Context
Observations from an RLCR session (3 rounds to completion, 10 acceptance criteria). All content is fully sanitized — no project-specific information.
Suggestions
1. Require AC-Level Self-Review Before Declaring Completion
The implementer declared all criteria met based on passing tests, but 40% were only partially complete. Adding a per-criterion evidence checklist to the summary template would catch this before the review round.
2. Review Fix Generality, Not Just Correctness
Narrow fixes that address the specific reported instance but miss the general case caused an extra round. Reviewers should explicitly check whether a fix handles analogous cases beyond the one reported.
3. Eliminate Low-Value Boilerplate Sections
The BitLesson Delta section said "none" in every round with zero information value. Consider making it optional or requiring meaningful reflection instead.
4. Distinguish "Tests Pass" from "Criteria Met"
A passing test suite created false confidence when the suite didn't cover all criterion requirements. The summary template should include an "Uncovered criteria or scenarios" field to force implementers to think about what tests don't cover.
5. Consider Consolidating Small Fix Rounds
When a review finds <3 residual issues in the same area as just-fixed work, deeper reviewer guidance (pseudo-code, specific constraints) could eliminate narrow-fix iterations and reduce total round count.
Impact
These changes could reduce the typical round count by ~1 round for sessions with 5+ acceptance criteria, primarily by catching partial implementations earlier.
Context
Observations from an RLCR session (3 rounds to completion, 10 acceptance criteria). All content is fully sanitized — no project-specific information.
Suggestions
1. Require AC-Level Self-Review Before Declaring Completion
The implementer declared all criteria met based on passing tests, but 40% were only partially complete. Adding a per-criterion evidence checklist to the summary template would catch this before the review round.
2. Review Fix Generality, Not Just Correctness
Narrow fixes that address the specific reported instance but miss the general case caused an extra round. Reviewers should explicitly check whether a fix handles analogous cases beyond the one reported.
3. Eliminate Low-Value Boilerplate Sections
The BitLesson Delta section said "none" in every round with zero information value. Consider making it optional or requiring meaningful reflection instead.
4. Distinguish "Tests Pass" from "Criteria Met"
A passing test suite created false confidence when the suite didn't cover all criterion requirements. The summary template should include an "Uncovered criteria or scenarios" field to force implementers to think about what tests don't cover.
5. Consider Consolidating Small Fix Rounds
When a review finds <3 residual issues in the same area as just-fixed work, deeper reviewer guidance (pseudo-code, specific constraints) could eliminate narrow-fix iterations and reduce total round count.
Impact
These changes could reduce the typical round count by ~1 round for sessions with 5+ acceptance criteria, primarily by catching partial implementations earlier.