[JoshTriplett]

Not using "link" to mean "symlink" is certainly a reasonable improvement. But using "link" to mean "hardlink" (outside of a context like "the link system call") creates the possibility of ambiguity.

You're suggesting "being comfortable with the correct language", but if it is not already the case that almost everyone means "hardlink" when they say "link", then it's not "the correct language". Your mental mappings should reflect actual language as actually used, so that you can understand and be understood. Those mappings should include knowledge of the ambiguity.

"hardlink" unambiguously means "hardlink" (modulo rare mistakes)

"symlink" unambiguously means "symlink" (modulo rare mistakes)

"link" means "probably hardlink, but possibly symlink being referred to sloppily, or possibly referring to the category that includes both; generally ambiguous without further contextual information".

[fsckboy]

the position you are advocating leads precisely to TFA which has it precisely wrong, that's why I went to the trouble to write a comment that makes people think. I don't need to hear a detailed counter argument for the status quo ante, just write "we'll just thoughtlessly do it the same shitty way you're arguing against"

[JoshTriplett]

No, as mentioned I am not suggesting using "link" to mean "symlink". I am suggesting being aware that not everyone (speaker or listener) will have the word "link" mapped unambiguously to "hardlink", and that if you hear "hardlink" it's likely the speaker means "hardlink", while if you hear "link" you can't be quite as sure.

In other words: I am advocating against deleting error-correcting bits from language, because language is a noisy channel and error-correcting bits help reduce errors.

[fsckboy]

thank you for describe the shitty status quo way again, you've finally convinced me my ideas are pointless