Joint likelihood for correlated observations O-35, O-36, O-37, O-38 (shared basis: S-65). Step 8 writes the single ## Likelihood block here; the member edges point back via group.
Likelihood
# CG-1 (HC-1) — ONE joint estimate over members E-49+E-50+E-51+E-52: all four observations (O-35 the DEFUSE
# consortium with WIV as an executing partner; O-36 chimera + hACE2-mouse plan; O-37 furin/human-specific
# cleavage-site insertion; O-38 chimeric-spike work) are extracted from the single DEFUSE proposal document
# S-65 (rule 1 — one witness; if the document is misread, all four are misread together). Judged as ONE
# pattern: in March 2018 the EcoHealth/WIV/UNC consortium, in the outbreak's own city, put in writing exactly
# the manipulations later debated in SARS-CoV-2's genome. This is a document about a PLAN, not evidence the
# plan was executed (A-47), so it LIFTS the engineered member over natural zoonosis but is bounded to "raises,
# cannot confirm" — a means-motive-opportunity update (A-46), not an observation of the virus itself. Anchored
# on H-43 = 1 (the member the whole pattern best fits); the rest priced as ratios to it (rule 7). Not
# double-counted with the prior: HC-1's prior explicitly excluded the DEFUSE feature-match (A-39 -> E-49/E-51)
# as belonging to this step-8 edge.
lik_defuse_H43 = 1.0 # anchor: engineered lab leak at these institutions is the only member for which the
# document is more than coincidence — the same institutional capability/intent that
# would build the construct is what wrote the plan (common cause). A-46 MMO update:
# documented intent (the furin/cleavage-site plan) + means (Baric's reverse-engineering
# track record, WIV sampling) + opportunity (WIV's own city, 20 months prior). Does NOT
# observe the genome — that evidence lives on the genome edges — so it raises, not confirms.
lik_defuse_H42 = 0.6 # 0.6x as expected as under H-43: H-42 (leak of an UNMODIFIED field-collected virus) is
# genuinely lifted by O-35 — DEFUSE documents an active WIV program sampling and handling
# exactly this bat-SARSr-CoV family, which is the collection/handling route H-42 posits.
# But the engineering-specific content (O-36/O-37/O-38) is not predicted by an unmodified
# virus, so H-42 earns the research-proximity lift only, well below the engineered member.
# ~1.4x over H-41. Confidence moderate — the split rests on which observations discriminate.
lik_defuse_H41 = 0.42 # 0.42x: under pure natural zoonosis with no research role, a specific matching plan
# existing in the outbreak city ~2yr prior is a coincidence, so the MMO update (A-46)
# penalises H-41 relative to the research members. BOUNDED, not extreme (A-47 caps the
# force): EcoHealth/WIV were the world's dominant bat-CoV labs, so a matching aggressive-
# work proposal near a natural southern-China / Wuhan-hub emergence is not astronomically
# unlikely; and the document is a REJECTED proposal, chimera work slated for UNC, on
# SARS-1-related WIV1/SHC014 backbones outside SARS-CoV-2's lineage — hence ~2.4x, not 10x.
lik_defuse_H44 = 0.55 # 0.55x: unlisted/hybrid origin, unconstrained — a middling value near H-42 (rule 3).
# Some residual histories involve these institutions (raising the document's
# expectedness), others do not; the residual neither strongly predicts nor forbids the
# pattern. Not set near 0 (would assert no unlisted origin could coincide with such a
# plan, which I don't know) nor near 1.
t_defuse = 0.90 # cap = trust_score of S-65 = 0.90 (the min over {O-35,O-36,O-37,O-38}; all four carry
# data_basis S-65). No dock: authenticity is very high (DRASTIC leak, DARPA-confirmed
# receipt+rejection, never disputed by EcoHealth) and the facts are documentary and
# directly checkable; the degraded-OCR quotes are reconciled to the widely-circulated
# verbatim text, and O-38's 'GoF-exempt' framing is EcoHealth advocacy but the
# observation is only that they ASSERTED it, which is faithful. So t sits at the cap.
evidence("HC-1", ["O-35", "O-36", "O-37", "O-38"], [lik_defuse_H41, lik_defuse_H42, lik_defuse_H43, lik_defuse_H44], t=t_defuse)