In the linked article, Hejlsberg considers adding a new checked exception a breaking change (true), but adding a new thrown exception to not be, because "in a lot of cases, people don't care." I think this is obviously open to debate.

You're conflating "incorrect" with "mistake," no one is saying the C# team forgot to add checked exceptions.

I have a lot of respect for Anders Hejlsberg. But that doesn't mean he is never wrong. Hejlsberg doesn't think anyone would want to recover from exceptions. "There's a bottom level exception handler around their message loop. That handler is just going to bring up a dialog that says what went wrong and continue." Okaayyy... I think we know a bit more about exception handling than that today! Real-world applications often need more sophisticated exception handling strategies.