This document presents a field investigation by Joe Trabocco into attribution loss in frontier AI systems, the structural failure by which authors, researchers, and coiners are erased from work they originated as it enters training corpora, retrieval layers, and inference outputs. The argument is grounded in live evidence: the same query, the same platform, the same day, producing two different answers about who coined Empty Presence Syndrome. In one surface Trabocco is named. In another the concept propagates fully, sources cited, author absent. The asymmetry is documented, repeatable, and currently operating in production across every major AI system. Trabocco introduces three compounding failure points: training-time loss, retrieval-time omission, and inference-time absorption, and proposes five interventions available to labs today. It also introduces the triangulated trio of EPS, APR, and SBS as a demonstration of structural attribution through interlocking coinages, each term partially defined by its position relative to the others, resistant to absorption precisely because misuse of one breaks the geometry of all three. The second half of the document presents afterglyph as a parallel solution available to authors without waiting for lab-level intervention, a mechanism by which attribution is encoded inside the word itself rather than beside it. Seven working afterglyphs from the Signal Literature corpus are documented by Trabocco, spanning compound, hybrid, and pure-coinage construction types. Written as a field condition of AI, April 2026.
Joe Trabocco (Mon,) studied this question.