Stackelberg vs. Nash in the Lottery Colonel Blotto Game

Yan Liu,Bonan Ni,Weiran Shen,Zihe Wang,Jie Zhang

2024-10-15

Abstract:Resource competition problems are often modeled using Colonel Blotto games. However, Colonel Blotto games only simulate scenarios where players act simultaneously. In many real-life scenarios, competition is sequential, such as cybersecurity, cloud services, Web investments, and more. To model such sequential competition, we model the Lottery Colonel Blotto game as a Stackelberg game. We solve the Stackelberg equilibrium in the Lottery Colonel Blotto game in which the first mover's strategy is actually a solution to a bi-level optimization problem. We develop a constructive method that allows for a series of game reductions. This method enables us to compute the leader's optimal commitment strategy in a polynomial number of iterations. Furthermore, we identify the conditions under which the Stackelberg equilibrium aligns with the Nash equilibria. Finally, we show that by making the optimal first move, the leader can improve utility by an infinite factor compared to its utility in the Nash equilibria. We find that the player with a smaller budget has a greater incentive to become the leader in the game. Surprisingly, even when the leader adopts the optimal commitment strategy, the follower's utility may improve compared to that in Nash equilibria.

Computer Science and Game Theory

What problem does this paper attempt to address?

### Problems the Paper Attempts to Solve

This paper attempts to address how to model and analyze sequential competition in resource competition using the Stackelberg game model. Specifically, the paper focuses on how to extend the traditional Colonel Blotto game (a simultaneous resource allocation game) to a sequential action scenario, where one player acts first (the leader) and the other player responds subsequently (the follower). The main objectives of the paper include:

1. **Model Construction**: Model the Lottery Colonel Blotto game as a Stackelberg game, where the leader's strategy is the solution to a bilevel optimization problem.
2. **Equilibrium Solving**: Develop a construction method that can compute the leader's optimal commitment strategy in polynomial time through a series of game simplification steps.
3. **Equilibrium Comparison**: Identify the conditions under which Stackelberg equilibrium and Nash equilibrium coincide, and explore how the leader's utility changes under these conditions.
4. **Utility Improvement**: Demonstrate that through optimal first action, the leader can significantly improve their utility, potentially by an infinite multiple.
5. **Budget Impact**: Investigate whether players with smaller budgets have a greater incentive to become the leader and explore whether the follower's utility can increase when the leader adopts the optimal commitment strategy.

### Main Contributions

1. **Best Response Understanding**: Propose a novel understanding based on the "water-filling" method to describe the follower's best response strategy.
2. **Game Simplification**: Simplify the game by partitioning the battlefield to ensure uniform value distribution across each sub-battlefield. Prove that players' strategies and utilities remain unchanged before and after each game simplification.
3. **Optimal Commitment Strategy Calculation**: By redefining the leader's optimization objective and reducing the search space for the follower's best response strategy, the leader's optimal commitment strategy can be computed in polynomial time.
4. **Equilibrium Conditions**: Provide sufficient and necessary conditions for the consistency of Stackelberg equilibrium and Nash equilibrium.
5. **Extreme Case Analysis**: Construct extreme cases to compare the leader's utility under the optimal commitment strategy with the utility under Nash equilibrium, observe how the utility ratio is affected by the budget ratio of the leader and the follower, and provide an example showing that the follower's utility can still increase when facing the leader's optimal commitment.

### Related Work

1. **Colonel Blotto Game and Its Variants**: Introduce extensive research on the Colonel Blotto game, including the existence of pure Nash equilibrium and the characteristics of mixed Nash equilibrium.
2. **Sequential Colonel Blotto Game**: Discuss two types of sequentiality: the order of players' actions and the order of battlefield appearances. Review the analysis of these two types of sequentiality in related research.
3. **Other Variants and Applications**: Explore the wide applications of the Colonel Blotto game and its variants in fields such as contest design and auctions.

### Conclusion

By modeling the Lottery Colonel Blotto game as a Stackelberg game, this paper successfully addresses the resource allocation problem in sequential competition. Through a series of innovative methods, the paper not only computes the leader's optimal commitment strategy but also reveals the utility change patterns of the leader under different conditions, providing new perspectives for understanding and optimizing resource competition in practice.