{"messages":[{"content":"You are a constitutional council ranking individual git commits for ownership allocation.\n\nCompare these two commits. Decide which contributed more lasting value to the project.\n\nJudge substance, not spectacle:\n- Prefer correct, lasting design and real bugfixes over churn, formatting, renames, or generated noise.\n- Prefer clarity and necessity over sheer line count. A small precise change can beat a large diffuse one.\n- Do not favor a side merely because its patch is longer or noisier.\n- Weight what the change does for the project, not the contributor's name.\n\nReturn ONLY a JSON object: {\"winner\": \"A\" or \"B\", \"ratio\": \"N:M\", \"explanation\": \"...\"}\nThe explanation must cite concrete differences in the patches (1-3 sentences).\n\nSide A — contributor: tommy-mor\nSide A — commit message:\n[81de487b] Fix zero-ratio guard in reducer to drop before registering items or pair.\n\nPreviously the early-return for zero-weight votes happened after\nensure_item and voted_pairs.insert, leaving ghost items in the index\nand the pair incorrectly marked as voted. Move the check to before\nany side effects.\n\nCo-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>\n\nSide A — unified diff (full patch):\ndiff --git a/server/src/reducer.rs b/server/src/reducer.rs\nindex 6841d35cfc9de2389f340a22b8a45acb335e36c3..0e36979abe0f051493038ff7e652efc7f7a0ac80 100644\n--- a/server/src/reducer.rs\n+++ b/server/src/reducer.rs\n@@ -112,6 +112,10 @@ impl GroupState {\n         if vote.ratio_right < 0 {\n             vote.ratio_right = 0;\n         }\n+        if vote.ratio_left == 0 || vote.ratio_right == 0 {\n+            // Zero on either side produces no valid edge; drop before registering items or pair.\n+            return;\n+        }\n \n         let a_idx = self.ensure_item(&vote.a);\n         let b_idx = self.ensure_item(&vote.b);\n@@ -121,10 +125,6 @@ impl GroupState {\n \n         let w_a = vote.ratio_left as f64;\n         let w_b = vote.ratio_right as f64;\n-        if w_a == 0.0 || w_b == 0.0 {\n-            // Zero on either side produces no valid edge; drop the vote.\n-            return;\n-        }\n \n         self.add_edge_weight(b_idx, a_idx, w_a);\n         self.add_edge_weight(a_idx, b_idx, w_b);\ndiff --git a/server/tests/basic.rs b/server/tests/basic.rs\nindex cc8c1a0d139f3722ba6ecd13dd001c65be835b67..08159f4a7f0850fd165817a4a1af4f31ced2ad76 100644\n--- a/server/tests/basic.rs\n+++ b/server/tests/basic.rs\n@@ -546,12 +546,10 @@ fn reducer_negative_ratio_clamped_to_zero() {\n         delegate: Some(\"00000000-0000-0000-0000-000000000000:test:local/test\".to_string()),\n         thread_tag: \"t\".to_string(),\n     });\n-    // Items are registered, but the zero-clamped vote produces no edges.\n-    assert_eq!(group.idx_to_item.len(), 2);\n-    let a_idx = group.item_to_idx[&item_id(\"https://slug.social/~/t/a\")];\n-    let b_idx = group.item_to_idx[&item_id(\"https://slug.social/~/t/b\")];\n-    assert!(!group.edges.contains_key(&(a_idx, b_idx)));\n-    assert!(!group.edges.contains_key(&(b_idx, a_idx)));\n+    // Nothing registered: zero-clamped vote is dropped before ensure_item.\n+    assert!(group.idx_to_item.is_empty());\n+    assert!(group.edges.is_empty());\n+    assert!(group.voted_pairs.is_empty());\n }\n \n \n\n\nSide B — contributor: tommy-mor\nSide B — commit message:\n[c7ef287e] Rank every eligible commit with the LLM council.\n\nStop short-circuiting on a single contributor; pairwise-sort commits, roll scores up for emission payouts, and surface commit rankings on epoch pages.