<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0">
  <channel>
    <title>Supposing blog</title>
    <link>https://supposing.app/blog</link>
    <description>Rankings judged by AI, and how we make them.</description>
    <language>en</language>
    <item>
      <title>Pairwise comparison beats individual scoring</title>
      <link>https://supposing.app/blog/pairwise-comparison-beats-individual-scoring</link>
      <guid>https://supposing.app/blog/pairwise-comparison-beats-individual-scoring</guid>
      <pubDate>Sat, 21 Nov 2026 00:00:00 GMT</pubDate>
      <description>An AI judge ranked 100 essays from the ELLIPSE dataset by comparing them in pairs. The ranking matched trained human raters on most pairs, and on nearly all where the raters saw a clear difference. Scoring each essay individually did much worse.</description>
    </item>
  </channel>
</rss>