What was not being seen
- Grades assigned for compliance, behavior, group work, and access to outside resources — with no measurement of individual learning
- A single aggregate score that told learners, families, and support staff nothing about where a student was or what they needed next
- SPED teachers receiving disaggregated scores via handwritten notes or emails, because the grading system produced no shareable data structure
- A quarterly workflow requiring the same five scores entered into five to seven separate destinations by hand, with no system communication
- No building-wide mechanism for evaluating whether grading practices were equitable, consistent, or aligned to learning objectives
What I saw
If a gradebook is supposed to communicate proximity to learning objectives, then the objectives themselves — not genre labels, not assignment titles — are what belongs there. That one design decision makes every score legible to every stakeholder the moment it’s entered: the learner knows what was measured, the family knows what it means, the SPED teacher has disaggregated data without a separate communication step, and the data warehouse has exactly the structure it needs for upload. The workflow fix wasn’t engineered separately. It became visible because the architecture was already correct.
What I built
- Standards-based gradebook: five permanent domain entries — consistent quarter to quarter, genre to genre, across up to seven consecutive years of secondary English — with student- and family-facing descriptions, pretest and revision scoring structure, and auto-calculated aggregate
- Gallery walk rubric: eight grading malpractice criteria mapped from Feldman’s Grading for Equity, structured for pre-walk review, mid-walk reference, post-walk conference documentation, and recursive comparison across prior walkthroughs
- Gallery walk protocol: randomized small groups tour a randomly selected host teacher’s projected gradebook each week within existing meeting infrastructure — no additional time commitment required
- “If the Target Were Equity”: blog post documenting the origin, design rationale, and organizational change argument behind the full system
Why I built it this way
The gradebook-as-rubric is not a writing solution. It is a measurement solution that happened to be demonstrated first in an English classroom. The same philosophy — make the skill domain the gradebook entry, not the assignment — applies to every subject area that generates grades. The gallery walk rubric scales that logic across a building by making one correctly designed gradebook the calibration standard for every other. When small groups of teachers tour one another’s gradebooks against eight grading malpractice criteria and document what they find, they are not receiving professional development about equitable grading. They are doing it — and generating a longitudinal record that becomes the baseline for the next cycle. The instruction has no choice but to rise to meet the measurement.
Scope
- Subject area: any class that generates grades — not limited to English or writing assessment
- Organizational scale: one classroom → department → building → district
- Implementation infrastructure: functions within existing weekly faculty and PLC meeting schedules
- Demonstrated at Pennfield Schools (175 students, English department): data visibility up 5x, data entry touchpoints down 80%, workflow steps down 62.5%, 10K+ error-risk operations eliminated annually, SPED communication overhead eliminated entirely — department-wide adoption
- Designed to outlast the designer: gallery walk documentation builds its own longitudinal baseline; gradebook architecture self-reinforces once objectives are the measurement units
Why it’ll last
Once the skill domain is the gradebook entry, the system cannot revert to opaque aggregate grades without also reverting the measurement to meaninglessness — and every stakeholder who has learned to read those domain scores will notice. The gallery walk protocol compounds this: each walkthrough produces documentation that becomes the comparison point for the next, so the calibration is recursive by design. No one has to remember to sustain it. The architecture remembers for them.