(In reply to re(4): please read
by Ady TZIDON)
OK, suppose the 4 audience members chose 20, 40, 80 and 100 (with the same FCDE sequence). B would pick the first (only) big gap (between 40 and 80) count up to 20 (corresponding to FCDE) and add card 60 to the set. Now, A would still see 20, 40, 60, 80, 100. Same as before. He would think that B added card 40, (as you state in your last post) which is wrong.
Don't you see that any method you propose is doomed? B can only select up 75 million+ different sets of 5 cards (combinations of 100 cards taken 5 at a time) and give them to A. But there are 94 million+ different possibilities (permutations of 100 cards taken 4 at a time) to be communicated to A.