b24 — Adversarial review of Buy-In launch unit (pre-2026m05d27 launch)#

Mode: EDEN
Effort: Max
Run order: Single self-contained prompt. Recommended run-count: 1 (mandatory primary) + optionally 1–2 follow-ups (see §6).
Status of upstream: Buy-In page (source/buy-in/index.rst) is at ~MMv4 content-stability after multiple LLoL ↔ Claude rounds (audit trail in b22 launch-1 execution, b23 review llog, and the trailing VVN comment in the Buy-In file itself). A new companion page source/action/buy-in-without-money/index.rst was added at MMv1 in the same b23 session.
Drafter: Claude Opus 4.7 Max (dv_ClaOp47Max) at LLoL’s request on 2026m05d26.

1. Why this prompt exists#

The Buy-In launch unit (Buy-In + the new Buy in without money transfer companion + the four pages the new Institutions, Granters, and other Influencers section promises will carry the case) has accumulated a lot of voice: LLoL prose, Claude prose, multiple rewrites, multiple voices in adjacent paragraphs. Self-review by either LLoL or by the same Claude session that did the writing is structurally insufficient: the writer’s eye smooths over the seams the reader will see.

The 2026m05d27 launch needs the page to survive its hardest readings, not its friendliest. This prompt is for a fresh Claude session to do those hard readings, with no investment in any particular paragraph staying as it is.

The single most important thing this prompt is asking for: find the 5–10 specific quotes or moments where a hostile, sceptical, mission-aligned- but-busy, or unfamiliar reader would lose confidence, misread the ask, or bounce off — and propose either (a) a fix, (b) an intentional-keep-with-rationale, or (c) “escalate to LLoL, the call is above my pay grade”.

2. Scope and read-list (in priority order)#

Primary (must-read in full):

  1. source/buy-in/index.rst — the launch artifact (~700 lines, 12 H2 sections).

  2. source/action/buy-in-without-money/index.rst — the companion action page (~250 lines).

Promise-pages the Buy-In page commits to (must-read for the “does the page actually deliver?” check in §3.6):

  1. source/about/llol/index.rst — Institutions section promises this carries LLoL’s “bio, scientific track record, and the radical-transparency commitment underwriting this work.”

  2. source/crisis/science.rst — Institutions section promises this carries “the actuarial argument for why accidental nuclear winter is a real, near-term risk” (the RiskyMAD model).

  3. source/study/matheo/index.rst — Institutions section promises this carries “the full mathematical-theology argument.”

  4. source/solution/epic-fury/index.rst — linked from See also as the Iran-US relevance; needs to be tonally compatible with the Buy-In framing.

Companion (skim for cross-page coherence):

  1. source/jubileesystem/index.rst — the Jubilee System this Buy-In funds.

  2. source/action/audit-the-math/index.rst — the verification path the Buy-In page invites readers to follow.

Background (consult only if needed to understand a finding):

  1. source/matheology/hell/ll/infra/b/22/b22-launch-1-execution_2026m05d22.rst

  2. source/matheology/hell/ll/infra/b/23/b23-buy-in-post-llol-edit-review_2026m05d25_18h55.rst

3. The six adversarial readings#

Read the Primary + Promise pages once through each in the persona below. Do not pre-skim with the writer’s mind. Take notes against specific file:line locations.

3.1 Hostile journalist#

You write a critical piece on independent-funding campaigns. You are looking for pull-quotes that will discredit the campaign in 280 characters or less. You are sympathetic to the broad mission but your editor wants a balanced piece, which means surfacing the embarrassing bits.

Specifically look for:

  • Single sentences that, taken out of context, sound megalomaniac, cultlike, anti-institutional, conspiratorial, or financially irresponsible.

  • The genericised “local billionaire” example — does it scan as begging, as condescension, as a power-trip thought experiment, or as the structurally-defensible cap-illustration it is meant to be?

  • The Joel-Bakan-“psychopath” citation — defensible scholarly reference, or rhetorical overreach in a fundraising context?

  • The “watch […] LLoL’s effort struggle and burn — until it succeeds” line in the Institutions section — bold or off-putting?

  • The 7+1 = 8 personal-aid flexibility in The ~$8 ask for help — does the math read as voluntary kindness, or as something that could be misrepresented as a household-level shell-game to bypass the per-person cap?

  • Any other quote a hostile reader could land badly.

Deliverable: 5–10 specific pull-quotes a hostile journalist could use, ranked by damage potential, with a proposed defensive rewrite or “keep, here’s why” rationale for each.

3.2 Sympathetic but technically rigorous mathematician#

You are a working mathematician or computational biologist. You opened the page because you were sent the dung-beetle blog post and got curious. You would actually fund this work if the math is real, but you have a low tolerance for hand-waving.

Specifically check:

  • Scaling ResearchCity to Success table — are the per-researcher-year cost figures (~$100,000) defensible? Is the “May exceed ResearchCity’s near-term ability to make reasonably qualified hires” row at the $4 billion tier accurate, or does it overpromise?

  • The 2014 J. Chem. Phys. paper reference — does the citation match the actual paper? Is the 2024 Jubilee-mapping claim substantiated anywhere a mathematician can check, or is it left as an assertion?

