Slow network or network died all of a sudden - you wouldn't want your experiments to fail just because there is a network issue. Once network is restored - everything is sent as it should be to the backend. This is the resilience I was talking about

Regarding mistake in URL - I guess it is your responsibility to keep track of not using wrong URLs. As I said before - How would you suggest the SDK know what is the 'right' URL and what is the 'wrong'...