The only way I see to harmonize LC's desire for generic expression records and some PCC members' desire for more specific authorities for expressions is as a hierarchy. The former represent in principle a class of expressions--all those for a given language--even when only one translation exists; and the latter represent particular translations in that language by extending the generic heading with additional specifying components. Both forms of access should be considered as authorized access points.
What seems to me unsustainable about LC-PCC PS 6.27.3 is the LC Practice instruction: "If there is a name authority record with an authorized access point for
an expression that includes an additional characteristic LC would not
have added, use the form of the access point in that authority record;
this action is consistent with the LC/PCC policy of using authorized
access points in existing name authority records." This will inevitably lead to split entries, generic for some translations and specific for others in the same language, and confusion for catalog users. "Using the authorized access points" in such cases would split entries for the "same" kind of resources, which is contrary to the fundamental purpose of authority control.
A more manageable approach would be to accept that either kind of authority is acceptable depending on the local cataloging agency's policy, but not both (except in rare cases when a second resource's description needs to specify a particular translation). Since the more specific translation headings are built by adding on to the generic heading, automated routines could change the specific heading to its generic counterpart on incoming copy to ensure consistency of access. As for libraries preferring more specific access points for translations, they have their work cut out for them; but that was always going to be the case.
As for the dates--a generic authority for a set of expressions could indicate the date of the first instance of the class of translations it represents. An authority for a specific translation could indicate a later date for its creation. What's lacking is clear consensus and policy about what the date means--which concept of expression it relates to--because of the ambiguity about whether generic and specific authorities can co-exist. The implication in the LC Practice statement that a specific AAP must trump a generic AAP throws the more viable co-existence option into uncertainty.
Stephen