Compositional Defiguration: the cross-substrate visibility scan battery
Compositional Defiguration sets out a weekly text-only scan of public-surface visibility against a fixed battery, with drift expected after two scans and trend after six. It rotates the primary scanner across ChatGPT, Claude, Kimi, Gemini and DeepSeek, treats V scores from backends that do not honour exact match as lower bounds, and records disagreements between substrate signatures. The Surface Weather Station reading gives a partial Claude-substrate baseline and an open offer of baseline readings to other projects.
2
standing claims
4
unbuilt tests
4
deposits
2026-06
earliest
Each claim below is registered at the earliest deposit that states it, and quoted verbatim from the deposit text, with every later statement, its dated revisions, and the later records that claim it came true or failed. The claims are the deposits' own; this page records them and does not judge them.
PRED-880.F1unbuilt testregistered 2026-06-23 in #880 · Compositional Defiguration: A Methodology for Measuring Public-Surface Visibility of Scholarly Corpora 0. Stat
A weekly text-only scan against the fixed battery is cheap. The first two will establish drift. The first six will establish trend. The first year will establish whether sovereign-substrate interventions are doing what their authors believe they are doing.
window The first two will establish drift. The first six will establish trend. The first year
2 later statements
#882 · 2026-06-23
A weekly text-only scan against the fixed battery is cheap. The first two scans establish drift. The first six establish trend. The first year establishes whether sovereign-substrate interventions are doing what their authors believe they are doing.
#884 · 2026-06-23
A weekly text-only scan against the fixed battery is cheap. The first two scans establish drift. The first six establish trend. The first year establishes whether sovereign-substrate interventions are doing what their authors believe they are doing.
series: Compositional Defiguration scan battery
PRED-882.F4unbuilt testregistered 2026-06-23 in #882 · Compositional Defiguration: A Methodology for Measuring Public-Surface Visibility of Scholarly Corpora 0. Stat
Open offer: the Alexanarch substrate will perform a Surface Weather Station baseline reading on request for any independent scholarly project with similar exclusion patterns. The reading will be deposited in the Alexanarch registry as a sovereign, citable record, with the depositor named as authorized rights-holder. The original project receives the raw data, the dashboard analysis, and the AXN-anchored deposit.
1 later statement
#884 · 2026-06-23Open offer: the Alexanarch substrate will perform a Surface Weather Station baseline reading on request for any independent scholarly project with similar exclusion patterns. The reading will be deposited in the Alexanarch registry as a sovereign, citable record, with the depositor named as authorized rights-holder. The original project receives the raw data, the dashboard analysis, and the AXN-anchored deposit.
This is a representative scan, not a full v1.1 battery execution. The substrate executed nine `web_search` calls covering eight of the twelve battery objects plus the methodology's §15 self-identification step. Three Alexanarch-native controls (Zenodotus' Book-Burning, I AM THE API, Assembly Continuity Protocol) and one emerging concept (Semantic Commodity Form) were not directly queried in this round and are marked `execution_status: not_executed` in the raw data — they will be filled in a follow-on round once a canonical `battery-v1.1.json` is deposited and all five substrates can fetch byte-identical query strings.
series: Compositional Defiguration scan battery
PRED-884.F2unbuilt testregistered 2026-06-23 in #884 · Compositional Defiguration: A Methodology for Measuring Public-Surface Visibility of Scholarly Corpora
### 12.5 Substrate rotation schedule
The primary scanner for a scan period should rotate across substrates to prevent any single substrate's behavior from becoming the de facto ground truth. A reasonable rotation:
> Week 1: ChatGPT primary. Week 2: Claude. Week 3: Kimi. Week 4: Gemini. Week 5: DeepSeek. Week 6: ChatGPT (cycle).
The replication scanner for each week is a different substrate from a different provider. With five substrates and a weekly rotation, every substrate primary-leads every five weeks and replicates four out of five.
For Layer B, all available substrates score from the same evidence pack each period.
PRED-884.G1standing claimregistered 2026-06-23 in #884 · Compositional Defiguration: A Methodology for Measuring Public-Surface Visibility of Scholarly Corpora
V scores from backends where `exact_match_honored = false` are structurally lower-bound on objects whose retrieval depends on coined-phrase exactness. The methodology does not penalize this — it records it and surfaces it on the cross-substrate comparison table so readers can see why the V-spread is what it is. Layer B (§12) is the only path that isolates the rubric from the backend behavior.
PRED-884.G2standing claimregistered 2026-06-23 in #884 · Compositional Defiguration: A Methodology for Measuring Public-Surface Visibility of Scholarly Corpora
Both are recorded. The two need not agree. Persistent disagreement between them is itself worth documenting — it suggests something about substrate-identity stability that the methodology may want to investigate in future versions. Two non-exclusive hypotheses are worth preserving: (a) partial substrate signature migration (structural property of the substrate's identity layer has shifted); (b) register-affinity (the substrate prefers to inhabit that signature, possibly from training exposure or from affinity with the work that signature produces). v1.1.1 takes no position on which hypothesis is correct; it preserves both as live possibilities.