Short-Story · Short fiction
What it measures and how
600–800-word stories with 10 creative requirements, identical prompts, three judges and bidirectional comparisons. Use the main Thurstone score; BT is diagnostic only and adds no votes. Multiple tiers follow fixed selection rules.
How this evidence is used
Under observation: integration awaits clear formal data reuse terms and resolution of differences between the README and machine-readable results.
Limitations and attribution
Model judges have literary preferences; some models have not completed every story. Short-story performance does not establish long-form fiction or Chinese writing quality.
Data license: No explicit leaderboard data reuse license found. A public GitHub repository does not imply open authorization.
Results are published by Lech Mazur. Raw scores and the e/acc consensus index use different scales and cannot be added directly.