WFGY/ProblemMap/GlobalFixMap/RAG/hybrid_failure.md

Hybrid Retrieval Failure — Guardrails and Fix Pattern

When hybrid retrieval (BM25 + dense, HyDE + reranker, multi-vector) performs worse than a single retriever.
Instead of increasing recall, the hybrid path introduces instability, wrong ranking, or noisy snippets.

---

Open these first

• Visual map and recovery: RAG Architecture & Recovery
• End to end retrieval knobs: Retrieval Playbook
• Traceability schema: Retrieval Traceability
• Snippet contracts: Data Contracts
• Query path splits: Pattern: Query Parsing Split
• Ranking drift: Rerankers

---

Core acceptance

• Hybrid recall ≥ single retriever recall
• ΔS(question, retrieved) ≤ 0.45 for top-1 result
• λ stable across three paraphrases and two seeds
• Coverage ≥ 0.70 to the target section

---

Typical symptoms → exact fix

Symptom	Likely cause	Open this
Hybrid returns unrelated snippet	query parsing split not locked	Pattern: Query Parsing Split
Hybrid recall < single recall	wrong weighting or missing normalization	Retrieval Playbook
Dense retriever dominates BM25	metric mismatch	Embedding ≠ Semantic
Reranker undoes good hits	λ flips, entropy collapse	Rerankers, Entropy Collapse

---

Fix in 60 seconds

1. Measure baseline
   Run BM25 alone and dense alone. Log coverage and ΔS. If hybrid < baseline, do not ship.

2. Stabilize query parsing
   Split HyDE prompts, keyword queries, and dense embeddings into deterministic branches. Lock weighting ratios.

3. Reranker probe
   Compare recall before and after reranker. If entropy rises, clamp with variance control or drop reranker.

4. Enforce snippet schema
   Always require snippet_id, section_id, offsets, tokens. Hybrid paths must normalize schema fields.

---

Copy-paste probe prompt

I uploaded TXT OS and the WFGY Problem Map.

My issue:
- hybrid retrieval returns worse results than BM25 or dense alone.

Tell me:
1) which layer fails (query parsing, weighting, reranker),
2) which WFGY fix page to open,
3) minimal steps to restore ΔS ≤ 0.45 and coverage ≥ 0.70,
4) reproducible test with BM25 vs dense vs hybrid.

---

🔗 Quick-Start Downloads (60 sec)

Tool	Link	3-Step Setup
WFGY 1.0 PDF	Engine Paper	1️⃣ Download · 2️⃣ Upload to your LLM · 3️⃣ Ask “Answer using WFGY + ”
TXT OS (plain-text OS)	TXTOS.txt	1️⃣ Download · 2️⃣ Paste into any LLM chat · 3️⃣ Type “hello world” — OS boots instantly

---

🧭 Explore More

Module	Description	Link
WFGY Core	WFGY 2.0 engine is live: full symbolic reasoning architecture and math stack	View →
Problem Map 1.0	Initial 16-mode diagnostic and symbolic fix framework	View →
Problem Map 2.0	RAG-focused failure tree, modular fixes, and pipelines	View →
Semantic Clinic Index	Expanded failure catalog: prompt injection, memory bugs, logic drift	View →
Semantic Blueprint	Layer-based symbolic reasoning & semantic modulations	View →
Benchmark vs GPT-5	Stress test GPT-5 with full WFGY reasoning suite	View →
🧙‍♂️ Starter Village 🏡	New here? Lost in symbols? Click here and let the wizard guide you through	Start →

---

👑 Early Stargazers: See the Hall of Fame — ⭐ WFGY Engine 2.0 is already unlocked. ⭐