docs: archive two-axis optical result and falsifiers

harness/DECISIONS.md

@@ -498,3 +498,22 @@ editing remains separate from semantic reconnect. The [decision and evidence
 map](edge-attachments-20260922.md) records alternatives, lifecycle rules,
 independent falsifiers, evaluators and the Process/Scenario applicability limits.
 This is branch-specific work; it does not establish an integrated release.
+
+## Two-axis Focus optical scope — 2026-10-01
+
+Use a private, opt-in inert runner to measure complete-scene magnification before
+implementing an editable Focus workspace. Width-only recovery was rejected for
+this question because height limits Fit; a hand-built blank Baseline would not
+represent real chrome. Freeze complete model/projection/routes/labels/ports and
+reuse production drawing with actual measured viewports. Reject unmeasured or
+stale captures; draw acknowledgment is CPU evidence, not GPU completion.
+
+The [predeclared comparison](experiments/focus-optical-20261001/plan.md) and
+[native result](focus-optical-validation-20261001.md) establish substantial
+two-axis gains on the mixed and nested examples while conserving whole recorded
+values/bytes. They do not establish unassisted reading: fresh-reader availability
+failed and shared runs/foreign-label crossings remain. Retain a workspace
+hypothesis without promoting the experiment or closing #38. Category recognition
+and individual connection identity are distinct next questions. Editable
+promotion must separately prove draft/caret/focus/input, camera, gesture,
+navigation, history/save/reopen and native behavior with #43 coordination.