← Gallery

About

EADAM (Efficiency-Adjusted Deferred Acceptance) Pareto-improves on student-optimal DA by identifying "interrupters" — students who take slots at schools they later leave, blocking others — and asking them to waive priority with consent.

EADAM Algorithm (Kesten 2010) 
Phase 1: Run student-proposing DA
         → matching M_DA

Phase 2: Identify "interrupters":
  Student s is an interrupter at
  school X if:
  - s was held at X at some point
  - s was later rejected from X
  - s's presence at X caused
    another student to be rejected

Phase 3: Ask each interrupter for
  CONSENT to waive their priority
  at the school where they interrupt

Phase 4: Rerun DA with consenting
  interrupters' priorities removed
  → matching M_EADAM

Result: M_EADAM Pareto-dominates
        M_DA (weakly better for all)

Setup

Students: s1, s2, s3, s4, s5
Schools: A[2], B[2], C[1]

Preferences:
s1: A>B>C | s2: A>B>C
s3: A>B>C | s4: B>A>C
s5: B>C>A

Priorities:
A: s1>s3>s2>s4>s5
B: s2>s4>s1>s3>s5
C: s5>s4>s3>s2>s1

Controls

Properties

✓ Pareto-improves on DA (weakly better for all)
✓ Stable in a weakened sense (priority-neutral)
✓ Respects consent (no one forced to waive)
✗ Not fully strategy-proof (but hard to manipulate)

Real-World Deployment

EADAM has been implemented by the Flemish Ministry of Education in Belgium (2024), marking the first real-world deployment of this mechanism for school assignment.

Event Log

EADAM: Efficiency-Adjusted DA

Kesten (2010) — Pareto-improving on Student-Optimal Deferred Acceptance