Erik says that all of the strings can be decoded with the same method. Since the input string can result in multiple different valid encoded forms that implies

Erik says that all of the strings can be decoded with the same method. Since the input string can result in multiple different valid encoded forms that implies to me that the method for decoding these strings is also encoded in the string (it seems that otherwise we would have to guess magic inputs). However, assuming that approach, it is interesting to me that each resulting string ended up the same length. Trying to tease out some encoding code from the outputs now...