Breadlines Research
When the Window Determines the Result
A landed-ledger methodology note on Solana’s 350ms → 300ms transition.
Finding
The measured before/after success-rate difference changed from +0.22 percentage points in a one-hour comparison to −10.65 percentage points in a buffered six-hour comparison. The study does not support a simple causal boundary claim.
The paper documents the deterministic sample contract, window and bucket results, stability criteria, a matched prior-day control, and explicit limits of landed-ledger evidence.
It does not infer sender geography, dropped traffic, end-to-end latency, identity, intent, or causality.