Reverse engineering
in Game Theory
Social Choice: aggregation of preferences of different participants toward a single joint decision
Mechanism Design: implement desired social choice in a strategic setting (participants are rational and selfish)
Examples:
Elections
Auctions
Government Policy
Borda count
Each candidate gets n-i points for every voter who ranked him in place i.
Example:
Strategic Voting
Suppose that voter i’s preference A > B > C but all other voters “hate” A. Then voter i may have incentive to misreport e.g. B > A > C
Plurality
We would like to avoid voting rules where strategic voting is encouraged
It is not transparent
the interaction of strategic voters is complex
Finite set of alternatives A={a,b,c,…}
A set of N voters
Example:
Pareto Efficiency:
Dictatorial:
1
If there is an individual i such that F(L1,…,LN) = Li
L1 = (A >1 B >1 C)
L2 = (B >2 A >2 C)
L3 = (B >3 A >3 C)
F(L1,L2,L3)= L2 = (B > A > C)
==> Voter 2 is a dictator
2
L1 = (A >1 B >1 C)
L2 = (B >2 C >2 A)
L3 = (B >3 A >3 C)
F(L1,L2,L3)= L2 = (B > C > A)
==> Voter 2 is a dictator
We would like to design s.w functions that satisfy
Pareto Efficiency
Independence of Irrelevant Alternatives
Non-Dictatorship
Theorem: If |A|≥3 and F satisfies
Pareto Efficiency
and IIA, then F is a dictatorial social welfare function.
strategically manipulated
by some player i, if for some ranking profile L=(L1,…,LN), and for some L’i, we havef(L’i,L-i)≠f(L) and f(L’i,L-i) >i f(L)
Pareto Efficiency: whenever an alternative a is on the top of every individual’s ranking Li then f(L1,…,LN ) = a
1
L1 = (B >1 C >1 A)
L2 = (B >2 A >2 C)
L3 = (B >3 A >3 C)
==> f(L1,L2,L3) = B
2
L1 = (D >1 B >1 C >1 A)
L2 = (D >2 A >2 B >2 C)
L3 = (D >3 C >3 A >3 B)
==> f(L1,L2,L3)=D
3
L1 = (D >1 B >1 C >1 A)
L2 = (B >2 A >2 D >2 C)
L3 = (D >3 C >3 A >3 B)
==> No useful implication for this input
4
Monotonic: If f(L1,…,LN)=a and for every individual i and every alternative b the ranking L’i ranks a above b if Li does, then f(L’1,…,L’N)=a
1
2
Dictatorial: If there is an individual i such that f(L1,…,LN)=a iff a is at the top of i’s ranking: f(L1,…,LN)=f(Li)
1
2
Strategy-Proof or Incentive Compatible: for every individual i, every ranking profile L=(L1,…,LN), and every L’i, f(L’i,L-i)≠f(L) implies that f(L) >i f(L’i,L-i) (and also f(L) <’i f(L’i,L-i))
Monotonic: if f(L1,…,LN)=a and for every individual i and every alternative b the ranking L’i ranks a above b if Li does, then f(L’1,…,L’N)=a
Pareto Efficient: whenever an alternative a is on the top of every individual’s ranking Li then f(L1,…,LN )=a
Dictatorial: If there is an individual i such that f(L1,…,LN)=a iff a is at the top of i’s ranking
We would like to design s.c functions that satisfy
Strategy-Proofness
Non-Dictatorship
Theorem: If |A|≥3 and f is strategy-proof and onto, then f is a dictatorial social choice function.
Onto: For any alternative a 2 A there exists a profile ranking such that f(L)=a.