Playbooks
Practitioner essays on course design, Backward Design, AI in higher education, and the discipline of getting Stage 1 right. Written for faculty, instructional designers, and program directors who want to do the work properly.
Each course looks fine alone; together a program over-teaches some outcomes and never assesses others. The curriculum map rolls program outcomes across every course — exposing the outcome each course assumes someone else covers and the redundancy nobody planned.
The take-home essay assumed the artifact proved the understanding, and AI severed that link. The instrument that survives is the oral defense — now runnable for the whole class, not just the honors section.
Multi-section courses grade inconsistently because a holistic rubric hides the standard inside adjectives. Convert it to an analytic grid your co-instructors apply the same way — in an afternoon.
A re-lecture gives students the feeling of knowing without the practice of doing. Build the last class from the exam blueprint: retrieval practice matched to what the final asks, attempts before explanation, and a study plan students own.
Most final exams are written in the last weeks of term, from memory of what was covered. A blueprint decides first what the exam should measure — outcomes by kinds of thinking, weighted on purpose — while there's still time to teach toward it.
Assigned in week one, ignored until the last weekend, graded as if it measured the whole term. Designed backward instead: milestones that are real pieces of the finished work, feedback timed to be used, and a process record that is hard to fake.
End-of-term evaluations are exit interviews — the students who answer can't benefit. The week-6 three-question instrument, the count-first de-prescribed reading, one sustainable change plus one principled refusal, and the report-back that makes the loop real.
A hard question and a broken question are identical in the gradebook — and only one should cost points. The one-afternoon item grid, three reads: discrimination and distractors, outcome clusters against the alignment matrix, and the at-risk rows worth a named invitation in week five.
Three weeks before term, an adjunct inherits an LMS export and a syllabus — the furniture without the floor plan. The six-part handoff packet: outcomes with intent, weights with reasons, calibrated rubrics, known failure points, and the drift log that turns handoffs into iteration.
A list of good assessments is not a coherent plan of assessment. Stage 3 is where you make a semester of graded work cover every outcome once, well, without redundancy or waste.
Every syllabus makes promises; the gradebook records what got measured. The alignment matrix checks the two against each other — exposing the outcome nothing tests and the exam mapped to nothing.
Stage 1 gives you outcomes. Stage 2 decides how you'll know students met them — and AI detonated the old shortcut of treating finished work as proof. The two kinds of evidence that survive.
Most faculty say they use Backward Design. Most syllabi reveal Stage 1 was skipped. AI didn't break our assessments — it revealed that Stage 1 was already broken.