Single-file exits, single points of pause
Most L2s run one sequencer — a single operator ordering transactions — and young L1s halt under bugs or consensus faults, some famously and repeatedly. When it happens, the market does not pause with the chain: prices on centralized venues and other chains keep moving while every on-chain position sits frozen mid-trade, mid-liquidation, or mid-exit.
What resumes is not what stopped
Restart is a repricing event: oracles refresh to wherever the world moved, marks jump discontinuously, and queued liquidations execute into the first blocks. Positions that were healthy at the halt can be underwater at the resume with no tradable moment between — gap risk, manufactured by infrastructure.
The structural residue
Chains carry their outage history as a discount: reliability is priced like a fundamental, which is why halt-prone networks trade at persistent skepticism and why Solana’s maturation out of its outage era was a repricing story in itself. Escape hatches — forced-inclusion queues, decentralized sequencer roadmaps — are the category’s answer, mostly still promissory.
Reading it
Size on-chain leverage to survive a market that moves while you cannot; know the chain’s halt history and the venue’s oracle-restart behavior. Downtime is rare exactly until it is the only thing that matters.