为什么最满意的分配反而不稳定？

from collections import deque


def gale_shapley(student_prefs, school_prefs):
    """
    学生提出型延迟接受算法。

    约定：
    1. 学生与学校数量相同。
    2. 每个参与者对另一侧均有完整、严格的偏好排序。
    3. 每所学校只有一个名额。
    """

    students = list(student_prefs)
    schools = list(school_prefs)

    if len(students) != len(schools):
        raise ValueError("本例要求双方人数相同")

    n = len(students)

    if any(set(order) != set(schools) or len(order) != n
           for order in student_prefs.values()):
        raise ValueError("学生志愿必须是所有学校的完整排列")

    if any(set(order) != set(students) or len(order) != n
           for order in school_prefs.values()):
        raise ValueError("学校偏好必须是所有学生的完整排列")

    # 学校的排名，便于常数时间内比较候选人
    rank = {
        school: {
            student: i
            for i, student in enumerate(order)
        }
        for school, order in school_prefs.items()
    }

    next_choice = {
        student: 0
        for student in students
    }

    held = {
        school: None
        for school in schools
    }

    free = deque(students)

    while free:
        student = free.popleft()

        choices = student_prefs[student]
        index = next_choice[student]

        if index >= n:
            continue

        school = choices[index]
        next_choice[student] += 1

        current = held[school]

        if current is None:
            # 学校还没有暂留学生
            held[school] = student

        elif rank[school][student] < rank[school][current]:
            # 新申请人更符合学校的排序
            held[school] = student

            if next_choice[current] < n:
                free.append(current)

        else:
            # 新申请人被拒绝，继续尝试下一个志愿
            if next_choice[student] < n:
                free.append(student)

    return {
        student: school
        for school, student in held.items()
        if student is not None
    }


student_prefs = {
    "甲": ["A", "B", "C"],
    "乙": ["A", "C", "B"],
    "丙": ["B", "C", "A"]
}

school_prefs = {
    "A": ["乙", "丙", "甲"],
    "B": ["甲", "丙", "乙"],
    "C": ["乙", "甲", "丙"]
}

matching = gale_shapley(
    student_prefs,
    school_prefs
)

for student, school in matching.items():
    print(f"{student} → {school}")

甲 → B
乙 → A
丙 → C