PAF doesn't create duplicate records on its own. There seems to be no limit to the amount of information a single record can hold. I paste obits, bios and every census or newspaper article I can find into the notes for people in my database. I have heard that in PAF5 there is no limit.
What happens though, is that as we do our research we will find the same individual in multiple contexts and therefore create multiple records for them. For example, supposing I find that Herbert and Susan Block have a son John, christened on 16 April 1832 in Rogate, Sussex, England. So I create a record for John as a child of Herbert and Susan. Later I find that Amelia Prudence Jones has married a John Block and so I add him again. Later, I discover that the Prudence Block who married my great grandfather's cousin Alexander Scott has parents named John and Amelia, and so I add him again.
It is only later, when I am scrutenizing my data that I ask myself, "Is this the same John Block in all three cases, or in two of the cases, or are they all different people?" As the evidence builds up I am faced with the decision of merging the records for John, and Amelia, and perhaps other members of their family.
As usual, I start with a new backup of my data using a new name. Then I carefully plan the merges, and proceed, with the assurance that if I mess it up I can always restore my backup.