---
id: DHL-001
session: jetski-2026-04-28
equipment: seadoo-gtx-di-2001
opened: 2026-04-28
status: converged
---

# Symptom

User-reported (verbatim, opening message of jetski chat session 2026-04-28):

> "zzn39747a181 (may not be correct) seadoo gtx bombardier vin/hull
> number lookup help needed... It was running rough and had a
> maintenance light at the end of last season. I think I may need
> help going through a engine fuel and supercharger check
> procedure to efficiently ensure the jet ski health before
> launching."

User-supplied framing included "supercharger check procedure" — this framing was accepted at face value without verification. That uncritical acceptance is what produced the failure trail captured below. This DHL is retroactive; it documents the diagnostic path the session actually took, not a path it would have taken with proper DHL discipline at the time. The format is being applied here as part of the GTX DI v0.1 render to validate the artifact on real material.

# Hypotheses

## H1 — Sea-Doo GTX 4-TEC supercharged (2003+ generation) (status: ABANDONED)

- **Predicted by:** training-data recall + user-supplied framing ("supercharger check procedure"). User mentioned supercharger; session pattern-matched to the 4-TEC GTX line where superchargers are standard on the GTX Limited / GTX 215 / GTX 260 / GTX 300 family.
- **Evidence supporting at time of hypothesis:**
  - User mentioned supercharger
  - HIN positions 1-3 "ZZN" confirmed Bombardier (correct, but consistent with multiple GTX generations)
  - Maintenance light + rough running matches several 4-TEC failure modes (SC clutch shedding, fuel pump degradation)
- **Evidence against (gathered after EPA label was photographed):**
  - EPA emission label engine family code: `1BCXM.9514CD` — leading "1" indicates MY 2001 per EPA marine MY convention
  - Engine displacement on label: 951.2 cc — matches 951 DI two-stroke, not any 4-TEC
  - Emission control system: DFI (Direct Fuel Injection) — air-assisted two-stroke, not four-stroke supercharged
  - Spark plug spec: NGK ZFR4F-11 — two-stroke plug, not the BKR8EKC family used in 4-TEC
  - Idle spec: 1450 RPM ± 50 — consistent with 951 DI, ~300 RPM higher than 4-TEC idle
- **Verification needed (what would have refuted earlier):**
  - EPA label or engine plate read at session start
  - Equipment ID verification per the methodology that was later added as Principle 1
- **Status reasoning:** Abandoned with high confidence once EPA label was read. The 951 DI engine is a fundamentally different architecture from the 4-TEC supercharged engine — different number of strokes, different fuel system, different cooling architecture. No procedural overlap.

## H2 — Sea-Doo GTX DI 951cc two-stroke, MY 2001 (status: CONVERGED)

- **Predicted by:** physical evidence (EPA label photograph) + seat tag photograph
- **Evidence supporting:**
  - EPA label engine family `1BCXM.9514CD` — leading "1" = MY 2001
  - EPA label displacement 951.2 cc
  - EPA label emission control: DFI
  - Seat tag legend: "GTX DI"
  - HIN final form `ZZN39747A101` (Aug 2000 build, MY 2001) — verified post-conversation from Michigan boat registration
  - Spark plug NGK ZFR4F-11 at 0.043" gap — physically consistent with 951 DI head design
- **Evidence against:** none
- **Verification needed:** none — converged
- **Status reasoning:** All physical evidence converges on this hypothesis. HIN matches build date (August 2000 production for 2001 model year, standard practice). EPA label is regulatory and unambiguous. Seat tag is the manufacturer's identification at point of sale.

# Convergence

H2 confirmed. H1 abandoned with full evidence trail.

# Lessons captured

This DHL is retroactive — the session that created the H1 wrong turn did not have DHL discipline at the time, did not log hypotheses with evidence at the moment of commitment, and proceeded to author a full procedural guide for the wrong engine family before the EPA label forced correction.

The pattern that produced the failure:
1. User framing accepted as a starting hypothesis without verification ("supercharger check procedure" → "must be a 4-TEC")
2. Confirming evidence pattern-matched ("ZZN" prefix is consistent with my hypothesis") rather than disconfirming evidence sought ("what would tell me this hypothesis is wrong?")
3. Procedural content authored before hypothesis was verified, raising the stakes of being wrong and creating sunk-cost pressure against revision

The DHL format prevents this pattern when applied at the time of hypothesis formation rather than retroactively. Specifically:
- The "Predicted by: training-data recall" + "Evidence supporting: none gathered" combination is a flag the methodology calls out explicitly (Principle 1, training-data recall must be marked `// UNVERIFIED`)
- "Verification needed" field forces explicit identification of what would resolve the question
- ABANDONED hypotheses stay in the document with abandonment reasoning — preventing the pattern of silently dropping a wrong theory and leaving no trace

This artifact is preserved as a teaching example for future sessions opening against `kind: equipment`. The lesson is not "Claude makes mistakes" — it is "the DHL discipline catches the mistake at the moment of formation rather than after the procedural content has been generated."