The frontier · questions before answers
All twelve named GP gates have packet-complete research contracts, and all twelve remain evidence-open. The public queue below separately labels five packet-complete, three component-supported, and four deferred routing states. A model run can improve a packet. It cannot manufacture its own world contact.
Twelve typed GP sockets · [C] or [I/C]All twelve contracts are packet-complete; none is evidence-complete. Five packet-complete · three component-supported · four deferred describe the public routing state, not whether a research contract exists.
Evidence-openThe required contact has not happened or does not yet discriminate. Separately, the execution board has three ready-to-freeze, three component-contact, one formal-only, and five deferred items.
No consensus promotionIndependent agreement can prioritize a test, never replace it.
Draft evaluation packets · FIN01-02 remains [C]
Three candidate protocols are typed as drafts. Exact materials, external review, an applicable ethics determination, and freeze remain outstanding. No participant sessions have been run, no external reviewer has been engaged, and no comparison has been preregistered.
Fresh readers must identify the category, the bounded practice, its non-authorization and unproven-efficacy limits, and a real Exit—then transfer the distinctions without parroting our words. This tests communication only.
Conflict-declared outsiders must inspect prior art, comparator fairness, measurement, harm, custody, and public language. Their dissent remains intact. Review is criticism, not endorsement or replication.
Four arms separate a Finity-branded worksheet, content-identical brand-masked prompts, a strongest component-matched Ordinary Decision Record, and a person's usual method. Correction, omissions, burden, harm, and attrition remain separate outcomes.
None of these gates validates an ontology. Comprehension can clear the copy; review can improve the search; a controlled result can move only the preregistered low-stakes comparison. The ordinary method is allowed to win.
“Allowed to win” is cheap unless the losing conditions are written down first, so here they are, quoted from the frozen protocol rather than paraphrased:
“Retire or neutralize the brand if F adds pressure/confusion or its apparent advantage over U is only demand, prestige, novelty, or expectancy.”
“If U and F do not beat C by the preregistered meaningful margin after burden, harms, and attrition, preserve the Ordinary Decision Record or other better rival and keep Finity only as an optional selected worksheet if justified.”
“If F/U/C perform alike and beat M, narrow to the supplied structured worksheets in the sampled context; do not call structure itself causal from this contrast alone.”
“Treat non-significance as inconclusive unless equivalence/noninferiority margins were powered and preregistered.”
Those four sentences commit us, in advance, to retiring our own branding, to preferring an ordinary paper form over our worksheet if the ordinary form does better, and to refusing to read a null result as a quiet win. The comparison also cannot begin at all until an ethics determination, a final consent script, a hashed immutable stimulus bundle, and a frozen analysis plan all exist — a no-start boundary, written before any reader is contacted.
The honest status, which is the whole point of saying this now: none of the three has been run. No participant session, no engaged external reviewer, no registered comparison. Publishing the losing conditions before the study is the only part of a preregistration that is worth anything, and it is the only part we have so far done. Full protocols:
03_METHODOLOGY/03_PREREGISTRATIONS/finity_practice/ — 01 fresh-reader comprehension, 02 independent review, 03 controlled comparison.
Twelve open sockets
Showing all twelve questions; public routing state, exact execution requirement, and programme state remain distinct.
Public Grand Puzzle boundary. This is the public twelve-GP research-queue projection of the Grand Puzzle Assembly Ledger; it does not expose or replace the ledger's assembled spine, source custody, or semantic owners. The labels here are public routing state, evidence, and execution labels; they do not state that the ledger's question, rival, discriminator, kill, and survivor packet fields are absent, and they do not change any source owner's packet fields.
Loading the research queue…
This source-level contract is not yet packet-complete and is not counted among the twelve-question queue. Test whether shared semantic and normative language predicts durable coordination after controlling for genetic relatedness, material incentives, contact frequency, institutions, and coercion. Treat trade as a thin cross-language protocol: bounded exchange can work without a shared worldview.
At the individual level, test actual-power stakes such as survival, status, resources, and reproduction. At the group level, test whether shared representations, rituals, norms, and teleology add explanatory power beyond security dilemmas, elite strategy, institutions, geography, and material capacity.
This programme does not establish language determinism, reproduction as a universal cause of war, or all wars as cult or egregore wars. Each proposed layer needs its own measure, rivals, discriminator, and kill criterion.
Laboratory constitution
Every proposal begins without Emergentist vocabulary. The native science keeps its warrant.
Agreement and disagreement remain visible. Synthesis never erases the minority report.
Every packet includes a serious rival and the possibility that the socket should remain empty.
Measures, exclusions and failure conditions cannot be rewritten after the result arrives.
Simulation checks formal consequences. Only the appropriate external evidence moves a world-facing claim.
Its attraction, fracture, survivor and replacement stay attached to the place it once occupied.