I'd add that it's obviously the client software's job to decide what timeouts (and what behind-the-scenes retries and fallbacks) are appropriate, and when a tim

I'd add that it's obviously the client software's job to decide what timeouts (and what behind-the-scenes retries and fallbacks) are appropriate, and when a timely response has arrived or not. The end user doesn't need to know whether, say, waiting 15 seconds is sufficiently generous: there are software engineers for that.