Subject Matter Requirements (SMRs)

If the previous lesson was the org chart, this one is the core data structure. Subject Matter Requirements (SMRs) are the Commission's formal definition of what a teacher must know in a given subject — the standard every competency decision is measured against, and therefore the backbone of any transcript-analysis system.

What an SMR actually is

For each credentialable subject (mathematics, English, science, ...), the Commission publishes a structured document dividing the subject into domains — major content areas — and each domain into sub-areas with specific competency statements. Mathematics might break into algebra, geometry, number theory, probability and statistics, and calculus; the geometry domain then enumerates what "knows geometry" concretely means (parallelism, congruence and similarity, three-dimensional objects...). Two properties matter to you as a developer. First, SMRs are hierarchical and enumerable — subject → domain → sub-area — which is why they can be modeled as data rather than prose. Second, they are versioned in time: SMRs get revised when standards change, and a candidate evaluated under the 2020 mathematics SMRs must stay traceable to that version even after 2025 revisions land. Model the version as a first-class fact, not an afterthought.

Two roads to "competent": exam or coursework

A candidate demonstrates subject matter competency in one of two main ways (statute adds variants, but these dominate):

• Examination. Pass the state's subject examination for that subject — in California, the CSET. Clean signal: pass/fail per subtest, dated, machine-verifiable. Scores can expire, and exams map to subtests covering groups of domains, not always one-to-one.
• Coursework. Show that completed college coursework (or a degree major in the subject) covers the SMR domains. Rich signal, but messy: someone must judge whether "MATH 3B: Analytic Methods" actually covers the calculus domain.

Mixing is allowed in many cases — coursework covering some domains, subtest passes covering the rest — which turns evaluation into a coverage problem: every domain must be satisfied by some qualifying evidence.

The mapping problem, developer's view

Strip away the terminology and the coursework route is a matching engine: on one side a normalized SMR structure (Subject → Domain → SubArea, versioned); on the other, evidence extracted from transcripts (courses with titles, descriptions, units, grades) and exam records; between them, mapping judgments — this course satisfies that domain, at this confidence, per this rule or reviewer. The output is a coverage determination: which domains are met, by what evidence, and which remain as gaps a candidate can close with targeted coursework or a subtest. Hold onto that shape. When later lessons discuss AI-assisted transcript analysis, the AI's role is confined to proposing mappings inside this structure — the structure itself, and the final determination, stay human-defined. Getting the SMR model right is what makes that division of labor possible.