LLMs as Judges: Measuring Bias, Hinting Effects, and Tier Preferences | Dark Hacker News