This question came to mind when I was dealing with this question: A card from a pack of 52 cards is lost. From the remaining cards of the pack, two cards are drawn at random and are found to be clubs. Find the probability that the lost card is a club.
My query: The probability that a club is lost is $13/52$. The next event of taking two cards from this incomplete set of cards happens after one of the cards is lost. How can this affect a past event of losing a card from the pack?
On
Let us extend your example. If we lose 13 cards, and after that open 52 - 13 cards and all of them are not clubs, what is the probability of all 13 lost cards to be clubs? It is clearly $1$.
Bayes theorem is not about affecting past events. It is about recalculation of probabilities when you have some more information. Probability of drawing two clubs after the card was lost depends on what card was lost. If the lost card is a club drawing two clubs after that becomes less probable than before. So, you can readjust your estimation of probability based on this information.
On
You are confusing correlation with causation. Two variables are correlated when there is any (nontrivial) relation between them. Two variables have a causal relation when a human can control one to influence the other.
Correlation is what mathematics and science deal with. Causation is what engineers and designers deal with.
There is a correlation between the future event and the past event. But you cannot "control" the future event in a way to change the past event.
The event has not changed. Your information regarding the event has changed.
For instance, say you were interested if the Club Ace was lost.
Now if the two cards happen to have the Club Ace, would your estimate of the chances still be as before?
Basically, the probability is based on what information you have at hand.