Ghost Town Directory

How we research

Every page on Ghost Town Directory follows the same process, from the first source to the person who approves it for publication.

Deciding what belongs

Before anything is written, we check that a place meets our definition: a formerly named settlement with fewer than 10 residents that is no longer incorporated. If research shows a living community, with more people, a town government, or an active post office or school, we set it aside and write nothing.

Where the facts come from

Research draws on public records, historical societies, library and university archives, newspapers and published local histories. We record what each source says, and the page is written only from those notes. Nothing is added from memory or assumption.

Sources don't always agree on a founding date or the reason a mine closed. When they differ, the story follows the best-supported account, and a short note at the top of the page's sources says where the records disagree. If something can't be confirmed, we leave it out.

Sources on every page

Every history ends with its sources, linked where they are online, so you can read the records for yourself. A source we relied on is a source we list.

How AI is used

AI lets us research at a depth no small team could manage by hand. For each town, our research tools work through county histories, government records, newspaper archives, historical society publications, academic papers and visitor reports: often dozens of sources for a single place. They cross-check dates, names and populations between sources, flag where accounts disagree, and turn up details a quick search would miss. Where sources conflict, we follow the best-supported account and note the disagreement on the page.

AI drafts each article from those source notes, but it doesn’t decide what’s true. Before anything is published, one of our editors reads the draft against its sources, corrects errors, edits the writing and chooses the photographs and videos. If something still slips through, tell us at and we’ll fix it.

Photographs and video

We use licensed archival photographs and Creative Commons images, most of them from Wikimedia Commons. Every photo is credited where it appears and links to its source. Some pages include videos by other creators, played from YouTube and credited to their channels.

Keeping pages right

Records turn up and links break. When we learn a page is wrong, we fix it and note substantive changes on the page. See our corrections policy.