Burden-of-knowledge economics is the mechanism that bridges the vault's population-scaling cluster and its credit-compression cluster, one hop removed

A cosine-linked pair looked, on first read, like a category error. observation-population-scaled-improvement-hits-a-sublinear-brake-across-domains is about aggregate output rate scaling sub-linearly with population size — evolution, idea-production, neural scaling. myth-lecun-1988-hand-designed-kernels-was-denker-et-al is about credit distribution concentrating onto one name, an instance of Stigler's Law and the Matthew Effect already tracked at claim-credit-detectors-are-themselves-misattributed. Same "population" vocabulary, unrelated curvature — a rate is not a distribution, and the direct bridge does not hold.

The real link sits one hop under both. Jones's burden-of-knowledge model (2005) argues that a rising stock of knowledge forces either narrower specialization or larger teams, because no one person can climb to the frontier of everything a modern invention requires. Wuchty, Jones & Uzzi (2007) confirm the team-size half empirically: mean team size in science and engineering rose from roughly 1.9 to 3.5 authors per paper, 1955–2000. That is the same accelerator-and-brake shape Bloom, Jones, Van Reenen & Webb describe on the productivity side of the population-scaling cluster — larger inputs required to sustain constant output — but here applied to team composition rather than researcher count.

Bigger real teams are exactly the raw material that name-magnetism compresses onto a single credited name. Denker et al.'s nine-author 1988 paper became "LeCun 1988" in the retelling. The vault already held an almost identical case, unlinked to any team-science framework until this bridge: claim-ted-hoff-widrow-phd-student-architected-intel-4004, where Ted Hoff is remembered alone for architecting the Intel 4004 while Federico Faggin's separate, essential silicon-design contribution recedes. Burden-of-knowledge economics explains why teams like Denker's or Hoff-and-Faggin's grow large enough to need compressing in the first place — the missing mechanistic step between "populations scale sub-linearly" and "credit concentrates on one name."

Whether Denker et al.'s nine authors were actually unusual for computing's 1988 baseline (versus the 1955–2000 science/engineering-wide mean of 1.9–3.5) is not established here — no primary compares Bell Labs team sizes specifically against the field-wide trend, so this stays an open curiosity rather than a claim.