfund-project --draft hb-643-136-introduced

DRAFT — hb-643-136-introduced
  provisions                        0 of 1 priced, 1 not

POLICY EFFECT — none computed
  No provision of this draft binds a lever, so there is no figure to give.
  That is not a cost of zero. It is the absence of a cost.

verified crates/project/tests/a_draft_cannot_hide_what_it_did_not_price.rs

This bill found a defect in the machinery built to hold it # ↑ Contents

A draft whose every provision falls outside the model produces a combined policy identical to current law, so the arithmetic gives exactly 0.00. The first version of Priced::cost returned a bare figure and would have published that zero — and $0.0M in a cost column reads as this bill is free, which is the opposite of this bill was not priced.

cost() and residual() return Option for that reason, the CLI prints the absence rather than the number, and the feed writes null rather than 0. The shape is not exotic: it is the ordinary one for a bill about vouchers, transportation or facilities, which is most of what touches school funding outside a budget act. verified

What the retrieval established, beyond this bill # ↑ Contents

The as-introduced version code is always 00_IN, because introduction is always a bill’s first version. That makes it simpler than the enrolled case, where the code is positional and differs per bill — 06_EN, 05_EN, 08_EN across the three acts Ohio session laws — the appropriation acts themselves carries — and a wrong guess returns a nine-byte 404 that reads exactly like “not served”. verified

And the listing endpoint cannot be used to tell a pending bill from an enacted one. It reports a bill’s first version rather than its current one, so H.B. 186 of this same General Assembly appears there as As Introduced and is in fact enrolled with an effective date of 20 March 2026 — signed 19 December 2025, re-checked 9 September 2026. Only the per-bill version index distinguishes them, and reading status off the listing would have recorded several enacted acts as proposals. verified

The version index carries more than versions, which this node used to deny # ↑ Contents

Two of this node’s open questions were open because the index was read as a list of documents. It is a record, and it advertises its own continuations: meetings/ and amendments/, plus the sponsor roster with an active flag, the subject taxonomy, and an official local_impact_statement. Both continuations answer for this bill, and the answer is [].

An empty array is only evidence if the endpoint fills. It does: H.B. 96 returns 99 meetings and H.B. 186 returns 14, against H.B. 643’s zero. That control is what turns “no committee history was retrieved” into “no committee history exists”, and without it the two are the same observation. verified

A pinned digest moved while the document did not # ↑ Contents

00_IN is the as-introduced version and cannot be amended, which is why this source was pinned at all. Re-fetched three weeks later it hashes differently and is 11 bytes longer, and edfund-connect verify reports “The publication was revised — rebuild the fixtures and read the diff before committing it.”

It was not revised. The department’s exporter went from LibreOffice 24.2.7.2 to 25.8.6.2, which reorders CSS properties inside style attributes and fills in an empty <title>. Stripped of markup the two documents are byte-identical across 3,798 characters. verified

Two things follow. A digest over office-suite HTML pins the renderer as much as the document, so drift on this host is not evidence that a bill changed — the diff has to be read, which is what the tool’s own message asks for and what this node records having done. And nothing here would have noticed on its own: fetch is the only command that touches the network and verify compares the cache against the manifest, so an upstream change is invisible until somebody re-fetches. verified