Reading an amending act

“The acts themselves” wired H.B. 215 and H.B. 282 and deliberately left two acts alone. H.B. 215 appropriated FY1999 as a single undifferentiated line — the whole year in 200-405, against a promise to itemize it by January 1998 — and H.B. 650 and H.B. 770 kept that promise. Neither was read, for a stated reason:

Both print every amended row twice — the struck figure inline and the inserted one on a continuation line beneath — and H.B. 650 reprints H.B. 215’s fund-group totals unchanged, so the reconciliation this reader relies on would pass against the superseded number. A wrong answer that satisfies the check is the worst property a source can have.

So FY1999 sat in the fixture as fifty-one zeros and one lump, and the four pre-2002 years stayed out of the published feed because charting that draws fifty-one programs going to zero.

Wire H.B. 770 and not H.B. 650, and make the vintage explicit rather than implicit.

H.B. 770 reprints Section 50 as already amended by H.B. 650 and then amends it again, so its columns carry everything H.B. 650 established without the hazard that kept H.B. 650 out. Both H.B. 215 and H.B. 770 are held; session_laws::OPERATIVE names which act answers for which year, and every total reads through it.

Put the four years in the feed. Contract 33.0.0, and source gains a third value, act.

The amendment is positional, not typographic, and that is why this was tractable. Strike-through does not survive into the PDF’s text layer and does not need to. The row line carries the figures as they stood; H.B. 770’s replacements are printed beneath, right-aligned under the column each replaces:

GRF 200-100   Personal Services   $  10,744,925 $  10,756,210
                                                   11,256,210

So a replacement is a line of nothing but digits whose numbers end where the row’s amounts end. That last condition is what keeps page furniture out: the acts print a bare page number on its own line, and 247 under a row whose columns end at 59 and 74 amends nothing. No blacklist, no page-header rule.

The real hazard was not the one the previous phase named. A replacement can sit two lines below its row, after the wrapped second half of its own title. 200-545 Vocational Education Enhancements is amended from $184,298,314 to $201,991,432 that way. A reader that only looks at the next line loses $17,693,118 — and it was the reconciliation that found it, not inspection.

FY1999 has two enacted readings and they are not close. Net of tax reimbursement it is $6.15bn as H.B. 215 passed it and $6.33bn as H.B. 770 left it, and the composition differs by about fifty line items: 200-405 struck to zero, 200-501 Base Cost Funding restored from $0 to $2,986,915,811, 200-504 Special Education and 200-507 Vocational Education consolidated away into new numbers. “As enacted” is not a single thing for this year, which is why OPERATIVE exists rather than a rule about which file was read last.

The series is continuous FY1998-FY2027 and joins without a seam. Enacted, net of reimbursement: 5.91 → 6.33 → 6.84 → 7.24 → 7.75bn into FY2002. The formula’s own share: 2.79 → 3.65 → 4.13 → 4.46 → 4.88bn. No step at the General Assembly boundary and none at the change of publisher, which is the evidence that the acts and the workbooks are counting the same thing.

One line is renumbered twice in four years. The lottery half of the formula is 017 200670 School Foundation - Basic Allowance in FY1998, 017 200610 Base Cost Funding in FY1999, and 017 200612 from FY2000 — same money, same fund, three numbers. The FY1999 identity is confirmed from outside the acts: the greenbook’s FY1999 actual for 200612 is $666,093,028, which is what H.B. 770 appropriates 017 200610, to the dollar.

And 200610 is not a key on its own: 454 200610 is Guidance and Testing in the same tables, about half a million a year. session_laws::FOUNDATION keys on fund and number together for that reason, with a test that fails if the colliding line disappears — because the day it does, the test stops guarding anything and should say so rather than pass quietly.

The act’s own footing is wrong in one place and is resolved rather than tolerated. H.B. 770’s FY1999 grand total omits the Education Improvement Fund’s $1,443,401 while its FY1998 total includes it. appropriation-lines.csv already carried 200689 Hazardous Waste Removal as a FY1999 actual of exactly that, from the greenbook — so the rows are right and the footing is wrong, and the allowance is a named entry of a fixed size rather than a tolerance. The list now has two entries and each is aimed at one of the two additions an act can get wrong.

Wire H.B. 650 as well, for completeness. Rejected, and the reason has not changed: it reprints H.B. 215’s fund-group totals unchanged, so the reconciliation would pass against the superseded number. Everything it establishes is legible in H.B. 770 in settled form. Wiring a source whose self-check cannot fail is worse than not wiring it.

Read strike-through from the PDF. Not needed, and it was the assumption that made this look harder than it is. Whether the strike survives into the text layer was never established because the positional reading made the question moot. Anyone reaching for it should check first.

Keep FY1999 as H.B. 215 enacted it and footnote the amendment. Rejected. It is the reading that was superseded before the fiscal year began, and a footnote under a chart showing fifty-one programs at zero is not a correction.

Drop H.B. 215’s rows now that H.B. 770 supersedes them. Rejected. The deferral is the finding — a legislature appropriating a year’s money while declining to say what it is for — and it is only visible by holding both readings. deferred_years() reports it and moves no total.

Fold the act years into appropriations::enacted_lines so there is one series. Rejected. The acts are a different publisher reading a different document, and source: "act" on the row is what tells a reader that. The join is made once, in bundle, where the other two publishers are already joined.