> I didn’t know it was that controversial. I feel like that article is making a bit of strawman argument from the get go by talking about 1-3% failure rate from

> I didn’t know it was that controversial. I feel like that article is making a bit of strawman argument from the get go by talking about 1-3% failure rate from the backend. If I had a system that flaky, I’d be taking a good hard look at myself before blaming anything on my optimistic UI :-) It's not a 1-3% failure rate for the backend, it's a 0.01-0.1% failure rate for each request between dozens of microservices each with their own bespoke error handling, each owned by a team with more political clout than the frontend guys/gals who keep reporting Sentry errors due to "UnknownNetworkError" or some other such nonsense, all wrapped in some spaghetti code frontend thats written like the whole thing is running locally.