# Faultprint Operating/Postmortem Crosswalk 002

## Purpose

This calibrated v2 pass compares the operating condition library against the available postmortem evidence base. The point is not to say that operating institutions have failed. The point is to test whether live operating conditions resemble structural conditions already observed in known breakdown cases.

## Inputs

- Operating pathology mapping: 680 rows
- AI postmortem registry: 300 rows
- Grid master evidence table: 124 cases
- Grid co-occurrence table: 45 pathology pairs

## Output Files

- `FAULTPRINT_OPERATING_POSTMORTEM_CROSSWALK_002.csv`
- `FAULTPRINT_CROSSWALK_PATHOLOGY_SUPPORT_002.csv`
- `FAULTPRINT_CROSSWALK_SYNTHESIS_002.md`

## Resemblance Bands

- moderate_resemblance: 371 rows
- elevated_resemblance: 244 rows
- low_resemblance: 58 rows
- high_resemblance: 7 rows

## Pathology Support

| Pathology | Operating Rows | AI Postmortem Rows | Grid Cases | Support Band |
|---|---:|---:|---:|---|
| Contestability Lag | 96 | 92 | 121 | strong |
| AI Authority Acceleration | 299 | 135 | 0 | moderate |
| Record-Authority Capture | 83 | 16 | 111 | moderate |
| Evidence-Container Illusion | 0 | 3 | 119 | moderate |
| Documentation Burden Transfer | 25 | 24 | 97 | moderate |
| Physical-World Authority | 0 | 4 | 81 | moderate |
| Access Interruption Before Review | 87 | 13 | 64 | moderate |
| Multi-Actor Handoff Diffusion | 11 | 0 | 71 | limited |
| Deadline Compression | 25 | 5 | 41 | limited |
| Classification Gatekeeping | 33 | 7 | 0 | limited |
| Remedy Path Split | 0 | 1 | 0 | limited |
| Provisional Relief Reversal | 21 | 0 | 0 | not_yet_supported |

## Highest Concentrations By Institution

- Chase Account Restriction and Dispute: 5 elevated/high rows
- Amazon Seller Central: 5 elevated/high rows
- ChexSystems: 5 elevated/high rows
- The Work Number Employment Data Dispute: 5 elevated/high rows
- Con Edison Payment Agreement and Termination: 5 elevated/high rows
- Citizens Property Claims: 5 elevated/high rows
- Credit bureaus / furnishers: 5 elevated/high rows
- State DMV/OMV agencies: 5 elevated/high rows
- MTA Transit Adjudication Bureau: 5 elevated/high rows
- Child Protective Services: 5 elevated/high rows
- E-ZPass Toll Violation Dispute: 5 elevated/high rows
- Airbnb: 4 elevated/high rows
- Philadelphia Eviction Diversion: 4 elevated/high rows
- Emergency Rental Assistance Documentation: 4 elevated/high rows
- Stripe: 4 elevated/high rows

## Highest Concentrations By Sector

- Insurance telematics: 21 elevated/high rows
- Health insurance: 8 elevated/high rows
- Employment screening / consumer reporting: 8 elevated/high rows
- Tenant screening: 7 elevated/high rows
- Education finance / federal student aid: 7 elevated/high rows
- Credit / financial AI: 6 elevated/high rows
- CPS intake and investigation: 5 elevated/high rows
- Employment and income verification / consumer reporting: 5 elevated/high rows
- Marketplace seller enforcement: 5 elevated/high rows
- Credit reporting disputes: 5 elevated/high rows
- Banking / deposit accounts: 5 elevated/high rows
- Electric gas and steam utility: 5 elevated/high rows
- Property insurance / insurer of last resort: 5 elevated/high rows
- Driver and vehicle reinstatement: 5 elevated/high rows
- AI lending / credit underwriting: 5 elevated/high rows

## Evidence Quality

- anchored_and_falsifiable: 680 rows

## Initial Read

The strongest bridge between the operating and postmortem libraries is not a company-level match. It is a condition-level match: current documents repeatedly show record authority, burden transfer, contestability lag, AI acceleration, and handoff diffusion, while postmortem records show the same conditions as recurring parts of breakdown anatomy.

The most commercially useful result is the ability to say: this client corpus contains conditions that are common in current operations, but some of those conditions also recur in postmortem cases when paired with missing controls, delayed contestability, compressed timing, or weak evidence of correction.

## Current Limits

This is still a first calibration pass. The operating pathology mapping is deterministic and has been calibrated once to prefer narrower primary pathologies over generic record-authority assignment. The postmortem mapping combines a purpose-built AI postmortem registry with an older grid evidence table whose pathology names are not identical to the Faultprint taxonomy. The output is defensible as an analytical bridge, not yet as a final actuarial or predictive model.

## Next Defensibility Step

The next step is manual calibration of a sample of rows from each resemblance band. For each sampled row, the test is whether a reviewer can trace the operating finding, pathology assignment, postmortem support, and falsification evidence back to the source fields without relying on narrative inference.