{
  "id": 1214,
  "url": "https://arxiv.org/abs/2606.10569v1",
  "title": "Hidden Consensus:Preference-Validity Compression in Human Feedback",
  "summary": "Standard RLHF pipelines often reduce heterogeneous human judgments into a single scalar reward target. We argue that this reduction can mis-measure alignment in structurally plural societies, where disagreement may reflect culturally, historically, linguistically, regionally, or normatively grounded interpretations rather than annotation noise. We call this failure Preference-Validity Compression, the collapse of multiple plural-valid response options into a single optimization target. Using Mal",
  "authors": "Dorcas Chia Ern Chua, Karen Myn Hui Lee, Jia Yue Tan, Zhen Xue Gue, Norzalena Abdul Hamid, Azima Binti Azmi et al.",
  "category": "research",
  "topics": "safety-alignment",
  "orgs": null,
  "regions": null,
  "published_at": "2026-06-09T08:32:11.000Z",
  "fetched_at": "2026-07-14T14:15:07.843Z",
  "source_slug": "arxiv-ethics",
  "source_name": "arXiv",
  "source_homepage": "https://arxiv.org",
  "ethics_ai_record_url": "https://ethics.ai/record/1214",
  "original_url": "https://arxiv.org/abs/2606.10569v1",
  "evidence_status": "source-only",
  "attribution": "via ethics.ai"
}