Bottom line
The public missing-persons system is not a system. It is a set of separate databases built by different organisations for different reasons, and nothing reconciles them. That is why the same person can sit unnoticed in two indexes under two spellings, and why a searching family re-enters the same details on six websites.
The layers
- NamUs (namus.nij.ojp.gov) — the national spine. Free, public search plus a restricted investigator tier. Run under the Department of Justice.
- Doe Network (doenetwork.org) — a volunteer cold-case index, mainly unidentified remains and long-term missing.
- Charley Project (charleyproject.org) — a volunteer index of long-term missing, with careful narrative profiles.
- NCMEC (missingkids.org) — the National Center for Missing & Exploited Children; minors.
- State clearinghouses — fifty separate offices, each with its own page and format.
- County ME / coroner offices — where an unidentified person is actually processed. Local, often on paper or a PDF.
What is missing between them
- No open API. There is no official NamUs API. None of the public layers expose a documented endpoint that a tool can query. Any "cross-reference" has to be done by a human reading each site in turn.
- No shared person key. A name is not an identifier. Two spellings, a maiden name, a nickname, a wrong birth year — and one person looks like two, or two look like one.
- No de-duplication. The same case can live in NamUs, Doe Network, and a state clearinghouse with slightly different details and no field that says "this is the same person."
- The authoritative index is closed. NCIC, the FBI's national crime database, is law-enforcement-only. The public cannot see it, and it is not merged with the public layers.
Why it matters in practice
A family searching by hand does the same work six times: type the name and details into each index's own search box, in case the person is in that one and not the others. An investigator does the same, then tries to hold six result sets in their head and spot the overlap that means something. The overlap — one person under two spellings, showing up in two indexes — is the actual finding, and nothing produces it automatically.
What we built from this
Since no index accepts a pre-filled query URL, an honest tool cannot "search them all." What it can do is hold the subject once, hand you the correct query links, keep a ledger of each hit, and flag when the same person appears in two indexes. That is the Missing Persons Console. It does not scrape or mirror any database, and it never touches NCIC. The honesty about what it did not search is a design constraint, not a disclaimer: a grieving family must never be handed false hope by a tool that overstates what it looked at.
Limits of this read
- Verified by checking each public index directly; no API documentation exists to cite because none is published.
- The layering and gaps are stable; individual site URLs and terms change.