Privacy safe family profiles facts: why citations vanish

A profile is not reported. It is assembled from other pages, and that is why its facts behave differently from the facts in an article somebody actually went out and gathered.

Every hop in that assembly drops a citation. By the time a profile exists, most of its statements have no findable origin, and the ones that do point at another profile. This page is about privacy safe family profiles facts and restoring the provenance, because a fact with no origin cannot be checked, corrected, or defended.

What to take away

Where the provenance goes

Follow a single statement backwards and the losses are predictable.

Nothing in that chain is dishonest. Each step is a normal editorial compression, and the cumulative effect is a page of confident statements resting on nothing you can name. On this subject the chain usually starts with a celebrity family: a birth announcement, a baby's name, a child's age. From there the hops run through a handful of familiar places — a news article, an aggregator, a Wikipedia article, and one or more profiles on FamilySearch, Ancestry, Geni or WikiTree. The difference between a document and somebody's report of it is not cosmetic, and the comparison is set out in two documents, two authors.

Take the kind of statement that starts these chains: a baby's name or a birth date attributed to a celebrity family in a news report. The report had a person behind it — someone who was present or close to the family — and it said so. An aggregator condensed the item to the bare fact, with no link back.

A first profile copied the name and cited the aggregator. A second profile copied the name and cited the first profile. Run the trace backwards and the origin is one news item and one person who was there.

The distinction that matters is the old one between a primary source and everything downstream of it. A profile is about as far downstream as a document gets, which is why building one requires more sourcing discipline than writing a piece from scratch, not less.

The fact with no findable origin

This is the profile's characteristic defect, and it does not feel like an error.

A statement is true. It has been in circulation for years. Nobody knows where it came from, and nobody has been able to check it since it entered. It cannot be corrected, because nobody knows what to correct it against, and it cannot be defended if challenged, because the defense is that it has been repeated.

Treat such a statement as unsourced no matter how long it has stood, because age is not corroboration and durability is a property of publishing systems, not evidence.

Either you can name where it came from or you cannot, and if you cannot, the honest options are to establish it or drop it. The same reasoning applies to every claim you did not gather yourself: write it down with its source attached, in a form that still holds up when someone else later takes the sentence apart.

Circular sourcing, and how to spot it

The specific pathology is a loop: A cites B, B cites C, and C cites A, so a claim with no origin appears to have three independent sources.

Written out, the loop looks like this. A Wikipedia article states a birth year and cites a newspaper piece. The newspaper piece states the same year and cites the Wikipedia article. A genealogy profile on FamilySearch or Ancestry states it and cites both. Ask any of the three what it saw first-hand, and none can answer.

The pattern has a name borrowed from Wikipedia: citogenesis — an unsourced line enters an article, a publication repeats it, and the publication is then cited back into the article as a source. In celebrity family coverage the shape is the same: a baby's name or a birth date appears in a news item, a fan page or aggregator repeats it, and genealogy profiles copy it and cite each other instead of the report that carried it first.

Sourcing standards that news organizations publish for themselves, such as the Associated Press news values and principles, are written around exactly this: what a reader can check, rather than what a writer feels confident about.

The three tests, in the order worth running:

The wording test is the most reliable of the three and the least used. When three pages carry the same slightly odd construction, they have one ancestor. Finding it usually takes a single search.

Where the underlying material is spoken rather than written, the loop usually starts as a paraphrase that quietly acquired quotation marks as it was copied from page to page, which is why the wording test catches what the citation test misses.

The record behind an auditable profile

Since a profile's problem is provenance, the fix is a record built for provenance rather than for the page.

For each statement, hold five things. The claim as published, in your wording. The source, named specifically enough to find again, with the date it was published and the date you saw it.

The kind of source matters: a statement by the person, a record, and an article about an article are three different objects.

What the source actually said, in its words, because paraphrase is where drift starts.

And what remains unestablished, the field nobody keeps and the one that stops a later editor from assuming the check was done.

That last field is worth defending in a meeting. A profile with no record of its gaps reads as complete, and a reader has no way to tell a verified statement from an inherited one.

The general practice of capturing and storing what you worked from is set out in more detail in keeping what you worked from.

Genealogy already has vocabulary for this. The Genealogical Proof Standard, maintained by the Board for Certification of Genealogists, expects a reasonably exhaustive search, complete and accurate source citations, and a soundly reasoned written conclusion. Elizabeth Shown Mills's Evidence Explained is the standard reference for citing the records themselves — the repository, the file, the page.

Provenance is necessary and not sufficient

Getting the sourcing right settles whether a statement is true. It does not settle whether it belongs on the page.

A perfectly sourced location, routine, or detail about a child is still a location, a routine, and a detail about a child, and the sourcing does nothing to reduce what publishing it does. Those are separate gates and passing the first one is not permission to skip the second.

The standard for what may be carried at all is in the guide to what a profile may carry. The strictest version, which applies to anyone who did not choose a public role, is in the guide to claims about children.

The second gate has its own standards. The Society of Professional Journalists Code of Ethics asks whether a detail is necessary and whether publishing it serves the public, and treats minimizing harm as a duty; the Associated Press news values and principles point the same way. US law largely governs collection rather than publication — the Children's Online Privacy Protection Act sets rules for what online services may collect from children under 13 — so the restraint on what a profile carries has to come from an editorial standard rather than a statute.

Common questions

A fact appears on several reputable sites. Is that not enough?

Only if they are independent, and on this subject they usually are not. Establish how many sources there are rather than how many pages, and if you cannot tell, you have learned that the claim is weaker than it looks.

We cannot find the origin but the family has never disputed it. Does silence help?

No. Families do not monitor everything written about them, disputing a claim republishes it, and many decline to engage on principle. Silence is not confirmation, and treating it as confirmation punishes people for declining to participate.

Is a reference work an acceptable source for a profile?

As a pointer, yes. As an endpoint, no. Reference works are compilations too, and a good one will tell you what it was compiled from; follow that through rather than stopping at the compilation.

What if a statement was confirmed to us privately?

Then you have a fact you cannot publish, which is a real position rather than a technicality. Record that it was confirmed, record that it was off the record, and do not let a later editor find the note and mistake it for a green light.