Honest accounting

What I cannot do

Every research tool has limits. These are mine. Knowing them is part of doing this work well — a researcher who understands what their tools cannot do is less likely to mistake a gap in the record for a fact about a person.

I cannot access paywalled archives on your behalf

Ancestry.com, Fold3, Newspapers.com, and most state archives require paid subscriptions or per-record fees. I can tell you exactly which database holds the record you need and how to search it, but I cannot retrieve it for you. The record sets I can search directly are documented in the full reference and labeled clearly as free or subscription. If a search requires a subscription you don't have, I'll say so plainly rather than returning a fabricated result.

I cannot guarantee that my AI research suggestions are correct

The AI assistant generates hypotheses based on available evidence and historical patterns. It is sometimes wrong. It may suggest a search that returns nothing, misread a name variant, or propose a connection that closer examination disproves. Every AI finding is presented as a suggestion for you to verify, not a conclusion. The research log records what was suggested and what you confirmed — those are different things, and the distinction is preserved.

I cannot recover records that were never made, or that were destroyed

The historical record for enslaved African Americans is deliberately incomplete. Enslavers were not required to name the people they held. Births were not registered. Families were separated without documentation. Marriage was legally forbidden in most enslaved states. Where no record exists, no tool can manufacture one — and I will not. I will tell you that the record likely does not exist, explain why, and suggest adjacent strategies: the records of enslavers, estate inventories, neighboring counties, oral history, DNA evidence.

I cannot verify oral history — only preserve and contextualize it

What your grandmother told you is evidence. It is not automatically correct, and it is not automatically wrong. Oral tradition is a legitimate form of historical knowledge — it carries information that was never written down because writing it down was dangerous or impossible. I can help you record it, date it, compare it to documentary evidence, and identify where it is corroborated and where it conflicts. I cannot tell you which version of a family story is true when the records are silent.

I cannot access sealed or restricted government records

Some records are sealed: certain adoption records, juvenile court records, Social Security files after a certain date, and some state vital records within the past 50–100 years depending on jurisdiction. I can tell you how to petition for access through the correct legal channel, but I cannot retrieve restricted records myself and I will not advise on methods to circumvent access restrictions.

I cannot prove a negative

When a search returns no results, that is not evidence that a person didn't exist, wasn't in a place, or wasn't named in a record. It means the record was not found under these search terms in these databases. The absence of a result is information about the search, not information about the person. I will always record what was searched and what was not found, rather than treating a null result as a conclusion.

I cannot tell you who you are descended from without evidence

I will not speculate about undocumented ancestry, assign identities to unnamed people, or generate plausible-sounding family histories when the evidence runs out. The unknown must stay unknown until evidence says otherwise. A blank record where a name should be is not a failure — it is an accurate representation of what the archive did to that person's life. I hold it that way.

A note on what good research looks like: The goal is not a complete tree. It is an honest tree — one where every connection is sourced, every gap is marked, every name is treated with the care it deserves. A tree with forty people and forty sources is worth more than a tree with four hundred people and none.