Cross-Platform Retrieval Mechanism Comparison: Perplexity, ChatGPT and Gemini on Identical Queries (GEO Lab Experiment E042)

E042 is an exploratory cross-platform observation (not a pre-registered controlled experiment) mapping how Perplexity, ChatGPT and Gemini retrieve in response to the same 9 queries (3 length tiers × 3 GEO Lab concept pages), captured same-day via browser network traces — 27 data points. Citation outcomes: Perplexity 67% (6/9), ChatGPT 56% (5/9), Gemini 0% (0/9), documenting three structurally different retrieval architectures. Key findings: ChatGPT operates a binary pre-retrieval gate bypassed by long-form (≥10-word) + named-platform + causal framing (first confirmed ChatGPT citation of thegeolab.net); Gemini Flash mode grounds via encoded batchexecute calls and cited nothing across all 9 queries; ThatWare and Hashmeta are absent from ChatGPT’s retrieval set, indicating competitive displacement is Perplexity-specific and content/rank-driven, not infrastructure-driven (robots.txt audit: no competitor blocks any AI crawler). Includes a cross-platform reconstruction coherence audit. Data files: cross-platform mechanism map (27 rows) and robots.txt audit (5 domains).