One of the most common administrative tasks that organizations perform is the assignment of people to positions based on the people’s preferences. A computer program that chooses an assignment automatically would be useful, but what assignment rule should it follow? I proposed the rule of least-unpopularity-factor matching, which is difficult to subvert and always designates a “best” assignment—but I proved that there is no practical way to actually compute that assignment for large numbers of people and positions. The result eliminates one possibility in the search for a rule suitable for an automatic assignment program but suggests where a suitable rule might be found.
[1]
Kurt Mehlhorn,et al.
Popular matchings
,
2005,
SODA '05.
[2]
Telikepalli Kavitha,et al.
Dynamic Matching Markets and Voting Paths
,
2006,
SWAT.
[3]
P. Gärdenfors.
Match making: Assignments based on bilateral preferences
,
1975
.
[4]
Andrew V. Goldberg,et al.
Scaling algorithms for the shortest paths problem
,
1995,
SODA '93.
[5]
Kurt Mehlhorn,et al.
Rank-maximal matchings
,
2004,
TALG.
[6]
A. Roth.
The Evolution of the Labor Market for Medical Interns and Residents: A Case Study in Game Theory
,
1984,
Journal of Political Economy.