Author-attached data only

A canvas shows data its author has seen. Reader-supplied data arrives by forking, with authorship attached.

#The rule

A canvas shows data its author has seen. Reader-uploaded data never enters a published internote — this is decided, not deferred.

Data reaches a document through @data, inline or attached as a CSV, and both forms are written by the author into the source.

#Why reader uploads are ruled out

An internote whose canvas shows data the author never saw has no author. The prose cannot make a claim about it, so it collapses into generic captioning — your data may show…, notice any outliers… — which is the narrative contract failing in its purest form. The feature would produce documents that look interactive and say nothing.

It is the same reasoning as the range rule: an author can narrate a range they chose. They cannot narrate data they have never seen.

#The fork is the path

Reader data is not impossible in the medium — it arrives with authorship attached. Fork the document, swap the @data declaration for your own, and rewrite the narrative you now owe. The result has an author again: you.

#What this asks of you

Look at the data before you write about it. The claims in the prose are claims about these rows.
Attach the real thing rather than a rounded excerpt, so a reader checking the numbers finds them.
Transform upstream. @data has no computed columns, filters or aggregates by design — what the document reads is what the file says.