Running nodes teaches you this the hard way... stability is a story you tell yourself after the fact. In practice, capacity drifts, operators disappear.. and nobody files a ticket on the way out.
Walrus does not fight that at all. #Walrus prices and coordinates around absence. Availability is not earned by being perfect... it is maintained by thresholds that assume someone won't show up. That changes the failure mode. Things donnot snap. They thin. And thinning is survivable.
That is the difference between storage that demos well and storage you’re still paying for a year later.

