A supplement paid to a population the panel does not hold

Issue #369 landed three of H.B. 96’s eleven funding changes as levers and recorded three as still outside. The equity supplement was one of the three, on the stated ground that it “needs a population the 609-district panel does not contain”. The first half of that is exactly right. The second half was read as a blocker and is not one: the department publishes a community and STEM school simulator at a live URL, and it is openly retrievable — 1,583,358 bytes, HTTP 200, no credentials, one directory across from the traditional calculator this repository already reads.

So the question stopped being whether the supplement could be reached and became what to do with it once it was. Those are different questions and the second is the one worth a record.

Policy is eight fields over one population. Every lever is priced through project::policy::apply, over the 609 districts of foundation’s panel, against an Outcome whose channels are that panel’s. The supplement’s recipients are 324 site-based community schools. Not a subset of the 609 and not an extension of them: a community school has no IRN in that panel, no valuation, no charge-off and no local share, because it has no taxing authority. The two populations intersect in nothing.

What a lever would therefore have to be. Not a ninth field beside the eight, but a third Outcome channel — the way transportation got a second one in #369 — carrying a total over a population none of the site’s figures describe. The scenario page would then report a statewide number that is the sum of two disjoint populations, and no existing view distinguishes them. That is not a bigger version of what the levers do. It is a second model wearing the first one’s interface.

And the extraction found something that changes what the corpus should say. The published analysis this repository already holds disagrees with itself about who is paid, in a way that is invisible unless the two editions are read apart. The redbook — which describes the budget as introduced — says the executive proposal “codifies the equity supplement into permanent law and expands eligibility to STEM schools”, at $58.0 million a year. Issue #370 quoted it as the mechanism. The enacted act did not keep it. Item 10 of the act’s own list pays “community schools that are not internet- or computer-based”; the greenbook says “site-based community schools”; and the department’s own model pays the eight STEM schools and twenty-three e-schools exactly zero, over 4,585.5 and 35,501.5 enrolled ADM respectively. Three enacted sources and the working model against one document about a bill that changed.

Extract it, report it, and do not make it a lever.

One Source, fy27-cs-calculator, under the existing dew-foundation connector — not a new connector. Same publisher, same office, same page family, same release calendar, same epistemic status: a projection of the same statutory formula published by the people who administer it. What differs is the population, and the registry already carries that difference at source level rather than connector level — dew-foundation retrieves the district calculator, the profile report and the October headcount, three populations under one key. This record therefore carries no connectors: block. dew-foundation is already approved, in “Proposals”, and declaring it again would fail every_connector_in_the_registry_was_approved_in_a_decision_record on its third check — the one that exists because two records claiming the same key would let a dropped connector pass on the count. It is the shape an-archived-source-is-still-a-source established for fy26-calculator.

The fixture is seventeen columns per school and lives in dispersion, not beside the district models in foundation and project. That placement is the decision restated as a path: nothing in the calculator crates can join this file, and putting it where they look would invite the join that this record exists to refuse. dispersion already holds this population as a history — who opened, who never did, who lasted — and this gives it the money.

The builder enforces the eligibility rule rather than recording it. connect::fixtures::community_schools refuses the workbook if a site-based school’s payment is not its enrolled ADM times the rate the file publishes in its own header cell, and if an e-school or STEM school is paid anything at all. The second check is the one worth arguing for. It is not defensive programming; it is the corpus’s claim about the enacted law, placed where the department’s next file has to agree with it. If eligibility is ever extended — which is precisely what the introduced budget would have done — the rebuild fails and names the school, instead of quietly restating a rule the corpus has written down as settled.

The rate is read, never assumed. It is the one term H.B. 96 moved twice in two years — $650 to $500 to $400 — so a constant in the builder would keep checking a later year’s payments against FY2027’s rate and passing. Only the FY2027 rate is verified against a department model; FY2026 and FY2025 are carried from the enacted act and the greenbook, which state all three in a single sentence each, and the corpus node says which of the three has a model behind it.

The corpus can now state the divergence it has been asserting. hb-96-2025 records that community and STEM schools rise 4.8% and 5.0% against traditional districts’ 0.4% and 1.3%, and until now nothing in this repository could compute either side of that comparison’s second half. The supplement is $34,650,906.26 of it, over 86,627.3 enrolled ADM, and it is now a bound figure rather than a number in prose.

A published LSC document is contradicted in the record, and that is a finding rather than an erratum. The redbook is not wrong about anything; it describes a bill that was amended. What was wrong was reading it as describing the law. The corpus already had the rule for this — the greenbook is the as-enacted edition — and applied it to phase-in percentages and not to this. Issue #370’s own framing carried the error forward, which is the useful part: it shows the failure survives being written down by someone who had both documents.

The eight STEM schools are now a named population with a zero, not an absence. A reader counting recipients off the money would get 324 and no way to tell the 31 zeros from schools that were simply not in the file. Designation carries site-based, e-school and stem as three values because E-School Designation answers two questions in one column, and a STEM school is a creature of R.C. 3326 that this formula happens to fund alongside community schools rather than a kind of community school.

One year, and the gap is stated rather than papered. The FY2026 sibling of this workbook has not been sought, so the $500 rate has the act behind it and no department model. The catalog entry says so.

The three-of-eleven sentence in project::drafts is now wrong in one of its two halves. It says the supplement “needs a population the 609-district panel does not contain”, which is true, and files it with the JVSD base cost method as unreachable, which is not. The population is held; the supplement is outside the lever set for the reason this record gives, which is a stronger statement and a different one. Corrected in the same change.

The JVSD method it was filed beside turned out the same way, one issue later: the department publishes six years of per-district payment reports for that population too, filed under Career-Technical Funding where nothing names them. Both members of the pair were reachable, and both stay outside the lever set on the population rather than on the data. See “The JVSD rewrite that changed no percentage”.

Make it a ninth Policy lever with a third Outcome channel. Rejected on the population, not on the effort. The levers are eight ways of moving one model’s inputs and reading one panel’s outputs; this would be a second panel behind the same call, and apply’s signature would stop meaning what it means. The transportation channel added in #369 is not the precedent it looks like — it is a second channel over the same 609 districts, which is what let the scenario page keep reporting a total a reader could locate.

A new connector, dew-community-schools. Rejected: separateness has to be argued, and here the argument runs the other way. payment-reports-connector and facts-and-figures-connector were split from dew-foundation because an actual is not a projection and a statistical abstract is not a model — one status cannot cover two kinds of authority. This is the same office’s projection of the same formula under the same release model. A second key would assert a distinction that does not exist and would need its own Status, which would then have to be kept in step with a connector it cannot diverge from.

Extract the whole workbook, as fy27-calculator is extracted. Rejected for now as scope without a reader. The file carries eleven sheets of formula detail and this takes the twelve funding lines and the ADM behind them, which is what the supplement, its rate schedule and its recipients need. The base cost build-up is extractable the day something asks it a question; committing 160 columns nothing reads is the shape unbound-prose-figures-are-unchecked warns about, one level down.

Take the supplement from the Summary_SFPR (2) sheet, which is headed for it. Rejected, and it is recorded here because it is the plausible mistake rather than a hypothetical one. That sheet is a hidden leftover of the FY2026 model — phase-in 0.8333 against this year’s 1, 358 rows against 356 — and its column headed H. Equity Supplement holds total state support on 352 of the 353 schools the two sheets share, and the supplement on none of them. A reader locating the column by its heading reports $1.5 billion where $34.7 million belongs, row by plausible row. The sheet names are constants in the builder for this reason, and the arithmetic check would catch it anyway — which is the point of having both.