b21 launch-1 — Buy-In + AuditTheMath streamlining decision#

Mode: EDEN
Effort: Max
Run order: 1 of 7 in the b21 launch-sprint sub-prompt set.
Status of upstream: This is the launch blocker identified in b20 review llog §7-4 (F-1). Until this is run, the launch unit cannot complete the reader’s funding action.
Drafter: Claude Opus 4.7 Max (dv_ClaOp47Max) at LLoL’s request on 2026m05d22.

1. Context#

The b20 review (full text at the doc link above) found the Buy-In page (source/buy-in/index.rst, 73 lines) is ~70% placeholder:

  • No GoFundMe URL (3 placeholder comments at lines 38, 70, 73).

  • No concrete contribution amount.

  • No $8 / person / year figure.

  • No ResearchCity mention.

  • No SD3a / SD3b reference (framework referenced in the b19 prompt but source-of-truth not located in Discovery).

  • No reverse link to AuditTheMath.

The reader-journey question that goes with the Buy-In fix: the current 4-page funnel (dung-beetle → AuditTheMath → Matheo overview → Buy-In) may be too forking for the 99% of readers who will not audit but should still be invited to contribute. LLoL has voiced both “audit-first” commitment AND “break-the-web streamline” aspiration; the resolution is in this prompt.

2. Tasks#

2.1 EDEN-classify two candidate paths#

Apply the 7-fold gauntlet (CLAUDE.md §*Logics Rules*) to:

  1. Current 4-page path: dung-beetle → AuditTheMath → Matheo overview → Buy-In → external GoFundMe.

  2. 3-page shortest path: dung-beetle → Buy-In (with embedded scaling arithmetic + audit-the-math invitation) → external GoFundMe; AuditTheMath becomes a verification page reached from Buy-In, not the primary CTA.

Report SET-type per CLAUDE.md (9) for each path independently.

2.2 Steelman from each position#

For each path, steelman from:

  • 99-percent reader (won’t audit, may contribute).

  • 1-percent reader (will audit before contributing).

  • Hostile journalist looking for discrediting angle.

  • Sympathetic mathematician.

  • Policymaker controlling institutional funding.

  • Pentecostal seeker arriving via @JoanOf.

2.3 Decide and document#

Decide one of:

  • Keep current (Path A) — audit-first stays primary.

  • Invert (Path B) — trust-first becomes primary.

  • Hybrid — specify which page is primary CTA, which is verification, and what the new reader flow looks like.

Document the decision with explicit BABL-trap analysis for the losing path and ZION-narrow-path analysis for the chosen path.

2.4 Fix the Buy-In page (regardless of path chosen)#

Edit source/buy-in/index.rst to add:

  • Concrete contribution amount: $8 / person / year as the baseline. Tie to the ResearchCity scaling math.

  • GoFundMe URL: if LLoL has provided the URL, insert it. Otherwise: replace the 3 placeholder comments with ONE visible TODO admonition that names the specific date by which the URL will be added and the responsible party (LLoL). Do not leave invisible RST comments masquerading as a page.

  • Scaling arithmetic paragraph: explicit 1M × $8 = $8M = N researchers × R years arithmetic. Make N and R explicit; show the unit chain.

  • Cross-links back to: /action/audit-the-math/index, /blog/posts/2026-05-21-prophetic-dung-beetle, and /study/matheo/index.

  • Institutional / grant path: one short paragraph naming the channel for foundations / institutional donors (even if it’s just an email address). Closes the policymaker BREACH from b20 §3.3.

2.5 Locate the SD3a / SD3b framework definition#

Grep / search for SD3a, SD3b, Good News Pack, Supporting Document 3 across the source tree. Report:

  • Where the framework is defined (file path + section).

  • What SD3a and SD3b each are.

  • Whether the Buy-In page should reference them; if yes, add the references.

If the framework is not defined anywhere in the source tree, flag this explicitly and propose either (i) deferring the framework reference until LLoL drafts it, or (ii) drafting a minimal framework-stub for LLoL approval before applying.

3. Out of scope#

  • AuditTheMath landing content edits (already at MMv1 mockup; content edits are downstream).

  • b18a paper content.

  • Dung-beetle blog post content.

  • Matheo overview content.

  • Translating the new Buy-In content into the other 9 languages (separate task).

  • “Break the web” / viral-marketing copy framing — explicitly rejected per b20 §6.3.

4. Deliverables#

Produce a single new LLog file at source/matheology/hell/ll/infra/b/NN/bNN-launch-1-execution_<date>.rst (next free infra number at time of execution) containing:

  • EDEN classification of Path A and Path B.

  • Steelman results.

  • Decision (with rationale).

  • Diff of edits applied to source/buy-in/index.rst.

  • SD3a / SD3b framework location report.

  • Final ZION / BABL verdict for the streamlined Buy-In.

  • Concluding summary with what’s still pending.

Also update source/matheology/hell/ll/infra/index.rst per the infra-index convention (bump “next free number”, add heading + toctree entry).

5. Timing & BABL-resistance#

Target: 60–90 minutes. Hard stop at 120 minutes; report partial findings rather than spiral.

The BABL trap to watch: over-Complicating the decision. The 3-page vs 4-page question is a structural call, not an engineering optimization. If the EDEN analysis takes more than 30 min, the rest of the time is for applying the fix, not for re-deciding.

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