Corrections & Error Reporting
Most TruckCodex data mirrors upstream public registries. Real corrections must flow through the original publisher; we mirror the corrected result.
Lag vs. error — triage first
Not every mismatch is a wrong fact. Sort the report before you spend time upstream:
- Import lag — our copy may trail the publisher by hours or days. Check /meta/freshness/ for table as-of dates and cadence before assuming an error.
- Display or join bugs — mislabeled fields, broken layout, stale cross-joins, wrong links on our page. Email us (below); we fix those in the product.
- Upstream facts — identity, authority, insurance, inspections, crashes, recalls that disagree with the publisher. Correct at the source (next section); we re-import after the publisher updates.
How to report an error
The TruckCodex display reflects the upstream record. If a USDOT, VIN, recall, inspection, or insurance row is wrong, the canonical fix is at the publisher:
- Carrier registry, authority, insurance, BOC-3, OOS, MCS-150 → FMCSA Motus (current registration system): motus.dot.gov. Update company facts and biennial MCS-150 filings there. Overview: fmcsa.dot.gov/registration/move-motus.
- Inspections, crashes, complaints → FMCSA DataQs: dataqs.fmcsa.dot.gov.
- Recalls, TSBs, investigations, vehicle complaints → NHTSA: nhtsa.gov/contact.
- EIA diesel prices → eia.gov/about/contact.
- IFTA tax rates → iftach.org.
The older Licensing & Insurance public search (li-public.fmcsa.dot.gov) may still be useful for lookup; it is a legacy surface after the Motus cutover, not the primary path to correct your record.
While an upstream fix is in flight, also email us at [email protected] with the page URL and the field in question. We will flag the record so future imports do not silently overwrite an in-progress correction, and so we can prioritize re-pulling the source once the publisher confirms the fix.
What we will and will not fix
We will fix:
- Display bugs (mislabeled fields, wrong formatting, broken layout).
- Broken links to upstream sources.
- Stale cross-joins (e.g., a chameleon cluster computed before a USDOT change propagated).
- Mistakes in our editorial copy (knowledge-base articles, explainers, FAQs).
We will not modify:
- Upstream data values. If FMCSA says a carrier had a violation, we do not unilaterally remove or rewrite it. The correction must flow through DataQs (or Motus for registration facts) first; we then re-import.
- The historical timeline. Authority transitions, prior addresses, prior insurers, prior officers retained even after the upstream record changes — this is the never-delete doctrine and it is core to the product.
- Missing post-cutover MC certificate PDFs. Letters on TruckCodex are an archive through about 2026-05-20; after Motus, FMCSA stopped publishing the old daily
LI_*bulk PDFs. Gaps after that date are an upstream/source gap, not something we can invent by editing.
Heuristic disclaimers
Two TruckCodex features compute derived signals from public records: the chameleon detector (USDOT inactive → reactive at same address/officer) and insurance-watch (carriers near insurance lapse). Both are heuristics, not legal determinations. They surface patterns that public data already supports — nothing more.
If you believe a heuristic-derived label about your business is incorrect:
- The canonical correction path is still upstream. If Motus (or the mirrored FMCSA open-data files) updates the underlying record (address, officer, authority status, insurance filing), the heuristic recomputes on the next import and the label changes accordingly.
- Email [email protected] with the page URL. If the upstream record is correct and our heuristic is misfiring, that is a bug we will fix in the algorithm.
Data removal
As a rule we do not remove entities from the surface. The never-delete doctrine exists because the historical record is the value — an authority that lapsed in 2019 is still relevant when the same officer reappears under a new USDOT in 2026.
For legal-hold or court-order scenarios we may suppress a page from search-engine indexing (<meta name="robots" content="noindex">) while leaving the underlying record in place. Outright deletion happens only when the upstream publisher itself withdraws the record and we re-import the absence.
Last reviewed: . See also /meta/freshness/, /meta/methodology/, and /meta/sources/.