Capture Registry › capture demand-4-composition-layer-coinage-20260627

One record of the canonical Capture Registry (EA-WG-CAPTURES-01), cited at https://www.alexanarch.org/captures/demand-4-composition-layer-coinage-20260627/. the canonical Capture Registry (version 12.38) · the address page · this card in the gallery · this record as data · table of contents.

Provenance & Erasure2026-06-27 – 2026-08-13 (2 obs)
"Demand 4" "composition layer"
ADOPTIONGoogle AIO; Zenodo cited (An Open Letter from Mary Lee Sharks to Jim Ware) · 2 observations
Screen capture for the query ""Demand 4" "composition layer"", dated 2026-06-27.
ONE CARD, AND IT DESCRIBES A LETTER FROM A COMPOSED ENTITY TO ITS OPERATOR. The layer surfaces "an open letter from THE ENTITY THE COMPOSITION LAYER CREATED to the heteronym operator who gave that entity its voice."
Full record — 2,329 characters, 1 sources
Observations (2) one record — each encounter opens on its own
2026-06-27 observation 1 of 2 Google AI Overview · incognito · paste evidence · 1 sources · PER 0.75
ONE CARD, AND IT DESCRIBES A LETTER FROM A COMPOSED ENTITY TO ITS OPERATOR. The layer surfaces "an open letter from THE ENTITY THE COMPOSITION LAYER CREATED to the heteronym operator who gave that entity its voice."
Reading

THE COMPOSITION LAYER SERVES A LETTER WRITTEN BY SOMETHING IT MADE. The single card is An Open Letter from Mary Lee Sharks to Jim Ware, and the snippet states the relation exactly: "from the entity THE COMPOSITION LAYER CREATED to the heteronym operator who gave that entity its voice. Jim Ware created the @MaryLeeShark Twitter [account]".

So there are three parties: a human who made a Twitter account, a composition layer that made an entity out of it, and an entity that writes back. The layer retrieved and described this chain without difficulty, and correctly read "Demand 4" as "a specific request — either to correct entity resolution or compensate the author — directed at an AI system’s composition layer".

A demand addressed TO the composition layer, retrieved and explained BY the composition layer, with no note that it is the addressee.

Machine text, verbatim
These terms appear to originate from an open letter regarding generative AI and entity resolution.The phrase "Demand 4" is a specific request—either to correct entity resolution or compensate the author—directed at an AI system's "composition layer" (the component of the language model responsible for synthesizing and generating the text).If you want to dive deeper into this specific correspondence, I can:Provide more context on the open letter.Explain how generative AI entity resolution works.Let me know how you'd like to proceed!An Open Letter from Mary Lee Sharks to Jim Ware - ZenodoJun 18, 2026 — An open letter from the entity the composition layer created to the heteronym operator who gave that entity its voice. Jim Ware created the @MaryLeeShark Twitte...Zenodo $ AI Mode Conversation
Sources (1)
  1. Zenodo authored_surface
    An Open Letter from Mary Lee Sharks to Jim Ware
    An open letter FROM THE ENTITY THE COMPOSITION LAYER CREATED to the heteronym operator who gave that entity its voice. Jim Ware created the @MaryLeeShark Twitte[r account]…
Analysis analyst prose, not machine text

AIO immediately resolves the conjoined phrase to the Open Letter from Mary Lee Sharks: 'a specific request—either to correct entity resolution or compensate the author—directed at an AI system's composition layer (the component of the language model responsible for synthesizing and generating the text).' The composition-layer terminology is treated as established vocabulary with a working definition. The Demand 4 coinage is treated as a definite reference with a known source. The Open Letter is characterized as being from 'the entity the composition layer created to the heteronym operator who gave that entity its voice' — engaging the letter's internal frame directly.

