Replace mega-channel size rule with explicit skip list

The >500-video threshold was too aggressive — it skipped tiebreaking
for legitimate large channels (1a-auto, forgotten-weapons, etc.) where
channel context correctly resolves ties. Replace with an explicit
MEGA_CHANNEL_SKIP_LIST in recon_domains.py. Only known non-topical
catch-alls (currently just "Transcript") skip the tiebreaker.

Removed _channel_video_count() helper and MEGA_CHANNEL_THRESHOLD
constant (no longer used).

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

docs/domain-assignment.md

@@ -30,15 +30,20 @@ For each `tied_pass_1` document:
 
 1. Identify the tied domains from Qdrant
 2. Look up the document's channel (`catalogue.category`)
-3. **Mega-channel rule:** If channel has >500 videos, skip tiebreaking → `tied_manual`
+3. **Skip-list check:** If channel is in `MEGA_CHANNEL_SKIP_LIST` (known non-topical catch-alls), skip tiebreaking → `tied_manual`
 4. Query Qdrant for domain counts across all other videos in the same channel (single batch query with `MatchAny` filter)
 5. Among the tied domains only, pick the one with the highest channel-wide concept count
 6. If resolved → `tied_pass_2`
 7. If still tied → `tied_manual` (alphabetical fallback assigned, flagged for review)
 
-### Mega-Channel Rule
+### Channel Skip List
 
-Channels with >500 videos (like the "Transcript" catch-all with ~9,200 videos) are not topically coherent. Scanning their concepts produces meaningless aggregate data. These go straight to `tied_manual` for dashboard review.
+Certain channels are known non-topical catch-alls where channel-wide concept aggregation produces meaningless noise. These are listed explicitly in `MEGA_CHANNEL_SKIP_LIST` (defined in `lib/recon_domains.py`) and skip tiebreaking entirely — their tied items go straight to `tied_manual` for dashboard review.
+
+Current skip list:
+- `Transcript` — Legacy catch-all (~9,200 videos), no topical coherence
+
+This is intentionally an explicit list, not a size threshold. Legitimate large channels (e.g., 1a-auto, forgotten-weapons) run the tiebreaker normally because their content is topically coherent. Adding a channel to the skip list requires a code change and a documented reason.
 
 ## Status Values
 

lib/domain_assigner.py

@@ -25,13 +25,11 @@ from collections import Counter
 from qdrant_client import QdrantClient
 from qdrant_client.models import Filter, FieldCondition, MatchValue, MatchAny
 
-from .recon_domains import VALID_DOMAINS, DOMAIN_CATEGORY_MAP
+from .recon_domains import VALID_DOMAINS, DOMAIN_CATEGORY_MAP, MEGA_CHANNEL_SKIP_LIST
 from .utils import setup_logging
 
 logger = setup_logging('recon.domain_assigner')
 
-# Channels with more than this many videos skip channel tiebreaking entirely
-MEGA_CHANNEL_THRESHOLD = 500
 
 
 def _get_qdrant_client(config):
@@ -222,14 +220,6 @@ def _channel_video_hashes(db, channel_name, exclude_hash=None):
     return hashes
 
 
-def _channel_video_count(db, channel_name):
-    """Count total videos in a channel."""
-    conn = db._get_conn()
-    row = conn.execute(
-        "SELECT COUNT(*) as cnt FROM catalogue WHERE category = ? AND source = 'stream.echo6.co'",
-        (channel_name,)
-    ).fetchone()
-    return row['cnt'] if row else 0
 
 
 def run_tiebreaker_pass(db, config, qdrant=None):
@@ -241,8 +231,8 @@ def run_tiebreaker_pass(db, config, qdrant=None):
     other videos in the same channel and picks the tied domain with the
     highest channel-wide count.
 
-    Mega-channels (>500 videos) skip tiebreaking and go straight to
+    Channels in MEGA_CHANNEL_SKIP_LIST (known non-topical catch-alls) skip
-    'tied_manual' for dashboard review.
+    tiebreaking and go straight to 'tied_manual' for dashboard review.
 
     Args:
         db: StatusDB instance
@@ -263,9 +253,6 @@ def run_tiebreaker_pass(db, config, qdrant=None):
     stats = {'resolved': 0, 'manual': 0, 'skipped': 0, 'errors': 0, 'total': len(tied_items)}
     logger.info(f"Tiebreaker pass: {len(tied_items)} items to resolve")
 
-    # Cache channel sizes to avoid repeated queries
-    channel_size_cache = {}
-
     for item in tied_items:
         file_hash = item['hash']
         channel = item.get('category', '')
@@ -283,16 +270,12 @@ def run_tiebreaker_pass(db, config, qdrant=None):
                 stats['resolved'] += 1
                 continue
 
-            # Check mega-channel rule
+            # Skip-list check: known non-topical catch-all channels
-            if channel not in channel_size_cache:
+            if channel in MEGA_CHANNEL_SKIP_LIST:
-                channel_size_cache[channel] = _channel_video_count(db, channel)
-
-            if channel_size_cache[channel] > MEGA_CHANNEL_THRESHOLD:
                 fallback = sorted(tied_domains)[0]
                 db.set_domain_assignment(file_hash, fallback, 'tied_manual')
                 stats['manual'] += 1
-                logger.debug(f"  {file_hash[:12]}: mega-channel '{channel}' "
+                logger.debug(f"  {file_hash[:12]}: skip-list channel '{channel}' → tied_manual")
-                             f"({channel_size_cache[channel]} videos), → tied_manual")
                 continue
 
             # Channel tiebreaker: count domains across all other videos in channel

lib/recon_domains.py

@@ -32,3 +32,15 @@ DOMAIN_CATEGORY_MAP = {
 VALID_DOMAINS = set(DOMAIN_CATEGORY_MAP.keys())
 
 CATEGORY_DOMAIN_MAP = {v: k for k, v in DOMAIN_CATEGORY_MAP.items()}
+
+# Channels whose tiebreaker is skipped because their content is non-topical
+# (catch-alls, miscellany dumps, etc.). Items in these channels with tied
+# domain counts go straight to tied_manual without channel-context tiebreaker.
+#
+# This is intentionally a hardcoded explicit list, not a size threshold.
+# Adding a channel here requires an explicit decision — only add channels
+# that are genuinely non-topical catch-alls where channel-wide concept
+# aggregation would produce meaningless noise.
+MEGA_CHANNEL_SKIP_LIST = {
+    'Transcript',  # Legacy catch-all, ~9,200 videos, no topical coherence
+}