This isn't new - people change their name for all sorts of reasons, the most common being marriage I expect.
Consistency in academic circles, and specifically publications is really important, hence ORCID iDs and other tracking systems to make sure that someone's output and activity is all linked, irrespective of changes of name, institution and so on.
But apparently one academic publisher, the Association of Computing Machinery now only keep records of a trans authors works under their previous name in some sort of locked vault, in case they are required under a legal subpoena? Effectively erasing the work's authors previous name.
This feels like a very retrograde and confusing step.
Article here: www.nature.com/articles/d41586-020-02145-3
Extracts:
The board’s initial position, that name changes could be applied only to metadata (and not PDF files), was a non-starter. This would juxtapose a trans author’s correct name with their previous name, cause confusion about how to cite the work and disclose to any casual reader that the author is transgender. If my deadname remains associated with my scholarship, readers of my work could either misname and misgender me, or apply anti-trans bias to my work before even encountering my ideas.
The plan is for the ACM to update all publicly accessible digital materials related to an author whose name has been changed. The previous version would be retained in a separate repository, accessible only when a subpoena is served. The draft policy was circulated, vetted by lawyers and voted through late last year. It’s imperfect: it doesn’t update citations by other scholars and, of course, print documents cannot be changed.