[tannevaled]

Why

The workbench page covers selecting and marking. Redact is the fourth
editing scope — and the only verb on that tab that destroys anything, so it is
the one most worth explaining.

What the section says

• what it does: removes the operator that drew the text, and says so —
  "removed 17 characters from 1 page — the text is gone from the file, not
  covered";
• why that is the whole point, in the spec's own words. ISO 32000-1 §12.5.6.23
  requires a conforming reader applying a redaction to "remove all traces of
  the specified content" and says "clipping or image masks shall not be used
  to hide that data". A rectangle drawn over a paragraph leaves the text in the
  stream for anyone to copy out from under it;
• where the verb sits and why — next to the selection, because a redaction you
  had to go and find on another tab is one you would do with a rectangle
  instead;
• the four refusals, and why each is a refusal rather than a best effort: a
  form XObject (shared bytes), an image (the pixels must be destroyed, and a
  mask over them is what the spec forbids), an annotation (it holds text of its
  own), and an operator that also draws words outside the selection. A redaction
  betrays its reader in two directions, and removing too much is the quieter
  one;
• the refusal a reader actually walks into: highlight a passage, then decide
  to redact it. The highlight is an annotation over those words and its note
  can quote them, so it has to go first;
• that marking and redacting answer the same question about where with the
  same arithmetic, so highlighting a line and then redacting it gets the same
  words both times.

Text only — no front matter, navigation or shortcode changes.

Companions: app#79 (the verb),
ops#68 and
ops#69 (the library).

🤖 Generated with Claude Code