IteratedUltimatumGame

Description

IteratedUltimatumGame is an environment for evaluating agents on fairness, negotiation, and strategic bargaining. This environment wraps the IteratedUltimatumGame implementation from TextArena, a framework for text-based game environments.

Capabilities

• Fairness and bargaining strategy
• Sequential decision-making with role alternation
• Reputation building and threat credibility
• Competitive gameplay against an LLM opponent

Compute Requirements

IteratedUltimatumGame does not require a sandbox. It has minimal compute requirements.

License

MIT.

Tasks

There are two splits: train (300 tasks) and test (300 tasks). Each split contains 50 tasks across each of 6 variants:

• IteratedUltimatumGame-v0
• IteratedUltimatumGame-v0-train
• IteratedUltimatumGame-v0-raw
• IteratedUltimatumGame-v0-alternate
• IteratedUltimatumGame-v0-alternate-train
• IteratedUltimatumGame-v0-alternate-raw

Each task is seeded for reproducibility.

Reward Structure

This is a sparse reward environment. Rewards are mapped from TextArena's native range of {-1, 0, 1} to {0.0, 0.5, 1.0} via (raw + 1) / 2.

We do not use LLM graders for this environment; reward is determined programmatically.

Data

Game state is generated procedurally by the TextArena engine using seeded randomness. No external data files are required.

Tools

Agents are given two tools:

• propose_split(amount): Propose a split as the proposer. Specify the amount (in dollars) you are offering to the other player.
• respond(decision): Respond to a proposal as the responder. Decision must be 'accept' or 'reject'.

Time Horizon

IteratedUltimatumGame is a multi-turn environment.

Environment Difficulty

Medium - requires understanding of fairness norms, credible threat assessment, and the ability to balance short-term gains with long-term reputation effects across multiple rounds.

Other Environment Requirements

This environment requires an OpenAI API key (passed via secrets) to power the LLM opponent.

Safety

Agents in IteratedUltimatumGame interact only with a bargaining game simulation and have no access to external systems, the internet, or sensitive data. The environment does not present safety risks.

Citations

@software{textarena2024,
  author    = {Guertler, Leon and Banting, Wilfried and Pignatelli, Eduardo},
  title     = {TextArena},
  year      = {2024},
  publisher = {GitHub},
  url       = {https://github.com/LeonGuertler/TextArena}
}