rests on an assumption that is rarely defended because it is rarely noticed.
The assumption is that a community observed over a long period behaves roughly as it would unobserved.
This is more than it sounds and less than the method requires.
People do adapt to a long-term observer, and the is not simply performance.
They come to treat the observer as a member of a category their society already contains.
Which category is chosen is itself a finding, and one that fieldworkers now report rather than .
An anthropologist classified as a kind of visiting official learns something about how officials are treated.
An earlier convention removed the observer from the text entirely and wrote in an voice.
The result read as a description of a society rather than of a series of , which is what it was.
Removing the observer did not remove the observation; it removed the reader's ability to weigh it.
A sentence saying what happens at a funeral conceals whether the writer attended one or was told about them.
Those are very different claims and the voice renders them identical.
Current practice restores the observer and has produced a difficulty of its own.
A text centred on the writer's experience can become a book about the writer.
The community then appears as , which is the failure the earlier convention was designed to avoid.
Neither convention is a solution and the choice between them is a choice about which error to risk.
The working compromise is to state the observer's position once, precisely, and then to write about other people.
It satisfies no theoretical position and is what almost every good now does.