# Strict Active Equivalence Sweep

Date: 2026-05-09
Actor: Codex
Scope: TIP active equivalence matches, DB-only evidence, Erik-safe

## Why

After the operator caught the false comparable product for FS.com `QDD-2FR4-800G`, Codex ran a broader DB-only safety sweep over active `approved` and `auto_approved` equivalence matches. The goal was to catch any remaining active 1:1 mistakes before they show up as trusted competitor products in the dashboard.

## Checks

Hard active-match risks checked:

- breakout/AOC/DAC/cable class mismatch
- reach mismatch when both sides have known reach
- fiber mismatch when both sides have known fiber
- primary wavelength mismatch when both sides have known wavelength
- missing core evidence on active matches

## Result

Rejected `16` active false positives:

- Flexoptix 400G/100G pluggable optics matched to ATGBICS AOC/breakout products
- Flexoptix `Q.851HG.03` 300m MMF matched to NADDOD 70m and 40km rows
- Flexoptix `Q.854HG.01.P` 100m MMF matched to a NADDOD 1m row

Backfilled numeric reach evidence:

- `269` rows with `km` reach labels received `reach_meters`
- `131` rows with `m` reach labels received `reach_meters`
- remaining reach labels without meters are only `N/A` accessory/control rows

Post-sweep active risk counts:

- active approved/auto-approved matches: `34051`
- breakout-class mismatches: `0`
- reach mismatches: `0`
- fiber mismatches: `0`
- wavelength mismatches: `0`
- missing core evidence: `0`

Live counters after the sweep:

- equivalences: `pending=0`, `approved=1987`, `auto_approved=32064`, `rejected=148382`, `due_research=0`
- product verification: total `17647`, price `11557`, image `11963`, details `11085`, fully `9861`

## Training Note

TIPLLM should treat unknown reach/fiber/wavelength as insufficient evidence for final 1:1 competitor approval, not as a wildcard. Breakout/AOC/DAC/cable class must be part of the equivalence key.

No external AI was used and no crawler wave was started.