(In reply to re: please read
by Ady TZIDON)
Charlie's argument is irrefutable. The fact is there are 94,109,400 ordered ways (permutations) in which 4 audience members can select 4 cards. When A looks at the cards, a card has been added by B and they've been shuffled. Thus the number of distinct arrangements of cards that can be given to A is the number of combinations (not permutations) of the 5 cards selected from 100. Thus A can receive no more than 75,287,520 distinct messages when he receives the 5 cards. So it isn't possible for all 94 million+ different possibilities of the original 4 persons' selections to be conveyed with 5 shuffled cards, since there are only 75 million+ ways to do the latter.
Ady, there must be something wrong with your algorithm--I'm sure if I studied it long enough, I could find two different sets of choices from the audience which, after B follows your method to add his 5th card, results in the same set of 5 cards, leaving A no way to determine which of the two cases it must be.
The only way this trick can be pulled off 100% of the time is using an additional trick, wherein A receives more information than just the 5 shuffled cards.