The underlying question (add death date where there is an open birth date, as in "Hartley, Walter, 1927- "?) is addressed with exquisite ambiguity in "Frequently Asked Questions on creating Personal Name Authority Records (NARs) for NACO"
in which Q no. 4 greatly discourages unnecessary changes (changes not needed for disambiguation)--for reasons which are abundantly clear in this discussion. Q no. 5, however, is a finely made waffle, which seems on the one hand to support updated LCRI 22.17 (February 1, 2006), which contains
"... the option to add death dates to personal name headings that have an open date, only. Remember that the policy to add death dates was intended to alleviate the perception of “wrong” or “misleading” headings that imply the deceased person is still living-- headings without open birth dates do not give such an impression. Catalogers must follow the existing practice for existing headings without dates: do not add dates to a heading where none existed previously (except in cases of conflict, etc. as usual). The principle of not making changes to headings when these are not necessary continues to hold true (cf. Response to question no. 12 below)
So which is correct? Add the dates according to the 2006 policy (which is apparently excepted from cases where there is no date)? Or stick to the principle of not unnecessarily changing a unique 100? This is a case where I would unambiguously stick to principle--my cataloging output is already sufficiently impeded by make-work procedures (e.g. 667 DO NOT USE). I don't think, under the circumstances, that a "perception" problem that had been elegantly addressed by the "b." tactic creates a condition of necessity.