Noted, and thanks.
Thanks for the feedback! So in total, it seems that researchers at least want the option to keep codings private, while not being opposed to selectively make them public as well.
After much thinking, I think I found a good solution that respects this requirement. The coding interface would not be in the editor, but rather work with Annotator (which see for a live demo, by selecting some text there … it’s quite nice). The storage would work using stand-off annotation technique: giving IDs to all words, and storing codings externally from the text by referencing the coded words’ IDs. (More in the software design wiki.) Still work in progress. For now, I’d like to know from you if word-level coding is sufficient? That is, do ethnographers require the option to code parts of words, or not?