Existing Wording
The composition section defines a section (or sub-section) of the document, but unlike in CDA, the section content is actually a reference to another resource that actually holds both the narrative and data content for the section. This design means that the narrative and data can be reused in many other ways. Note that the most common resource to use for the content of a section is the List resource.

Proposed Wording
The composition section defines a section (or sub-section) of the document. Like CDA, the section contains the human readable narrative (text). It will also contain a set of references to the underlying resources that were used to construct the text. This design supports the reuse of resources while enabling the construction of authenticated text within documents.

A composition can be authenticated as needed and when dealing with a narrative document, the composition as a whole needs to be reviewed.

Comments
A clinical document needs to exhibit a high degree of consistency with the view of the information that the legal authenticator reviewed when they signed the document. In CDA the section/text element was identified as the location for the authenticated narrative content. In FHIR narrative can be found in any domain resource, inclusive of the List resource. There is no clear statement as to where the authenticated narrative content can be found.

Grahame's Comments
I think the statement is pretty clear - the target resource holds the narrative. Yes, this is more complex than CDA, but it allows for more reuse. I do not think that duplicating the narrative will reduce complexity

Disposition Comment
FHIR defines the specific algorithm for rendering a document, including what narratives are rendered in what order and requires that this view be what is displayed to authenticators