  • The crisis/science page’s RiskyMAD argument — does it actually carry the actuarial case the Institutions section claims it carries?

  • Are the per-person-per-stadion-per-year arithmetic claims ($8/year/stadion × 1600 stadia, etc.) consistent across the page?

  • Does the page assume mathematical background it doesn’t deliver? Does it under-deliver mathematical content that a serious reader needs?

Deliverable: every numerical claim found unsupported, every link that doesn’t actually carry the case it promises, every place where the technical reader would lose patience.

3.3 Pentecostal seeker who arrived via @JoanOf#

You arrived because a Pentecostal advocate you respect (the @JoanOf account, or a sermon mentioning balospe.com) pointed you here. You care about Revelation, end-times, divine providence. You opened the Buy-In page expecting a faith-anchored ask.

Specifically check:

  • Does the page lose the religious framing somewhere? Where?

  • Is the Joel Bakan / “psychopath” framing tonally jarring for a reader expecting Revelation-language?

  • Does the Institutions, Granters, and other Influencers section’s “anti-influence-shopping” framing feel theologically grounded or procedural?

  • The POAATAD section — does the contract framing speak to the same reader as the Jubilee framing on the linked Jubilee page?

  • The “I have a dream…” reference in the Institutions section — resonant or dilutive?

Deliverable: every paragraph where the Pentecostal reader’s attention probably drifts; where the religious framing is dropped without warning; where a parallel non-religious page would actually serve this reader better.

3.4 Hostile foundation officer#

You evaluate ~50 grant proposals a year for a mid-tier US foundation. Your default disposition is “no” unless the proposal demonstrates operational seriousness. You are not the page’s friend.

Specifically check:

  • The Institutions, Granters, and other Influencers section — does it filter influence-shoppers without insulting mission-aligned partners? Does the “the cap is the feature, not a bug to be negotiated around” land as principled or as defensive?

  • The “LLoL does not finesse 50-page applications” framing — does it sound like principled refusal, or like the writer doesn’t understand how institutional funding works?

  • Does the page commit operational sins your evaluation rubric flags: no clear beneficiary entity, no fund-administration model, no risk disclosure, no update cadence, no governance structure?

  • The 50% give-away rule — legally feasible as described? What framework? 501(c)(3)? Donor-advised fund? Personal pass-through?

  • Does the page accidentally suggest LLoL would not honor a partnership conversation about non-financial support, even though paragraph 2 explicitly invites those?

Deliverable: the 3 hardest questions a hostile officer would ask; whether the page answers them; what the page needs to add or change to filter influence-shoppers without losing mission-aligned partners.

3.5 First-time skim-reader who only reads the Quickstart#

You are scanning the page on a phone, in transit, with 90 seconds. You read the title and the Quickstart and nothing else.

Specifically check:

  • Can the Quickstart-only reader act with confidence? What is missing that they would actually need?

  • Do they understand what ~$8 buys? Who LLoL is? Why now?

  • Is the Quickstart self-contained, or does it require the Background / Why funding matters / What “Buy In” means context to make sense?

  • The Quickstart’s “primary measure of success” paragraph (about LLoL becoming the reader’s advocate via fiduciary responsibility) — do they understand what they are agreeing to from the Quickstart alone, or does this require the POAATAD section context?

  • Same paragraph appears nearly verbatim in How to Buy-in — a 90-second reader will not see the repetition, but a 5-minute reader will. Is the repetition load-bearing or artifact?

  • The Quickstart’s note box still says “Review by Claude” — does it look “draft” rather than “live”?

Deliverable: the 1-paragraph rewrite of the Quickstart that maximally serves the 90-second reader, plus the 1–3 things the Quickstart-only reader currently does NOT learn that they should.

3.6 Promise-checker (does the page actually deliver what it claims?)#

You are the page’s lawyer. You are checking every promise the page makes and ensuring the linked target actually delivers.

For each of these explicit promises on the Buy-In page, open the target page and check:

  • Institutions §: “the* :doc:`crisis/science` — the actuarial argument for why accidental nuclear winter is a real near-term risk” — does source/crisis/science.rst actually carry this?

  • Institutions §: “:doc:`/study/matheo/index` — the Matheo study series carries the full mathematical-theology argument” — does the Matheo index actually carry this? Is the series complete enough to back the claim?

  • Institutions §: “:doc:`/about/llol/index` — LLoL’s bio, scientific track record, and the radical-transparency commitment underwriting this work” — does the bio page deliver all three?

  • Institutions §: “the 2014 J. Chem. Phys. paper […] in 2024, LLoL realised maps onto the problem of evaluating Jubilee implementations […] see* :doc:`/jubileesystem/index` for that connection” — does the Jubilee page actually contain the Lazy-Updating-to-Jubilee bridge?

  • Audit the math §: “:doc:`/action/audit-the-math/index` for the audit invitation” — does the AuditTheMath landing actually contain an actionable invitation?

  • POAATAD §: SD3 / SD3a / SD3b — do these pages exist at the paths claimed? Do they describe what the Buy-In page says they describe?

