I believe the problem this policy aims to solve is one of HN submitters inserting their own bias into self-written titles.

Whereas if the original title is used, then any bias in the title is confined to that of the article author / publisher.