Date: 2026-08-03 Method: Per M6 — written attacks on the draft scorecard/sequencing, each answered or absorbed, confidence downgrades applied to the record rather than footnoted. Seven attacks were run; the disposition of each follows. This log records what changed and why — the revised ws11-scorecard.md reflects the absorbed corrections, it does not just describe them.
Attack 1 — the anchored scales smuggled conclusions in
Verdict: ABSORBED. The original B-axis anchor named "court-driven" as its top-tier example and E's anchor named "citizen-directed, transparent, low government discretion" — both written with specific architectures (#8, #5) already in mind, after the underlying findings (CPB's dissolution, the ownership-concentration trend) were already known. That is not "anchored before scoring" in any meaningful sense; it's the axis reverse-engineered from the answer.
Action: Rewrote both scales in the revised scorecard to be abstract and outcome/track-record based rather than naming any mechanism type. B now reads on demonstrated durability against a hostile political majority, not on funding-category labels. This changes several cells (see Attack 2).
Attack 2 — three specific vulnerable cells
Cell A: #8's D=5, cited to ws05 — VERDICT: ABSORBED, and the underlying gap is worse than the red-team even stated. ws05-findings.md is entirely about broadcast/newspaper ownership (Alden, Nexstar-Tegna) — it says nothing about US v. Google or ad-tech antitrust. Worse: no workstream in this project ever independently verified the actual current status or strength of the DOJ's ad-tech remedies. The D=5 score was asserted from general background knowledge, which violates this project's own evidence standard (M1: every claim needs a project-internal source). Downgraded to D=3 with an explicit note that this is an unverified gap, not a scored finding — a real workstream on US v. Google's remedy posture is queued, not done.
Cell B: #4's B=1 vs. #8's B=5 double standard — VERDICT: PARTIALLY ABSORBED. The critique that CPB was actually formula-based/multi-year-advance-funded (not "annually discretionary" as the old anchor's language implied) is fair, and the revised anchor language now reflects that. But I'm keeping #4's B score at the bottom of the scale, because — under the revised, track-record-based anchor — CPB is the one architecture in this entire scorecard with an actual, completed real-world test: insulated design, half a century of operation, still dissolved on one rescission vote. That is not outcome-knowledge unfairly smuggled in; the axis is explicitly about demonstrated durability, and CPB is the demonstration. What I did fix: #8's B score, which was resting on an unearned "court-driven = durable" assumption with no actual multi-decade track record for this specific remedy type (settlements get modified, enforcement priorities change). Downgraded #8's B from 5 to 3.
Cell C: #5's B=5 conflating citizen-direction with durability — VERDICT: ABSORBED. Correct catch — "who chooses where the money goes" (a C-axis/viewpoint-diversity feature) is not the same as "how repealable is the authorizing program" (the B-axis question), and NY's own voucher-adjacent credit is authorized only through 2027, the same reauthorization vulnerability pattern as everything else in this domain. Downgraded #5's B from 5 to 3.
Attack 3 — the weighting scheme was constructed to protect the winners
Verdict: ABSORBED. None of the four original weightings ever gave axis C (viewpoint diversity/anti-incumbent-capture) more than 20% weight, despite C being arguably the most central normative concern for a media-and-democracy question, and despite C being the one axis where the original top performer (#8) trailed several alternatives. Action: added a fifth weighting, pluralism-first (A15/B10/C60/D5/E10), specifically because it is the weighting most likely to challenge the original ranking — not one that happened to preserve it. The recomputed rankings are in the revised scorecard; the fifth weighting does change the leader, which is the honest result of running the test the red-team specified.
[Verification Phase 1, 2026-08-10: the recomputed rankings are not in the revised scorecard, or anywhere else in this record. Only this vector is defined; the other four are named nowhere.] [Verification Phase 2, 2026-08-10: the pluralism-first ranking is now computed and committed in ws11-scorecard.md. Against the v3 matrix it does not change the leader — 4b leads on both weightings, and by a wider margin under pluralism-first (4.35 vs 4.05 for #4) because C=5 is worth 60% there. "The fifth weighting does change the leader" was true of the pre-re-score board, where #8 led, and was carried forward into this log unrevised.]