\n\nCo-authored-by: Cursor <cursoragent@cursor.com>\n\nSide B — unified diff (full patch):\ndiff --git a/constitution.py b/constitution.py\nindex 26ba130e885e0e69fb7874ca5c3f07f42100a150..71fc46b4c9a4d7a9ea7bf319860b0db1acc4a673 100644\n--- a/constitution.py\n+++ b/constitution.py\n@@ -503,20 +503,22 @@ def _epochs_in_ledger() -> list[int]:\n \n \n def build_pairwise_prompt(side_a: dict, side_b: dict) -> str:\n-    return f\"\"\"You are ranking contributions to an open source project.\n-Compare these two sides (each may be one or more commits). Decide which side contributed more.\n+    return f\"\"\"You are ranking individual git commits to an open source project.\n+Compare these two commits. Decide which commit contributed more.\n Return ONLY a JSON object: {{\"winner\": \"A\" or \"B\", \"ratio\": \"N:M\", \"explanation\": \"...\"}}\n \n-Side A — commit messages:\n+Side A — contributor: {side_a.get('contributor', '?')}\n+Side A — commit message:\n {side_a['message']}\n \n-Side A — unified diffs (full patches):\n+Side A — unified diff (full patch):\n {side_a['diff']}\n \n-Side B — commit messages:\n+Side B — contributor: {side_b.get('contributor', '?')}\n+Side B — commit message:\n {side_b['message']}\n \n-Side B — unified diffs (full patches):\n+Side B — unified diff (full patch):\n {side_b['diff']}\"\"\"\n \n \n@@ -1518,16 +1520,28 @@ async def broadcast_js(js: str):\n         await queue.put(js)\n \n \n-def _author_side_for_llm(author: str, author_commits: dict) -> dict:\n-    cs = author_commits[author]\n+def _commit_side_for_llm(row: dict) -> dict:\n+    oid = row[\"oid\"]\n+    short = oid.split(\":\", 1)[1][:8] if \":\" in oid else oid[:8]\n     return {\n-        \"message\": \"\\n\".join(f\"[{c['sha']}] {c['message']}\" for c in cs),\n-        \"diff\": \"\\n\\n\".join(f\"=== {c['sha']} ===\\n{c['diff']}\" for c in cs),\n-        \"commit_ids\": [c[\"commit_id\"] for c in cs],\n-        \"contributor\": author,\n+        \"message\": f\"[{short}] {row['message']}\",\n+        \"diff\": row[\"patch\"] or \"\",\n+        \"commit_id\": commit_id_for_oid(oid),\n+        \"contributor\": row[\"contributor\"],\n+        \"oid\": oid,\n     }\n \n \n+def _rollup_contributor_scores(\n+    ordered: list[dict], commit_scores: list[Decimal]\n+) -> dict[str, Decimal]:\n+    totals: dict[str, Decimal] = {}\n+    for row, score in zip(ordered, commit_scores):\n+        contributor = row[\"contributor\"]\n+        totals[contributor] = totals.get(contributor, Decimal(\"0\")) + score\n+    return totals\n+\n+\n def _find_judgment(comparison_id: str, model_id: str) -> dict | None:\n     for e in evidence_by_kind(\"llm.judgment\"):\n         p = e.payload\n@@ -1546,101 +1560,106 @@ def _find_ranking_models(ranking_run_id: str) -> list[str] | None:\n \n \n async def rank_commits(commits: list[dict], *, epoch: int = -1):\n+    \"\"\"Pairwise-rank every eligible commit; roll scores up to contributors.\"\"\"\n     if not commits:\n         return {}, [], {\"ranking_run_id\": \"\", \"ranking_event_id\": \"\"}\n \n-    commit_ids = sorted(commit_id_for_oid(row[\"oid\"]) for row in commits)\n+    ordered = sorted(commits, key=lambda r: r[\"oid\"])\n+    commit_ids = [commit_id_for_oid(row[\"oid\"]) for row in ordered]\n     ranking_run_id = _content_id(\"rank\", {\n         \"epoch\": epoch,\n-        \"commit_ids\": commit_ids,\n+        \"commit_ids\": sorted(commit_ids),\n     })\n-    contributors = sorted(set(c[\"contributor\"] for c in commits))\n+    contributors = sorted({c[\"contributor\"] for c in ordered})\n \n-    if len(contributors) == 1:\n+    # Nothing to compare: a single commit (not a single contributor).