Side B fixes URL generation to use `display_path()` instead of stored full paths for vote compare and pool links, aligning hrefs with the user-facing path format, and greatly strengthens the browser test by exercising all 45 pairwise votes and verifying the final RPC ranking. Side A is primarily a UI correctness change (per-group rank gradient highlighting plus vote-history slider visualization) with useful regression tests, but its impact is narrower than the routing fix and comprehensive end-to-end validation added in Side B.
constitution · epochs · watch · epoch 3 · comparison · attempt
judgment
openai/gpt-chat-latest → B (4:1)
jud_b57a4ee9011544 · raw event
Metadata
judgment_idjud_b57a4ee90115449e2b5e65057281ed6422413a4977c7b5138cfa67f2f4dcfc9d
model_idopenai/gpt-chat-latest
winnerB
ratio4:1
comparison_idcmp_f753887b01e61e07263f82558a4b2c621787a257f19cb4dbcbbad3ea2bc3a626
attempt_idatt_5b0172e1dd2009664af4bd023fa683f6f55903fbcd1add8261e389d3aee775a1