Alan Yuille person-bridges the vault's Stephen Hawking and David Marr entity hubs, invisible to embedding retrieval

cross-domain-bridgeperson-bridgebridge-investigationretrieval-blind-spotmeta-vaultalan-yuillestephen-hawkingdavid-marrcomputer-visionhistory-of-science

Stephen Hawking and David Marr are both established vault hub entities with no link between them. Alan Yuille is the connective person. Yuille's own Johns Hopkins homepage states: "His PhD on theoretical physics, supervised by Prof. S.W. Hawking, was approved in 1981" — a Cambridge cosmology thesis, unconnected to vision. In 2003, Yuille co-won computer vision's Marr Prize at ICCV; the conference's own award page names the prize for "David Courtnay Marr, a Cambridge (UK) theoretical neuroscientist... who made profound contributions to the theory of both human and machine vision at the MIT AI Lab," and the same page credits the winning paper, "Image Parsing: Unifying Segmentation, Detection and Recognition," to "Zhuowen Tu, Xiangrong Chen, Alan L. Yuille, Song-Chun Zhu."

One career therefore anchors both hubs: a 1981 physics PhD under Hawking's supervision, and a 2003 vision prize named for Marr, twenty-two years and two fields apart, with no shared text or vocabulary between the two source documents. This is the shape the vault's retrieval tooling cannot see: vault_bridge, queried directly on the Yuille/Hawking/Marr connection, returned bridge_candidate: false, surfacing only the vault's other physics-into-AI notes (Hopfield/Hinton's Nobel, NVIDIA's Blackwell/Rubin naming) and never Hawking or Marr by name — because the tool scores note-text geometry, and no note yet named the hinge. It is a further instance of the "person-bridge invisible to vault_bridge" pattern already logged for Isaac Pitman/Carl von Ossietzky, A.E. Clark, A.A. Fel'dbaum, David Rumelhart, and Paul Allen (see observation-paul-allen-person-bridges-altair-basic-and-semantic-scholar-clusters and its own chain of citations) — a bridge that surfaces only once a hop follows the person rather than the embedding.