I am glad that someone else thinks that the generation of all these policies inevitably leads to problems.
As someone who spends a great deal of time delving into ancient charters and other legal documents in England, I am amazed that a fixed set of policies for the mediaeval period can be properly established and yet still be functional.
I thought that the overriding principle was to try and establish the earliest record of a name for an individual and place that as their LNAB. It matters not if their immediate family have different names, as long as you can prove the link.
I can show evidence of one person being named differently on different lines of a fourteenth century charter. I have evidence too of charters with confirmations of earlier deeds that clearly show that in the contemporary society in England, a male line could have as many as six surnames in use simultaneously within three generations. The document might say, for example: "Robert de Pul grants six acres of grassland in the manor of Haxted to the Abbot of Cherwell, in confirmation of a charter made by his father William the Armiger and of his grandfather Edward de Stapelton. it merely shows that until the fifteenth and sixteenth century it was still common for a person, usually a man, to hold lands and to take his name from them - if they were not the same lands of his forefathers, then the surname was different.