The smaller a perspective is, the easier it is, to split a perspective path into more smaller parts. This in turn makes it easier to construct such paths. Additionally, this makes it easier to create a new branch of the path, with minimal changes. Furthermore, it gets easier and faster to select fitting children, when the size of the perspective is decreased, provided a fast evaluation of network traversal's direction. Therefore, perspectives that do not represent lists, should prefer a minimal number of children of the same type.
Perspectives representing a list, should prefer having only elements of the same type, in order to make it clear, that these are lists. Thereby, perspectives, where it is hard to determine the searched path, are more separated from the rest of the network. This in turn makes the management of complexity more easy. Such perspectives can be viewed as archives, general storage or namespaces. Keep in mind, that in such archives, the ordering generally cannot be insured to be correct or up to date, when done manually. So one should assume, that the order of such a list is incorrect, if it is not enforced automatically.
The longer a path is, the more detailed and specific something is described by it. In other words the, the longer the path is, the smaller the set of things is that complies with the meaning of the path. This also holds true for links, as even absolute links can be or get incorrect over time.
The interpretation of the path element names should match the actually referenced object as close as possible.
Prefer words and names where there is no singular and plural version, as this helps to minimize the size of perspectives. Also prefer using only nouns, adjectives or verbs.