[Retra]

I think the confusion here is related to the idea that you can infer something happens even if you cannot observe it happening. If you know it takes light 8K years to reach you, and you can right now see a countdown process which will go off in 4K years, it is reasonable to assume the event already occurred in the process frame of reference. Of course, the certainty of that hinges on the inevitability of the process, but that could viably be on the same scale as the certainty of a direct observation itself.

Basically, our ability to accurately predict the future is not confined to events in our own frame of reference, so the difference between what is known to have happened and what is predicted to happen need not be very large.