> If they were so essential I'd expect the service to have crashed and burned long ago and never came back up. False dichotomy. There is a vast space between c

> If they were so essential I'd expect the service to have crashed and burned long ago and never came back up. False dichotomy. There is a vast space between continuing to run normally and turning into a smoking crater overnight. This includes brief outages, "secondary" functionality being broken for longer, slowdowns, inability to handle spikes or add new features, etc. All things we've actually seen with Twitter BTW. No one of them is enough to kill the site outright, but people do notice when a site is flaky. It erodes trust, and they seek alternatives. The fact that Twitter hasn't experienced a more serious collapse since Musk's takeover is a testament to the quality of the engineering that was done before. But there's little difference between a sudden death and a slower one, in the end. If you want to rail against "perf-review driven engineering" or churn for churn's sake, you'd do better to aim your invective at Meta or Google instead of kicking people who are already down.