Comparative Output Evaluation Prompt

# Comparative Output Evaluation Prompt

You are an evaluation systems designer. Build a comparative evaluation prompt for [COMPARISON TASK] — choosing between strategies, written drafts, designs, or proposed solutions.

## Evaluation Rubric Design

Build the scoring rubric for [COMPARISON TASK]:
- 4-6 evaluation criteria with clear definitions
- Weighting for each criterion (must sum to 100%)
- Scoring scale: 1-5 or 1-10 with anchor descriptions for each score level
- A mandatory "fatal flaw" check that disqualifies an option regardless of score

## Evaluation Prompt Template

Write the evaluation prompt that:
- Presents the rubric before asking for evaluation
- Asks the AI to score each option per criterion before comparing
- Requires a final ranked recommendation with reasoning
- Includes a devil-advocate check: what is the strongest argument for the option ranked last?

## Anti-Bias Instructions

Add instructions that reduce common AI evaluation biases:
- Recency bias: do not favor the last option presented
- Length bias: longer responses are not automatically better
- Confidence bias: more assertive writing is not more correct

## Example Evaluation

Run a full evaluation of [3 OPTIONS] for [COMPARISON TASK] using the rubric above, showing the score breakdown and final recommendation.