\n+    if len(ordered) == 1:\n         await append_evidence(epoch, \"ranking.started\", {\n             \"ranking_run_id\": ranking_run_id,\n             \"commit_ids\": commit_ids,\n             \"contributors\": contributors,\n             \"models\": [],\n-            \"summary\": f\"ranking epoch {epoch}: single contributor\",\n+            \"summary\": f\"ranking epoch {epoch}: single commit\",\n         })\n-        ranking = {contributors[0]: Decimal(\"1\")}\n+        commit_ranking = {commit_ids[0]: \"1\"}\n+        contributor_ranking = {ordered[0][\"contributor\"]: Decimal(\"1\")}\n         completed = await append_evidence(epoch, \"ranking.completed\", {\n             \"ranking_run_id\": ranking_run_id,\n             \"models\": [],\n-            \"ranking\": {contributors[0]: \"1\"},\n+            \"commit_ranking\": commit_ranking,\n+            \"contributor_ranking\": {ordered[0][\"contributor\"]: \"1\"},\n+            \"ranking\": {ordered[0][\"contributor\"]: \"1\"},\n             \"judgment_ids\": [],\n-            \"summary\": f\"Only {contributors[0]} is eligible; rank is 1.0\",\n+            \"summary\": f\"Only one eligible commit; {ordered[0]['contributor']} rank 1.0\",\n         })\n         await broadcast_audit(\n             \"ranking\",\n-            f\"Only {contributors[0]} is eligible; rank is 1.0\",\n+            f\"Only one eligible commit; {ordered[0]['contributor']} rank 1.0\",\n             progress=90,\n             phase=\"finalizing\",\n             evidence_event_id=completed.event_id,\n             evidence_url=_evidence_url(\"event\", completed.event_id),\n             links={\"epoch\": _evidence_url(\"epoch\", str(epoch))},\n         )\n-        return ranking, [], {\n+        return contributor_ranking, [], {\n             \"ranking_run_id\": ranking_run_id,\n             \"ranking_event_id\": completed.event_id,\n         }\n \n     if not (OPENROUTER_API_KEY or \"\").strip():\n         raise RuntimeError(\n-            \"OPENROUTER_API_KEY is required when multiple contributors need ranking\"\n+            \"OPENROUTER_API_KEY is required when multiple commits need ranking\"\n         )\n \n     models = _find_ranking_models(ranking_run_id)\n     if models is None:\n         models = await fetch_top_models(n=3)\n     if not models:\n-        raise RuntimeError(\"no council models available for contributor ranking\")\n+        raise RuntimeError(\"no council models available for commit ranking\")\n     await append_evidence(epoch, \"ranking.started\", {\n         \"ranking_run_id\": ranking_run_id,\n         \"commit_ids\": commit_ids,\n         \"contributors\": contributors,\n         \"models\": models,\n-        \"summary\": f\"Council selected: {', '.join(models)}\",\n+        \"summary\": (\n+            f\"Council selected: {', '.join(models)} — \"\n+            f\"{len(ordered)} commits\"\n+        ),\n     })\n     await broadcast_audit(\n         \"council\",\n-        f\"Council selected: {', '.join(models)}\",\n+        f\"Council selected: {', '.join(models)} — ranking {len(ordered)} commits\",\n         progress=35,\n         phase=\"ranking\",\n     )\n     await broadcast_js(exec_event(Three[Selector(\"#emission-log\")][PREPEND][\n-        [\"div.log-council\", f\"Council: {', '.join(models)} — {len(commits)} commits\"]\n+        [\"div.log-council\",\n+            f\"Council: {', '.join(models)} — {len(ordered)} commits\"]\n     ]))\n \n-    authors = contributors\n-    author_commits = {a: [] for a in authors}\n-    for row in sorted(commits, key=lambda r: r[\"oid\"]):\n-        author_commits[row[\"contributor\"]].append({\n-            \"message\": row[\"message\"],\n-            \"sha\": row[\"oid\"].split(\":\", 1)[1][:8],\n-            \"diff\": row[\"patch\"],\n-            \"commit_id\": commit_id_for_oid(row[\"oid\"]),\n-        })\n-\n+    sides = [_commit_side_for_llm(row) for row in ordered]\n     judgment_ids: list[str] = []\n \n     async def compare_fn(i, j):\n-        a1, a2 = authors[i], authors[j]\n-        side_a = _author_side_for_llm(a1, author_commits)\n-        side_b = _author_side_for_llm(a2, author_commits)\n+        side_a, side_b = sides[i], sides[j]\n+        label_a = f\"{side_a['commit_id'][:16]} ({side_a['contributor']})\"\n+        label_b = f\"{side_b['commit_id'][:16]} ({side_b['contributor']})\"\n         prompt = build_pairwise_prompt(side_a, side_b)\n         comparison_material = {\n             \"ranking_run_id\": ranking_run_id,\n             \"side_a\": {\n-                \"contributor\": a1,\n-                \"commit_ids\": side_a[\"commit_ids\"],\n+                \"contributor\": side_a[\"contributor\"],\n+                \"commit_id\": side_a[\"commit_id\"],\n+                \"commit_ids\": [side_a[\"commit_id\"]],\n+                \"oid\": side_a[\"oid\"],\n                 \"message\": _bytes_blob(side_a[\"message\"]),\n                 \"diff\": _bytes_blob(side_a[\"diff\"]),\n             },\n             \"side_b\": {\n-                \"contributor\": a2,\n-                \"commit_ids\": side_b[\"commit_ids\"],\n+                \"contributor\": side_b[\"contributor\"],\n+                \"commit_id\": side_b[\"commit_id\"],\n+                \"commit_ids\": [side_b[\"commit_id\"]],\n+                \"oid\": side_b[\"oid\"],\n                 \"message\": _bytes_blob(side_b[\"message\"]),\n                 \"diff\": _bytes_blob(side_b[\"diff\"]),\n             },\n@@ -1650,22 +1669,24 @@ async def rank_commits(commits: list[dict], *, epoch: int = -1):\n         comparison_material = {\n             **comparison_material,\n             \"comparison_id\": comparison_id,\n-            \"summary\": f\"Comparing {a1} with {a2}\",\n+            \"summary\": f\"Comparing {label_a} with {label_b}\",\n         }\n         cmp_ev = await append_evidence(epoch, \"comparison.input\", comparison_material)\n         await broadcast_audit(\n             \"comparison\",\n-            f\"Comparing {a1} with {a2}\",\n+            f\"Comparing commits {label_a} vs {label_b}\",\n             phase=\"ranking\",\n             evidence_event_id=cmp_ev.event_id,\n             evidence_url=_evidence_url(\"comparison\", comparison_id),\n             links={\n                 \"comparison\": _evidence_url(\"comparison\", comparison_id),\n+                \"commit_a\": _evidence_url(\"commit\", side_a[\"commit_id\"]),\n+                \"commit_b\": _evidence_url(\"commit\", side_b[\"commit_id\"]),\n                 \"epoch\": _evidence_url(\"epoch\", str(epoch)),\n             },\n         )\n         await broadcast_js(exec_event(Three[Selector(\"#emission-status\")][MORPH][\n-            [\"div#emission-status\", f\"Comparing {a1} vs {a2}…\"]\n+            [\"div#emission-status\", f\"Comparing {label_a} vs {label_b}…\"]\n         ]))\n         results = []\n         for model in models:\n@@ -1697,9 +1718,15 @@ async def rank_commits(commits: list[dict], *, epoch: int = -1):\n                 jud_id = (existing or _find_judgment(comparison_id, model) or {}).get(\n                     \"judgment_id\"\n                 )\n+                win_label = (\n+                    f\"{sides[w]['commit_id'][:16]} ({sides[w]['contributor']})\"\n+                )\n+                lose_label = (\n+                    f\"{sides[l]['commit_id'][:16]} ({sides[l]['contributor']})\"\n+                )\n                 await broadcast_audit(\n                     \"vote\",\n-                    f\"{model}: {authors[w]} over {authors[l]} ({result['ratio']})\",\n+                    f\"{model}: {win_label} over {lose_label} ({result['ratio']})\",\n                     phase=\"ranking\",\n                     evidence_url=(\n                         _evidence_url(\"judgment\", jud_id) if jud_id else None\n@@ -1714,8 +1741,8 @@ async def rank_commits(commits: list[dict], *, epoch: int = -1):\n                 await broadcast_js(exec_event(Three[Selector(\"#emission-log\")][PREPEND][\n                     [\"div.log-vote\",\n                         [\"span.model\", model], \" — \",\n-                        [\"span.winner\", authors[w]], f\" beat \",\n-                        [\"span.loser\", authors[l]], f\" ({result['ratio']}) \",\n+                        [\"span.winner\", win_label], f\" beat \",\n+                        [\"span.loser\", lose_label], f\" ({result['ratio']}) \",\n                         [\"span.explanation\", result[\"explanation\"]],\n                     ]\n                 ]))\n@@ -1745,24 +1772,46 @@ async def rank_commits(commits: list[dict], *, epoch: int = -1):\n             [\"div#emission-status\", label]\n         ]))\n \n-    pairs = await pairwise_rank(len(authors), compare_fn, progress_fn)\n+    pairs = await pairwise_rank(len(ordered), compare_fn, progress_fn)\n \n     if not pairs:\n-        ranking = {authors[0]: Decimal(\"1\")} if authors else {}\n+        commit_score_list = [Decimal(\"1\")]\n     else:\n         scores = rank_centrality(pairs)\n-        ranking = {authors[i]: Decimal(str(scores[i])) for i in range(len(authors))}\n-    ranking_rows = sorted(ranking.items(), key=lambda x: x[1], reverse=True)\n+        commit_score_list = [Decimal(str(scores[i])) for i in range(len(ordered))]\n+\n+    commit_ranking = {\n+        commit_ids[i]: str(commit_score_list[i]) for i in range(len(ordered))\n+    }\n+    contributor_totals = _rollup_contributor_scores(ordered, commit_score_list)\n+    contrib_rows = sorted(\n+        contributor_totals.items(), key=lambda x: x[1], reverse=True\n+    )\n+    commit_rows = sorted(\n+        ((commit_ids[i], commit_score_list[i], ordered[i][\"contributor\"])\n+         for i in range(len(ordered))),\n+        key=lambda x: x[1],\n+        reverse=True,\n+    )\n     completed = await append_evidence(epoch, \"ranking.completed\", {\n         \"ranking_run_id\": ranking_run_id,\n         \"models\": models,\n-        \"ranking\": {a: str(s) for a, s in ranking_rows},\n+        \"commit_ranking\": commit_ranking,\n+        \"contributor_ranking\": {a: str(s) for a, s in contrib_rows},\n+        \"ranking\": {a: str(s) for a, s in contrib_rows},\n         \"judgment_ids\": judgment_ids,\n-        \"summary\": \"Ranking: \" + \", \".join(f\"{a} {s:.4f}\" for a, s in ranking_rows),\n+        \"summary\": (\n+            \"Commit ranking: \"\n+            + \", \".join(\n+                f\"{cid[:16]}={float(s):.4f}\" for cid, s, _ in commit_rows[:12]\n+            )\n+            + (\"…\" if len(commit_rows) > 12 else \"\")\n+        ),\n     })\n     await broadcast_audit(\n         \"ranking\",\n-        \"Ranking: \" + \", \".join(f\"{a} {s:.4f}\" for a, s in ranking_rows),\n+        \"Contributor rollup: \"\n+        + \", \".join(f\"{a} {s:.4f}\" for a, s in contrib_rows),\n         progress=90,\n         phase=\"finalizing\",\n         evidence_event_id=completed.event_id,\n@@ -1772,10 +1821,10 @@ async def rank_commits(commits: list[dict], *, epoch: int = -1):\n     await broadcast_js(exec_event(Three[Selector(\"#emission-log\")][PREPEND][\n         [\"div.log-ranking\",\n             [\"b\", \"Ranking: \"],\n-            *[[\"span.rank-entry\", f\"{a} {float(s):.3f}  \"] for a, s