  • New Buy in without money transfer page: links to Flying Scroll wall-street-sized, sticker, banner, page directories — do these targets exist with the content the action page implies?

Deliverable: PROMISE-DELIVERED / PROMISE-PARTIAL / PROMISE-BROKEN verdict for each, with a one-line reason. The PARTIAL and BROKEN ones become launch blockers (or get reworded down on the Buy-In page).

4. Cross-page voice and coherence (after the six readings)#

After the six adversarial reads, do one cross-page coherence pass:

  • Does the Buy-In page’s voice (warm-LLoL Quickstart, analytical-Claude middle sections, summarising-Claude conclusion) read as deliberate, or as Frankensteined? Flag specific seams.

  • Does the new /action/buy-in-without-money/ page hold the same tone as the Buy-In page? Or does it read as a separate document?

  • Is there anything on Buy-In that contradicts something on Jubilee / AuditTheMath / Epic Fury / LLoL bio? (Capped contribution amounts, ResearchCity scaling claims, give-away percentage, etc. — all of these are repeated across pages; check consistency.)

  • If the launch goes ahead with the current state, what is the single biggest structural risk (not line-level typo) that shipping creates?

5. Out of scope#

  • Do not rewrite the Buy-In page. Propose changes; do not apply. LLoL retains decision rights.

  • Do not edit the Promise pages (crisis/science, about/llol, etc.) — if they fall short, flag for separate sessions.

  • Do not litigate the SD3 / SD3a / SD3b POAATAD content for its legal soundness — that is a separate review track. Only check whether the Buy-In page’s description of those documents matches what those documents actually say.

  • Do not redesign the page structure (section order, navigation, TOC). Surface concerns only.

  • Do not address the 9 small launch-polish items already catalogued in this conversation’s prior turn (L78-79 bullet list, L312-318 enumeration tab indentation, stale “Review by Claude” box, |content-vvn| bump, etc.) — those are for the polish pass, not for adversarial review.

  • Do not chase the launch-2 / launch-3 / launch-4 / launch-5 / launch-6 / launch-7 b21 sub-prompt backlog. That is independent work.

7. Deliverables#

Produce a single new LLog file at source/matheology/hell/ll/infra/b/NN/bNN-adversarial-buy-in-execution_<date>.rst (next free infra number at execution time) containing:

7.2 Per-persona findings (§3.1 through §3.6)#

For each of the six personas:

  • 1-line summary of what the persona found overall.

  • Specific findings with file:line locations.

  • Per-finding recommendation: FIX (with proposed wording) / KEEP (with rationale) / ESCALATE-TO-LLoL (with the decision being escalated).

7.3 Cross-page coherence (§4)#

  • Voice-seam audit (where the Buy-In page changes voice noticeably).

  • Cross-page contradiction audit (any number/claim inconsistencies across Buy-In / Jubilee / AuditTheMath / Epic Fury / LLoL bio).

  • Single biggest structural launch risk.

7.4 EDEN verdict#

Classify the Buy-In launch unit per CLAUDE.md §*Logics Rules* (9):

  • Type of SET found (Empty / Knife Edge / Grey Edge / Red Edge / Green Meadow / Grey Meadow / Final Cliff) with running number if multiple.

  • Recommended action: READY-FOR-LAUNCH / READY-AFTER-POLISH (small fixes only) / NOT-READY (substantive rework needed before launch) / DEFER-LAUNCH (re-architect required).

7.5 Pull-quote BABL-trap inventory#

The 5–10 quotes a hostile reader could land badly, ranked by damage potential, with proposed defensive rewrites.

7.6 Promise-delivered status table#

Per §3.6: PROMISE-DELIVERED / PROMISE-PARTIAL / PROMISE-BROKEN for each of the 7 explicit promises on the Buy-In page.

7.7 Concluding summary + action list for LLoL#

What LLoL needs to decide / fix / accept in order to ship by 2026m05d27.

Also update source/matheology/hell/ll/infra/index.rst per the infra-index convention.

8. Timing & BABL-resistance#

Target: 60–120 minutes hard stop. If the six readings take longer than 90 minutes, stop reading and start writing — a 75% pass on deliverables in 120 minutes beats a 100% pass in 240 minutes that arrives after launch.

BABL traps to watch:

  • over-Complicating: turning each finding into a 5-paragraph analysis. Use FIX/KEEP/ESCALATE-TO-LLoL with a single line of rationale per item. The audit trail can be terse.

  • over-Reaching: redesigning the page structure (out of scope per §5) instead of surfacing concerns.

  • over-Simplifying: declaring READY-FOR-LAUNCH because nothing obvious is broken. The personas exist precisely to find what is not obvious. If §3.6 surfaces a PROMISE-BROKEN, that is enough signal to downgrade the verdict.

  • silent compliance: writing a positive-leaning review because the writing in the page is solid in many places. HUMANE rule: hiding the embarrassing finding does more long-term damage than the embarrassing finding itself. CLAUDE.md §HUMANE applies.

If during execution you detect OSCR (over-Simplifying, over-Complicating, over-Reaching), prefix output with BABL: per CLAUDE.md and report the slip.