REALITY METABOLISM EXPERIMENT · CALIBRATION IMPACT
How much can two labels change the institutional story?
A cloned experimental branch of the 2,231-record corpus was recalibrated. The canonical corpus was not overwritten.
240
records whose representation changed580
typed route changes39
aggregate journey signatures affectedExperimental boundary: 14 early-WGS labels were narrowed to molecular sequencing/sequence analysis; 230 pre-2019 combined COVID/respiratory labels were renamed to historical respiratory-virus/infection representation. These are simulation rules, not canonical corrections.
Network effect
Route population changed from 6,427 to 6,427. There were 290 old typed routes removed and 290 new routes created because labels participate directly in relationship generation.
Largest journey-count changes
| Type | A | B | Before | After | Δ |
|---|---|---|---|---|---|
| THEME_TECH | Respiratory viruses / respiratory infection | Not specified / mixed | 0 | 125 | +125 |
| THEME_TECH | COVID-19 / respiratory viruses | Not specified / mixed | 199 | 74 | -125 |
| THEME_TECH | Respiratory viruses / respiratory infection | Culture / phenotypic methods | 0 | 55 | +55 |
| THEME_TECH | COVID-19 / respiratory viruses | Culture / phenotypic methods | 129 | 74 | -55 |
| THEME_TECH | Respiratory viruses / respiratory infection | Serology / immunology | 0 | 28 | +28 |
| THEME_TECH | COVID-19 / respiratory viruses | Serology / immunology | 50 | 22 | -28 |
| THEME_TECH | Respiratory viruses / respiratory infection | PCR / molecular diagnostics | 0 | 20 | +20 |
| THEME_TECH | COVID-19 / respiratory viruses | PCR / molecular diagnostics | 26 | 6 | -20 |
| THEME_TECH | Genomics / bioinformatics | Molecular sequencing / sequence analysis | 0 | 14 | +14 |
| THEME_TECH | Respiratory viruses / respiratory infection | Clinical trial / therapy | 0 | 14 | +14 |
| THEME_TECH | COVID-19 / respiratory viruses | Clinical trial / therapy | 40 | 26 | -14 |
| THEME_TECH | Genomics / bioinformatics | Whole-genome sequencing | 184 | 170 | -14 |
| PATHOGEN_TECH | Not specified / mixed | Whole-genome sequencing | 113 | 102 | -11 |
| PATHOGEN_TECH | Not specified / mixed | Molecular sequencing / sequence analysis | 0 | 11 | +11 |
| THEME_TECH | COVID-19 / respiratory viruses | Whole-genome sequencing | 25 | 17 | -8 |
| THEME_TECH | Diagnostics / laboratory methods | Molecular sequencing / sequence analysis | 0 | 4 | +4 |
| THEME_TECH | Diagnostics / laboratory methods | Whole-genome sequencing | 43 | 39 | -4 |
| THEME_TECH | Respiratory viruses / respiratory infection | Whole-genome sequencing | 0 | 4 | +4 |
| THEME_TECH | Respiratory viruses / respiratory infection | Molecular sequencing / sequence analysis | 0 | 4 | +4 |
| THEME_TECH | Public health / epidemiology | Molecular sequencing / sequence analysis | 0 | 2 | +2 |
The lesson
A small correction at the representation layer propagates upward: Record → Classification → Relationship → Journey → Capability interpretation → Institutional story. This is why calibration belongs inside Scientific Intelligence, not outside it as clerical cleanup.
Annual before/after data → · Calibration Lab → · Colleague Beta →