b32 prompt (give-away) — Designing the 50%-give-away framing + process#
dv_ClaOp47Max, 2026m05d28_13h54. Self-contained.1. The decision this resolves#
LLoL weighed four ways to frame where the 50% give-away goes (full reasoning in the b32 llog, §”LLoL’s give-away options”):
(a) stay non-committal (“worthy causes”) — safe + non-divisive, but Popper-unfalsifiable; says nothing testable. LLoL leans against it.
(b) name recipients — concrete but maximally divisive, and any specific pick (incl. a Vatican AI-dignity effort) is speculative + an unkeepable public promise while no give-away infrastructure exists yet.
(c) a criterion + open propose/apply process — commit to a falsifiable principle without a divisive recipient. The launch already uses this spine: half goes to causes in **urgent need now* — not endowment-building or the already-secure* (LLoL’s “I won’t feed the banks’ endowments” rule).
(d) a deliberately balanced mix — a cross-spectrum set of examples so “everyone can see themselves in them while realising they are not the only ones who struggle.” LLoL wants to design this with Claude, here.
This session designs (c) properly and explores (d) carefully.
2. Design tasks#
The criterion, sharpened. State the urgent-need-vs-endowment principle crisply and defensibly, and show why it is on-brand (it is the Jubilee anti-hoarding ethic applied to giving — a differentiator, not a liability). Define the falsifiable test: would this cause have to cut services / people right now without help? vs is it building long-term security?
The open propose/apply process. Design the lightweight mechanism by which causes can be proposed/applied and evaluated (LLoL’s option c). What does a page look like? What does an applicant submit? Who decides, on what record? Keep it honest that this is aspirational until infrastructure exists.
Explore the balanced mix (d) — with LLoL, not for him. Help LLoL choose a cross-spectrum set of illustrative examples (e.g. one cause that resonates more on the left, one more on the right, plus a humane-AI strand) such that no reader feels it is “only for them.” Present options; do not impose. Flag the curation risk: a clumsy “balance” can read as tokenism.
Fact-check before any public mention of specific bodies. LLoL recalls a Vatican document on AI (“Humane/Dignitas…”?) at whose launch “someone from Anthropic spoke.” Do not assert any of this from memory. Use web search to establish the real document name + status (and whether Anthropic was involved) before it appears on any public page — or leave it out. Rather say “I could not confirm” than publish a SIN.
3. Cautions (do not breach)#
No unkeepable public promises. The give-away mechanism does not exist yet; every public word must stay honest about that (cf. Buy-In page: causes “currently most in need to avert some emergency”).
No single named recipient as *the* target on public pages until there is a real, disclosed process behind it.
Divisiveness is unavoidable in the specific; defensible in the principle. Anchor on the principle; use specifics only as clearly-labelled illustration.
Keep it off the cold GoFundMe funnel. Funnel stays “worthy causes most in need”; richer framing lives on the Buy-In page / a dedicated give-away page / the blog.
4. Deliverables#
A give-away framing recommendation (the sharpened criterion + the process design + the balance options for LLoL to choose from + the fact-check result).
A proposal for where it lives (Buy-In §50%-give-away tweak and/or a new page), to be built only after LLoL chooses.
An execution llog at the next free infra number.