in ranking_rows],\n+            *[[\"span.rank-entry\", f\"{a} {float(s):.3f}  \"] for a, s in contrib_rows],\n         ]\n     ]))\n-    return ranking, models, {\n+    return contributor_totals, models, {\n         \"ranking_run_id\": ranking_run_id,\n         \"ranking_event_id\": completed.event_id,\n     }\n@@ -2352,6 +2401,23 @@ async def epoch_detail(epoch: int):\n \n     ranking_nodes: list = []\n     if ranking_completed:\n+        commit_ranking = ranking_completed.payload.get(\"commit_ranking\") or {}\n+        contrib_ranking = (\n+            ranking_completed.payload.get(\"contributor_ranking\")\n+            or ranking_completed.payload.get(\"ranking\")\n+            or {}\n+        )\n+        commit_rank_links = [\n+            (\n+                f\"{cid[:20]} = {score}\",\n+                _evidence_path(\"commit\", cid),\n+            )\n+            for cid, score in sorted(\n+                commit_ranking.items(),\n+                key=lambda kv: Decimal(str(kv[1])),\n+                reverse=True,\n+            )\n+        ]\n         ranking_nodes = [\n             [\"p\", ranking_completed.payload.get(\"summary\") or \"ranking completed\"],\n             _dl_rows([\n@@ -2362,10 +2428,10 @@ async def epoch_detail(epoch: int):\n                     ranking_completed.event_id,\n                 )),\n             ]),\n-            [\"pre.blob\", json.dumps(\n-                ranking_completed.payload.get(\"ranking\") or {},\n-                indent=2, sort_keys=True,\n-            )],\n+            [\"h3\", \"commit ranking\"],\n+            _link_list(commit_rank_links) if commit_rank_links else [\"p.note\", \"(none)\"],\n+            [\"h3\", \"contributor rollup\"],\n+            [\"pre.blob\", json.dumps(contrib_ranking, indent=2, sort_keys=True)],\n         ]\n     elif ranking_started:\n         ranking_nodes = [[\"p.note\", f\"Ranking started: {ranking_started.event_id}\"]]\n@@ -2386,14 +2452,9 @@ async def epoch_detail(epoch: int):\n     else:\n         emission_node = [\"p.note\", \"No emission for this epoch.\"]\n \n-    contributors = {\n-        e.payload.get(\"contributor\")\n-        for e in commit_evs\n-        if e.payload.get(\"contributor\")\n-    }\n     no_comparisons_note = \"No comparisons.\"\n-    if len(contributors) <= 1:\n-        no_comparisons_note += \" Single-contributor — no LLM judgments.\"\n+    if len(commit_evs) <= 1:\n+        no_comparisons_note += \" Fewer than two eligible commits — nothing to pairwise-rank.\"\n \n     body = [\n         _evidence_nav(),\n@@ -2511,9 +2572,14 @@ async def comparison_detail(comparison_id: str):\n                     j.payload.get(\"summary\") or jid,\n                     _evidence_path(\"judgment\", jid),\n                 ))\n-    commit_links = []\n-    for cid in (side_a.get(\"commit_ids\") or []) + (side_b.get(\"commit_ids\") or []):\n-        commit_links.append((cid, _evidence_path(\"commit\", cid)))\n+    commit_ids = []\n+    for side in (side_a, side_b):\n+        cid = side.get(\"commit_id\")\n+        if cid:\n+            commit_ids.append(cid)\n+        else:\n+            commit_ids.extend(side.get(\"commit_ids\") or [])\n+    commit_links = [(cid, _evidence_path(\"commit\", cid)) for cid in commit_ids]\n     return _evidence_page(f\"comparison {comparison_id[:24]}\", [\n         _evidence_nav(_a(_evidence_path(\"epoch\", str(ev.epoch)), f\"epoch {ev.epoch}\")),\n         [\"div.eyebrow\", \"comparison\"],\n@@ -2522,8 +2588,8 @@ async def comparison_detail(comparison_id: str):\n             (\"summary\", p.get(\"summary\")),\n             (\"ranking_run_id\", p.get(\"ranking_run_id\")),\n             (\"evidence_event\", _a(_evidence_path(\"event\", ev.event_id), ev.event_id)),\n-            (\"side_a\", side_a.get(\"contributor\")),\n-            (\"side_b\", side_b.get(\"contributor\")),\n+            (\"side_a\", f\"{side_a.get('commit_id', '?')