Oral history is almost always gathered too late

The theory of oral history is well covered elsewhere. This is about the version you meet in practice.

The timing is structurally wrong

Oral history projects tend to be triggered by anniversaries, closures, funding rounds or the death of a prominent participant. All of those occur long after the events, which means the interval between experience and recording is decades by design. The ideal interval is short enough that memory is fresh and long enough that people are willing to speak, and nobody plans around it.

The result is that most testimony is collected at close to the worst available moment.

Survivors are not a random sample

The people available to interview are those who lived longest, stayed reachable, and are willing and able to talk. That correlates with health, class, social connection and often with having a story they are used to telling. Interviewers know this and cannot fully correct for it, since the alternative sample does not exist.

For a while, any oral history collection should be read as evidence about its own selection as much as about its subject.

Retelling has already shaped the account

A person interviewed at eighty about something at twenty has told the story many times, and the version recorded is the polished one. Polished stories are more coherent, better structured and further from the original experience than a fresh account would be. Skilled interviewers work around this by asking for sensory and procedural detail rather than for the narrative.

Asking how something was done frequently produces better evidence than asking what happened.

The technical barrier has disappeared entirely

Recording good audio now requires a phone and a quiet room, where it once required equipment, expertise and money. Storage costs nothing and transcription is largely automated, which removes the two historical bottlenecks. Despite this, family recording remains rare, because the barrier was never really technical.

The obstacle is that it feels premature until the moment it becomes impossible.

Method matters more than equipment

Open questions, silence after answers, asking about routine rather than highlights, and returning for a second session all produce substantially better material. Consent, context notes and clear labelling with names, dates and places are what make a recording usable to anyone else later.

By the end, oral history associations in several countries publish free guidance covering ethics, consent and technique. A poorly labelled recording is close to worthless, and this is the most common failure by a considerable margin.

Institutions have the same problem at larger scale

Companies, hospitals, unions and public bodies lose operational knowledge continuously as staff retire, and formal capture is rare. Where it has been done — in nuclear decommissioning, in rail engineering, in certain manufacturing sectors — it was usually driven by a safety or liability requirement. Knowledge capture funded by regulation is the only kind that reliably happens.

By the end, everywhere else, the tape recorder comes out at the leaving party, which is far too late.

The takeaway

Everybody intends to ask. The interval between intending to ask and being unable to is the entire subject of oral history.

What replaced it works. That is not the same as it being nothing.