Reimian Empire

when the outcome of previous attempts is known in advance, the chance for a new attempt is basic, 5%.

No, you are wrong if you mean the object itself; what if it has already produced the same result a million times? If you mean the overall chance for an independent observer, then yes, but this only confirms what I said.

If two out of 99 readers died, then the chance of the hundredth dying is 5%. It does not increase, as these events are independent.

I don't remember exactly, but in any case, are you saying that if ninety-nine people flip the same (and this is important) coin and it lands on tails, will the probability for the last person still be 50%?