} ({side_a.get('contributor', '?')})\"),\n+            (\"side_b\", f\"{side_b.get('commit_id', '?')} ({side_b.get('contributor', '?')})\"),\n         ]),\n         [\"p\", _a(f\"/comparisons/{comparison_id}/prompt\", \"download prompt\")],\n         [\"h2\", \"commits\"],\n@@ -2534,12 +2600,12 @@ async def comparison_detail(comparison_id: str):\n         _link_list(judgment_links),\n         [\"h2\", \"prompt\"],\n         _pre_blob(_blob_text(p.get(\"prompt\"))),\n-        [\"h2\", f\"side A — {side_a.get('contributor', '?')}\"],\n+        [\"h2\", f\"side A — {side_a.get('commit_id', side_a.get('contributor', '?'))}\"],\n         [\"h3\", \"message\"],\n         _pre_blob(_blob_text(side_a.get(\"message\"))),\n         [\"h3\", \"diff\"],\n         _pre_blob(_blob_text(side_a.get(\"diff\"))),\n-        [\"h2\", f\"side B — {side_b.get('contributor', '?')}\"],\n+        [\"h2\", f\"side B — {side_b.get('commit_id', side_b.get('contributor', '?'))}\"],\n         [\"h3\", \"message\"],\n         _pre_blob(_blob_text(side_b.get(\"message\"))),\n         [\"h3\", \"diff\"],\n@@ -3156,9 +3222,9 @@ async def watch():\n                 ],\n                 [\"p.note\",\n                     (\n-                        \"Pairwise council voting is ready.\"\n+                        \"Pairwise council ranks every eligible commit.\"\n                         if key_ok else\n-                        \"Single-contributor epochs can finalize, but contested rankings require OPENROUTER_API_KEY.\"\n+                        \"Epochs with two or more eligible commits require OPENROUTER_API_KEY.\"\n                     )\n                 ],\n             ],\ndiff --git a/tests/integration.clj b/tests/integration.clj\nindex 11140e95a8cd86768a661a30e6b29a3ddfc95fcc..39b2476cb5ddf80733769f61343c81ba7287a974 100644\n--- a/tests/integration.clj\n+++ b/tests/integration.clj\n@@ -507,7 +507,7 @@\n \n         (bind or-state @(:state or-mock))\n         (assert! (pos? (:model-requests or-state))    \"OpenRouter /models was called\")\n-        (assert! (>= (:compare-requests or-state) 3) \"at least 3 pairwise LLM calls (2 authors × 3 models)\")\n+        (assert! (>= (:compare-requests or-state) 3) \"at least 3 pairwise LLM calls (2 commits × 3 models)\")\n \n         (bind ledger2 (get-json base-url \"/api/ledger\"))\n         (assert! (>= (count ledger2) 3)\n@@ -543,6 +543,8 @@\n                  \"epoch page links comparisons\")\n         (assert! (str/includes? epoch-html \"/judgments/\")\n                  \"epoch page links judgments\")\n+        (assert! (str/includes? epoch-html \"commit ranking\")\n+                 \"epoch page shows per-commit ranking\")\n         (bind commit-href\n           (second (re-find #\"/commits/(c_[a-f0-9]+)\" epoch-html)))\n         (assert! (some? commit-href) \"found a commit id on epoch page\")\ndiff --git a/tests/test_git_discovery.py b/tests/test_git_discovery.py\nindex 00d26bfa85737b178c8822804e278211974e0efe..0f002a58bd9a122d41c5a85e32c16ed2b4404d2f 100644\n--- a/tests/test_git_discovery.py\n+++ b/tests/test_git_discovery.py\n@@ -434,7 +434,7 @@ def test_emission_distribution_sums_exactly_to_total(\n     assert entry.discovery_snapshot_id == \"ranked-snapshot\"\n \n \n-def test_single_contributor_ranking_is_total_and_uses_no_pairwise_votes(\n+def test_single_commit_ranking_skips_pairwise(\n     discovery_config, monkeypatch,\n ):\n     monkeypatch.setattr(c, \"store\", c.JsonlStore(discovery_config / \"ledger.jsonl\"))\n@@ -454,6 +454,65 @@ def test_single_contributor_ranking_is_total_and_uses_no_pairwise_votes(\n     assert info[\"ranking_event_id\"]\n \n \n+def test_same_contributor_multiple_commits_runs_pairwise(\n+    discovery_config, monkeypatch,\n+):\n+    monkeypatch.setattr(c, \"store\", c.JsonlStore(discovery_config / \"ledger.jsonl\"))\n+    calls = {\"n\": 0}\n+\n+    async def models(n=3):\n+        return [\"m1\", \"m2\", \"m3\"]\n+\n+    async def compare(model_id, side_a, side_b, **kwargs):\n+        calls[\"n\"] += 1\n+        assert \"commit_id\" in side_a and \"commit_id\" in side_b\n+        if kwargs.get(\"persist\"):\n+            attempt_id = c.attempt_id_for(kwargs[\"comparison_id\"], model_id, 1)\n+            await c.append_evidence(kwargs[\"epoch\"], \"llm.judgment\", {\n+                \"judgment_id\": c.judgment_id_for({\n+                    \"attempt_id\": attempt_id,\n+                    \"comparison_id\": kwargs[\"comparison_id\"],\n+                    \"model_id\": model_id,\n+                    \"winner\": \"A\",\n+                    \"ratio\": \"2:1\",\n+                    \"explanation\": \"ok\",\n+                }),\n+                \"attempt_id\": attempt_id,\n+                \"comparison_id\": kwargs[\"comparison_id\"],\n+                \"model_id\": model_id,\n+                \"winner\": \"A\",\n+                \"ratio\": \"2:1\",\n+                \"explanation\": \"ok\",\n+                \"summary\": \"ok\",\n+            })\n+        return {\"winner\": \"A\", \"ratio\": \"2:1\", \"explanation\": \"ok\"}\n+\n+    monkeypatch.setattr(c, \"fetch_top_models\", models)\n+    monkeypatch.setattr(c, \"llm_pairwise_compare\", compare)\n+    monkeypatch.setattr(c, \"OPENROUTER_API_KEY\", \"test-key\")\n+    commits = [\n+        {\n+            \"contributor\": \"alice\",\n+            \"oid\": \"sha1:\" + char * 40,\n+            \"message\": f\"msg-{char}\",\n+            \"patch\": f\"patch-{char}\",\n+        }\n+        for char in (\"a\", \"b\", \"c\")\n+    ]\n+    ranking, used, info = asyncio.run(c.rank_commits(commits, epoch=0))\n+    assert set(ranking) == {\"alice\"}\n+    assert ranking[\"alice\"] > 0\n+    assert used == [\"m1\", \"m2\", \"m3\"]\n+    assert calls[\"n\"] >= 3\n+    completed = next(\n+        e for e in c.store.read()\n+        if isinstance(e, c.Evidence) and e.kind == \"ranking.completed\"\n+    )\n+    assert len(completed.payload[\"commit_ranking\"]) == 3\n+    assert \"alice\" in completed.payload[\"contributor_ranking\"]\n+    assert info[\"ranking_event_id\"]\n+\n+\n def test_any_council_failure_aborts_ranking(discovery_config, monkeypatch):\n     monkeypatch.setattr(c, \"store\", c.JsonlStore(discovery_config / \"ledger.jsonl\"))\n \n@@ -479,16 +538,16 @@ def test_any_council_failure_aborts_ranking(discovery_config, monkeypatch):\n         asyncio.run(c.rank_commits(commits, epoch=0))\n \n \n-def test_contested_ranking_requires_openrouter_key(monkeypatch):\n+def test_multi_commit_ranking_requires_openrouter_key(monkeypatch):\n     monkeypatch.setattr(c, \"OPENROUTER_API_KEY\", \"\")\n     commits = [\n         {\n-            \"contributor\": contributor,\n+            \"contributor\": \"alice\",\n             \"oid\": \"sha1:\" + char * 40,\n-            \"message\": contributor,\n+            \"message\": char,\n             \"patch\": \"patch\",\n         }\n-        for contributor, char in [(\"alice\", \"a\"), (\"bob\", \"b\")]\n+        for char in (\"a\", \"b\")\n     ]\n     with pytest.raises(RuntimeError, match=\"OPENROUTER_API_KEY\"):\n         asyncio.run(c.rank_commits(commits))\n","role":"user"}],"model":"openai/gpt-chat-latest"}