Part 8 · Distributing students across institutions
Matching & Distribution Model
Five distribution principles, a five-stage explainable matching algorithm with deferred-acceptance structure, nine specified edge cases, and a simulation design held to the evidence discipline.
Distribution principles
Students are never distributed by dividing the national cohort equally among offices. Distribution follows five principles: (1) discipline match first — a placement that wastes a student's training is a failed placement; (2) near-home priority — travel cost is the dominant equity risk (RSK-03), so home-district matching is the default; (3) capacity honesty — hosting density capped by supervisor ratio, never by building count; (4) transparent rules — the algorithm is published, configurable by the steering committee, and reproducible; (5) explainability — every student's placement outcome can be reconstructed from logged inputs, rules and seed.
The matching algorithm
Matching runs in five stages per placement cycle:
The recommended implementation is a constrained deferred-acceptance style matching (students ordered by priority score applying to seat queues; seats capacity-constrained), which is strategy-proof in the classic setting and produces reproducible, explainable outcomes. The exact weights are a steering-committee configuration item, published before every cycle and auditable afterwards.
Edge cases — specified, not improvised
| Case | Rule |
|---|---|
| More applicants than seats | Priority scoring + waitlist with carry-over priority; guaranteed-placement rule (RSK-02): a university cannot require the service semester until a confirmed placement exists for every mandated student in the cohort |
| More seats than applicants | Seats released to the national pool mid-cycle; utilization published (anti-hoarding) |
| Highly specialized programs | Reserved discipline-matched seats (e.g., clinical-adjacent for health students); cross-university exchange of unmatched specialized seats |
| Remote and underserved areas | Location weighting compensates hardship postings; travel/accommodation support from the need-based fund (UFM-08) |
| Students who cannot travel | Home-district default; accessibility accommodations (UFM-07); remote-eligible tasks where the seat specification permits |
| Students with disabilities | Accessibility-first matching: only seats meeting documented accommodation requirements; CRPD-aligned obligation on hosts |
| Cancelled placements | Automatic re-entry into the same window's pool with priority credit; no academic penalty for administrative cancellation |
| Students approaching graduation | Final-year urgency weighting — never left to compete with first-takers for scarce seats |
| Offices withdrawing seats | Published reallocation round; the office's future capacity validation flagged for review |
Simulation (hypothetical, labelled as such)
The master research protocol requires the allocation model to be tested with simulated data clearly labelled as hypothetical. In this blueprint's evidence discipline, that simulation is specified but not yet run on the real matched datasets (it is a PH-0 platform task with the pilot's real application data). The simulation design: 130,178 eligible applicants (PL-14), ~140,000 published seats across 10 sectors, five-cycle run, metrics = placement rate, discipline-match rate, home-district rate, unmatched rate, utilization by sector, equity-flag coverage. No synthetic results are presented as findings. The pilot produces the first real numbers; the gate criteria in the roadmap then decide expansion on evidence.