"J. P. Gilliver (John)" wrote

| And the date-of-taking would probably be embedded in the file anyway,
| these days (and for the last several years).

Yes. But I guess we shouldn't start up the
semantic debate over "what's metadata" again.

| I can have two (or more) copies of the same file, _with the same date
| stamps_, on the same disc, as long as they're in different directories;

With the same *creation* time? I don't see how.
As soon as you make a copy, that copy gets the
creation time for when it was created. If you look
at system files you'll see the same thing. It might
be creation time of 2013 and last modified 2010.
The creation time is when it was installed. Last
modified is when it was last changed. The latter was
stored in the Windows installer so that it could be
added during install.
Likewise if you take a file out of a ZIP. The lastMod
time is stored because it might be relevant. But creation
time is whenever that copy of the file was taken out
of the ZIP.

Otherwise, when is it created? You write an essay,
copy it elsewhere, rewrite that file, eventually have
a second file that bears no commonality at all with the
first... To say it was created when the first file was
created wouldn't make sense. You're then attaching
the creation to a theoretical file rather than to a
specific digital item. Creation time would then
always have to trace back to when a new file was made.
And if copying is not creation, what about recompiling?
Were all copies of all versions of the Windows kernel
created in 1992 or 1995 or some such just because there's
been a kernel32.dll file in existence during all that time?