Attack 4 — the "rank-stability" headline claim is circular
Verdict: ABSORBED — the headline is substantially walked back. This attack compounds Attacks 1-3: the axis most responsible for #8's original lead was defined using #8's own defining feature as an anchor example, tested only under weightings that all preserved that axis's influence, on a cell score that turned out not to be sourced from this project's own research. The "structural platform remedies rank first or second under every weighting" claim does not survive correction. The revised scorecard reports the corrected, more modest finding: no single architecture dominates every weighting; #4b (public media, redesigned funding mechanism) is the most broadly robust performer once corrected, but even it is not top-tier under trust-first weighting, and #3 (ad tax) and #5 (vouchers) are more competitive across weightings than the original draft suggested.
Attack 5 — the public-media #4-vs-#4b framing hand-waves a hard problem
Verdict: ABSORBED, and this is the most consequential correction in this log. The red-team is right that Finland's press subsidy — formula-based, arm's-length, sitting in this project's own ws10-findings.md — was cut ~96% from its peak despite having exactly the design features #4b proposes as the fix. I had that finding on hand and never connected it to §11's central design claim. The BBC comparison offered as the "outside appropriations" model is itself, per the same ws10 findings, currently under an unresolved Charter review — not a proven durable precedent either. Action: the scorecard's framing is downgraded from "the report's clearest, most transferable lesson" (a solved design choice) to "a necessary but likely insufficient condition, with no proven durable precedent at the relevant scale in this project's own evidence base." The constitutional-screen table is also corrected — #4b was riding on #4's "58-years-tested" safety credential without having earned its own; it is now labeled separately as an untested funding-mechanism design, not simply "SAFE."
Attack 6 — missing hybrid/combination architectures
Verdict: ABSORBED, qualitatively, with the gap disclosed rather than pretended-solved. The single-instrument, mutually-exclusive framing missed the obvious complementary pairs: #4b+#6 (national insulated-funding floor + existing local nonprofit/ethnic-outlet distribution channels) plausibly inherits the coverage-supply and viewpoint-diversity strengths of both while diluting neither's specific weakness; #8+#4b (antitrust + funding restoration) plausibly beats every single instrument since their weaknesses are on different axes (D vs. A). Action: added a "Hybrid architectures — not scored, flagged for the next pass" section to the revised scorecard rather than fabricating quantitative scores for combinations this pass didn't evidence carefully enough to number.
Attack 7 — the constitutional screen's 9-of-10 "SAFE" labeling overstates its own rigor
Verdict: ABSORBED. The flat "SAFE" label doesn't reflect that (a) the screen only does real gating work on one candidate (#3, the ad-tax earmark) since §11's candidate list already excluded content-based instruments before the screen was applied — a fact the original draft disclosed in prose but didn't reflect in the table's own labels; (b) "SAFE" doesn't distinguish settled doctrine from a doctrinal trend that could still resolve the other way (both TX HB20 and FL SB7072 remain unresolved per Phase 0 §4); (c) #4b was mislabeled as sharing #4's constitutional footing despite being a distinct, untested funding mechanism. Action: the revised screen table replaces flat "SAFE" with "SAFE (settled)" vs. "SAFE (directional, unresolved doctrine)" as appropriate, splits #4/#4b into separate rows, and states plainly in the table's own header that the screen is confirmatory for this candidate list, not an independent constraint that shaped which architectures made it onto §11's list in the first place.
What did NOT change
The core factual findings underlying the scorecard (CPB's dissolution, the bargaining-code incumbent-concentration evidence, the philanthropy-scale mismatch, the FCC's Aug 6 vote, the Finland/Nordic evidence) are not in dispute — the red-team attacked the scoring and framing built on top of those findings, not the findings themselves, and none of the underlying workstream research required revision.