The 85/15 calibration rule is the targeting system for productive struggle in the Lacefield Adaptive Learning System. It sets the session difficulty target so that approximately 85% of practice problems fall at the student's current growth edge — hard enough to force genuine engagement, not so hard that confusion collapses into frustration. The remaining 15% targets mastered material for fluency reinforcement and confidence maintenance.
This is not arbitrary. The ratio is grounded in Manu Kapur's productive failure research, Vygotsky's zone of proximal development, and seven years of direct classroom observation of where the productive struggle zone actually sits for adult learners with significant foundational gaps. The system uses the student's DLP to calculate the exact growth-edge target for each session and recalibrates after every session based on actual performance data.
This single rule is what turns generic adaptive learning into high-fidelity adaptive learning. Without calibration precision, difficulty adjustment is guesswork. With it, every session is structurally designed to produce growth.
Problems at the student's current capability boundary. Target accuracy: ~80%. Student should be working hard, succeeding most of the time. This is where schema development happens.
Problems on already-mastered material. Target accuracy: 90%+. Not filler — confidence infrastructure. Students who experience only difficulty lose belief that progress is possible.
Growth-edge accuracy ≥ 90%: Difficulty up. Student has exceeded the growth-edge target. Current tier is no longer the boundary — move up one tier and reassess.
Growth-edge accuracy 75–89%: Hold. Student is in the productive zone. No change to calibration state.
Growth-edge accuracy 60–74%: Flag for review. Student may be at the lower edge of productive struggle or approaching destructive frustration. Assess engagement quality before adjusting.
Growth-edge accuracy < 60%: Difficulty down. Not a student failure — a calibration failure. Drop back to where the student can succeed and rebuild from there.
The calibration_state object in the DLP stores the current calibration parameters and is updated after every session. It is per-student, not per-subject — calibration tracks the individual, not the curriculum level.
The 85% growth-edge target is not arbitrary. Kapur (2010, 2016) on productive failure shows that learning consolidates most effectively when students work at the boundary of current capability with appropriate scaffolding. Vygotsky's zone of proximal development defines the same region conceptually. The specific 85% figure emerged from classroom observation of where the productive struggle zone sits for adult learners with significant educational gaps — populations where the zone is narrower and more precisely calibrated difficulty matters more, not less.