School improvement connector

Written after the fact. The connector was declared and wired in 08488b0, and no decision record named it until “What approves a connector” found four connectors in that state. What follows is reconstructed from that commit and from the registry.

“Accountability domain” brought the federal regime into the corpus: the identification categories, the thresholds that trigger them, and the interventions that follow. All of it was the rule and none of it was the roster. The corpus could state what makes a school a Comprehensive Support and Improvement school and could not say whether any particular school was one.

The lists are published, and finding them took reading the department’s page rather than inferring a filename. Six probes against the report card storage all 404’d; the links sat on Identification-and-Requirements the whole time.

Declare dew-school-improvement as a connector in its own right, Wired with a recorded remainder, feeding school, education-agency and accountability-regime, with the three current identification workbooks as its sources: 231 CSI, 117 ATSI, 60 TSI.

Separate from dew-report-card although both are the department and both are accountability. The report card is a graded measurement republished annually with its history intact; this is a status list republished in place. They fail differently and a shared status could not say so.

The corpus can answer a question it had recorded itself unable to answer. Sheridan High School — Nathan DeRolph’s building — was ATSI in 2022 and escalated to CSI in 2025. Its node said one commit earlier that the corpus “cannot say whether Sheridan has ever been identified”; it now says what changed rather than reading as though it always knew.

And the route is the finding, not the fact. A four-star building with a Performance Index of 89.9 did not reach CSI by performing in the lowest 5%. It was ATSI — one or more subgroups at or below the CSI threshold — and did not exit within three years, which escalates automatically. Forty-two of the 231 CSI schools arrived that way. The whole-school figures and the status are not in tension; they measure different things, and the federal system exists to surface that gap.

The corpus’s worst-performing exemplar is on no list, and the corpus cannot say why. Anton Grdina, Performance Index 36.8, two in three students chronically absent, is not CSI, TSI or ATSI while Sheridan is. CSI selects among Title I served schools and nothing held here carries a Title I flag, so the absence reads three ways and none can be ruled out. What can be said is that a reader treating the CSI list as “the failing schools” will not find this building on it.

A remainder that is the publisher’s archiving practice. The department republishes each list in place under a dated filename rather than archiving prior cycles, so there is no history here: a school that exited before this file was written is indistinguishable from one never identified. Some CSI rows still carry a 2018 identification against exit criteria that allow three years.

The comma guard earned its keep. Year Identified is not always a year — for escalated schools it carries a history, ATSI 2022 , CSI 2025 — and the writer refused it on the first build. That cell is now the only place the escalation path is legible per school.

Fold the lists into dew-report-card. Rejected, above: same publisher, opposite archiving behavior, and one still_blocked string cannot describe both.

Wait for an archived cycle before wiring anything. Rejected: the snapshot is the only thing the department publishes, and refusing it would have left the accountability nodes unable to name a school for as long as the practice lasts, which is indefinitely. The limit is recorded on the connector instead of being the reason not to have one.

Infer Title I service from the CSI list itself. Tempting, and it would explain Anton Grdina in one line. Rejected: it assumes the conclusion. A school absent from a list that selects among Title I schools tells you nothing about whether it is one, and deriving the flag from the selection would make the absence self-explaining and unfalsifiable.