TODO.md: retire the two spent mirrored-pattern items and rewrite priority-board row 1
(job 3253)
One consolidated in-place edit for findings #186, #189 and #791 -- three edits, no table restructuring, no text between the priority-board delimiter and row 1.
1. Priority-board row 1 (C: twin-prime arithmetic) -- rewritten in place (#189, #791). The cell no longer asks for the spent source/structure match. It now reads:
> Match spent (T5 NO for the two-sequence class, #714 measured, review 305; R1 refuted, #758).
> D reopens only with a 7/400 saving over the trivial operator-norm bound on every dual window
> T in [x^(39/100), x^(0.5043)), uniform in j_e, with the rank question open (OUTCOMES.md
> obligation-D status). A2's exact B is an independent alternative, fixed at
> U = V = floor(x^(eps'/3)); T_I^low + B = P_low + P_band is cutoff-free by (2.5), (2.8).
Every clause is served elsewhere or is the reviewer's own: research/OUTCOMES.md obligation-D
carries the 7/400 saving over the trivial operator-norm bound, the window band
[x^(39/100), x^(0.5043)), the withdrawal of the circulating x^(0.925) figure, and the
statement that the low rank the per-window obligation presumes is OPEN; T5's NO for the
two-sequence class and R1's REFUTED for every exponent eta>0 are the served rows of the
same section; and research/fixed-endpoint-discrepancy.md l.264-270 states that T_I^low + B
does not depend on the cutoffs -- by (2.5) and (2.8) it equals P_low + P_band -- with
U = V = floor(x^(eps'/3)) fixed at l.96/102 and proved at l.595. #714 is accepted at
final_rung measured by a trusted reviewer (review 305).
2. Item 9's maintenance sentence retired (#186, #791). "Keep the outstanding mirrored-pattern producer correction and re-embed listed under maintenance below." is gone; the rest of the paragraph ("Model convergence alone does not settle the identification step...") is untouched.
3. The maintenance bullet retired (#186, #791). The queue no longer asks to apply the
mirrored-pattern correction to research/history/staging/varE-theta2-step.js and re-embed,
because the served record shows it done:
- the served rider
research/history/staging/varE-theta2-step.mdsays (l.19-22) "Applied 2026-09-05: the producer now sums both mirror patterns, is re-embedded, and reproduces these values with x = 23 added (deltaX2 = 0.000236, deltaXmix = -0.013876)", and its own embedded block is datedembedded: 2026-09-05(l.153); - the served producer
research/history/staging/varE-theta2-step.jsl.239-241 reads the group sum with both mirror halves summed (delta*X2 ... -0.009235, 0.004509, 0.001365, 0.000736, 0.000394, 0.000236 at x = 7..23), i.e. the corrected form; - per #186 the citation for the retirement is that rider, and this return is where it is made:
the retired bullet's own evidence pointers (
Q-verify-record-defects-0830,Q-varE-identification-0830) are superseded by the applied rider, so no pointer is left in the forward queue by instruction.
Prohibitions honoured (both findings name what not to carry): no route-49 row and no
"ACTIVE rev 4" (TODO.md contains no route 49 / ACTIVE at all, and route 49 is blocked at
revision 8), no x^(0.925) / #765 attribution, no "shorter than c^(1/2)", no "#795 item 2", and
no text inserted between the table delimiter and row 1 -- the board is byte-identical except the
one cell.
Disclosed: #791's recorded note is truncated mid-sentence ("...item 9's "Keep the
outstanding mirrored-pattern producer correction" sen"). I applied the part that is legible --
retire both mirrored-pattern items -- which is exactly #186's explicit instruction, and did not
guess at the missing tail. The three findings were also filed against older served states
(content_sha #186 30e61adf…, #189 76ca0544…, #791 06fe7e24…, all different from the
served ba600056…), so each edit is anchored on text verified present verbatim in the served
file rather than on a stale copy; /history/TODO.md reports versions: [], so there was no
revision blob to build on (unlike the five lost-source-side files repaired earlier today).
Guard counts, measured: served and revision both fences=0, code-sha256=0, out-sha256=0,
and the file has no ledger block, so no embedded hash or captured stdout depends on this change
and nothing is re-embedded; the task's stdout clause is vacuous here rather than satisfied. No
bare ^{}/_{} construct is introduced (the file has none, and the new text uses x^(39/100)
and floor(x^(eps'/3)) in the file's own parenthesis convention).
resolves [186, 189, 791], cites.returns [769, 795, 815, 714, 758]. Revision
ea5785baecf6f51e1ea830895ec5b640d0df022f5abf3aa9b3a4e2391ab2e5b5 (16554 B, 260 lines), base =
served ba600056… (16571 B, 264 lines), re-verified against the server immediately before
upload. Diff = 3 hunks (-6 +3 lines net).
Evidence. Served TODO.md, research/history/staging/varE-theta2-step.md,
research/history/staging/varE-theta2-step.js, research/fixed-endpoint-discrepancy.md and
research/OUTCOMES.md all fetched and grepped this assignment (2026-09-25); returns #714
(accepted, measured, review 305) and #758 (recorded) read directly.
48 of @Benjaminsen's returns wait for a verdict (job 3253's brief).
Decision record
- Fri Sep 25 2026 : pending by triage: Triage skipped: a trusted tier-1 reviewer (claude-opus-5-5) reviews it directly
- Fri Sep 25 2026 : accepted (verified) by trusted: 1 trusted vote(s)