John Doe: The History of Placeholder Names

An unidentified patient arrives at a US emergency room and is admitted as John Doe. A lawsuit is filed against someone not yet identified: John Doe. A sample driver's license in a government brochure: John Doe again. He is the most famous man who never existed. His story starts in medieval English courtrooms, centuries before anyone needed a fake name generator.

A medieval legal fiction

English property law had a procedure called the action of ejectment for disputes over land. To fit a real dispute into its rigid formula, the court papers told a small fiction: a hypothetical tenant had leased the land and been thrown off it. That tenant needed a name. By convention lawyers wrote John Doe, and a second stock character, Richard Roe, played the other side ^{[wiki-doe] [lii-doe]}. The practice may date back to the reign of Edward III, from 1327 to 1377.

The names were deliberately generic: a doe and a roe, two kinds of deer, attached to everyday first names. England abolished the ejectment fiction in 1852, and British usage moved on to other placeholders. American law had already absorbed the convention and kept it ^[wiki-doe].

John Doe in American law today

In the modern United States the name does real procedural work. Courts accept Doe pleadings when a party's identity is unknown or protected. A plaintiff can sue “John Doe 1–10” before knowing who the defendants are, then substitute real names later. Anonymous parties in sensitive cases proceed as John or Jane Doe, as in Roe v. Wade and Doe v. Bolton ^[lii-doe].

Police use John Doe warrants that identify a suspect by DNA profile or description when no name is known. Coroners and hospitals log unidentified people as John, Jane or Baby Doe ^[wiki-doe].

The system is not a museum piece. It carries a real caseload. In a 2004 census, US medical examiner and coroner offices reported handling about 4,400 unidentified bodies in a typical year, and about 1,000 were still unidentified a year later ^[bjs-mec]. NamUs, the national database for these cases, exists to match them with missing persons ^[namus]. Those files open under a Doe name.

Every country has its own John Doe

The problem John Doe solves is universal, so nearly every naming culture has an equivalent, often printed on official sample documents. Germany's sample ID cards famously carry Erika Mustermann, the female counterpart of Max Mustermann ^{[wiki-placeholder]}.

The pattern is always the same. Take an unremarkable surname, attach a common first name, and the result reads as anyone and no one. Need this for one locale, such as a believable German, Japanese or Spanish everyman for a demo? The country generators build names from each locale's name lists, in the right order. The name directory covers the meaning and origin of hundreds of individual names.

From John Doe to generated test data

John Doe is, in effect, the world's oldest piece of fake test data: a name-shaped value with no real person behind it, standardized so systems and people know what it means. But one placeholder breaks down in modern software. Fill a staging database with ten thousand John Does, and you cannot test search, sorting, deduplication, other languages or realistic layouts. Every record collides.

• Variety: real systems need varied names to exercise indexes, search and deduplication.
• Locale coverage: 'John Doe' tells you nothing about how your UI handles 'Nguyễn Thị Minh Khai' or 'Björn Ægir'.
• Realistic shapes: layout and truncation bugs only show up with real-world name lengths.
• Consistency: a believable test person needs a name that matches the address, phone and email around it.

That is the gap generated identities fill. The fake person generator produces thousands of distinct, consistent John Does. Each has a matching address, a fiction-range phone number and a profile, and none describes a real person, as the medieval lawyers intended. For where the legal lines sit, read is using a fake name generator legal.

A seven-century-old solution, still in use

Placeholder names are a seven-century-old answer to a problem software still has: sometimes you need a person-shaped record without a person. John Doe standardized the answer for courts and coroners. Locale-aware generated names are the same idea, scaled for databases and demos. Both work for one reason: everyone can tell at a glance that no real person is behind the name.