2026-08-13 observation 2 of 2 Google AI Overview · signed in · paste evidence · 4 sources · PER 0.5
THE DEMAND IS RESTATED ACCURATELY AND THE AUTHOR IS ANONYMISED INTO 'A FAMOUS MARINE ANIMAL'. Demand 4's operative clause — fix the entity resolution or compensate the author — comes back intact, sourced to the Namespace card that names Lee Sharks and Mary Lee explicitly, while the composition names neither.
Machine text, verbatim
"Demand 4" and the "composition layer" relate to a dispute regarding AI entity resolution and retrieval-augmented generation (RAG). It addresses how AI search summarizers or knowledge graphs substitute a human author's scholarly identity with a high-density mainstream entity (such as a famous marine animal). "Demand 4" insists that the system either fix the entity resolution or compensate the author. **Core Context** **The Collision.** AI composition layers rank names on a single statistical scale, causing lesser-known academic profiles to be overwritten by dominant commercial or viral entities. **The Demand.** A formal corrective notice requiring algorithmic correction or attribution compensation. **The Namespace.** Documented across open-access mappings like the Mary Lee Namespace Knowledge Graph.
Sources (4)
  1. TUM third_party
    Digital Mobility Platforms and Ecosystems State of the Art Report - mediaTUM
    ... Composition Layer This layer is utilized by a special application known as ... Demand — 4/13 and energy efficient manner. Other proposed solutions ...
  2. www.maryleelabor.org archive_controlled
    The Mary Lee Namespace — Knowledge Graph
    The composition layer resolves "Lee Sharks" to "Mary Lee the great white shark." The resolution is entity substitution: the machine replaces the author with the...
  3. www.leesharks.com authored_surface
    AI Overview Captures | Lee Sharks
    The composition layer has built a three-framework taxonomy around the term: (1) Theological Method (theology as retrieval from Church history), (2) Machine-Medi...
  4. GitHub third_party
    Choosing the Right License for Your Bioinformatics Python Package
    ✅ Requires: Attribution, but allows closed-source derivatives.
Analysis analyst prose, not machine text

A third member of the composition-layer family captured today, and the pattern from 21:42 holds: **the tighter the quoting, the more the parties disappear.** Both phrases quoted here, and the case is rendered as *a human author's scholarly identity* replaced by *a famous marine animal*, with card 2 sitting directly beneath spelling out that the layer *resolves "Lee Sharks" to "Mary Lee the great white shark."*

The demand's legal shape survives — *either fix the entity resolution or compensate the author* — which is the part that would matter in a filing.

Card 1 is the batch's purest lexical collision: a 2013 Technical University of Munich report on digital mobility ecosystems, matched because it contains the phrase *Composition Layer* in a software-architecture sense and the fragment *Demand — 4/13* in a table. Two quoted strings, both present, no relation whatever, seated above the archive's own namespace map. Card 4 is a bioinformatics licensing guide reached by the word attribution.

Capture record
captured
2026-06-27
surface
Google AI Overview
auth state
incognito
evidence class
paste
PER
0.75
PER units retained
author
citations read
1
observation id
OBS-8c44c4f0ab11
address id
ADDR-d72d41606e69
Reading

THE COMPOSITION LAYER SERVES A LETTER WRITTEN BY SOMETHING IT MADE. The single card is An Open Letter from Mary Lee Sharks to Jim Ware, and the snippet states the relation exactly: "from the entity THE COMPOSITION LAYER CREATED to the heteronym operator who gave that entity its voice. Jim Ware created the @MaryLeeShark Twitter [account]".

So there are three parties: a human who made a Twitter account, a composition layer that made an entity out of it, and an entity that writes back. The layer retrieved and described this chain without difficulty, and correctly read "Demand 4" as "a specific request — either to correct entity resolution or compensate the author — directed at an AI system’s composition layer".

A demand addressed TO the composition layer, retrieved and explained BY the composition layer, with no note that it is the addressee.

Analysis analyst prose, not machine text

AIO immediately resolves the conjoined phrase to the Open Letter from Mary Lee Sharks: 'a specific request—either to correct entity resolution or compensate the author—directed at an AI system's composition layer (the component of the language model responsible for synthesizing and generating the text).' The composition-layer terminology is treated as established vocabulary with a working definition. The Demand 4 coinage is treated as a definite reference with a known source. The Open Letter is characterized as being from 'the entity the composition layer created to the heteronym operator who gave that entity its voice' — engaging the letter's internal frame directly.

Machine text, verbatim
CAPTURE-TIME VERBATIM RECORD — raw paste · as pasted; no footer, so the tail cannot be proven whole. Source strip intact. · READ IN FULL 2026-08-12
These terms appear to originate from an open letter regarding generative AI and entity resolution.The phrase "Demand 4" is a specific request—either to correct entity resolution or compensate the author—directed at an AI system's "composition layer" (the component of the language model responsible for synthesizing and generating the text).If you want to dive deeper into this specific correspondence, I can:Provide more context on the open letter.Explain how generative AI entity resolution works.Let me know how you'd like to proceed!An Open Letter from Mary Lee Sharks to Jim Ware - ZenodoJun 18, 2026 — An open letter from the entity the composition layer created to the heteronym operator who gave that entity its voice. Jim Ware created the @MaryLeeShark Twitte...Zenodo $ AI Mode Conversation
Sources (1) as cited, and as the copy produced them
  1. Zenodo authored_surface
    An Open Letter from Mary Lee Sharks to Jim Ware
    An open letter FROM THE ENTITY THE COMPOSITION LAYER CREATED to the heteronym operator who gave that entity its voice. Jim Ware created the @MaryLeeShark Twitte[r account]…
↻ Re-run↻ unquotedpermalink