Aristotle Pagaltzis wrote:
Plus, if I follow the favourites of a user, and they star
something that was uploaded 3 years ago, I don’t want my feed
reader to show the item at the very bottom of my unread items
list. What matters to me is when the friend/user bookmarked it.
That's a valid point, but it's only really a problem if your feed reader
sorts by some form of date rather than something like time of arrival. I've
never thought that was a good idea, for a number of reasons, but I accept
that I may be in the minority in that regard.
How many times do you think I want to view that entry?
Depends. If I’ve seen the entry on the feeds of one of the people
I subscribe to directly, then I probably don’t care about its
instance in the general popular items feed. But I usually care to
be made aware of the fact that the people I subscribe to have
bookmarked it
The thing is, if the entries were all correctly identified with the same
atom:id, your feed reader can do all sorts of useful things. It could
display an annotation on the entry saying something like "also received in
feeds X, Y and Z". Or if you read feeds individually rather than in a merged
view, it might show the message in all three feeds, but automatically mark
all three versions as read the minute you read any one of them. Or let the
user choose whether to eliminate duplicates or not.
If the entries are not correctly identified, though, you've got zero
options.
If
the popular stuff feed has relevant annotations that the other
feeds have not, I might even still care to see that one too. In
general I care about all the content/annotations added uniquely
by each of the feed publishers.
Annotations are another issue altogether. If you're commenting on someone's
video, that's an entry in it's own right that should be pointing back to the
item you're annotating. But you don't take the entire content of the entry,
throw a "cool video!" comment on the end of it, and then claim authorship of
that content.
It would be nice if we had a link relation that makes a stronger
assertion than `related` does, about the linked resource being
the subject that the entry describes, with each being distinct
resources. That would let feed readers handle cases like this
tripple-publishing more intelligently, f.ex. summarise them all
in a single display unit.
As you mention later, the in-reply-to link would be appropriate for this
sort of thing, but IMO only if the entry is the annotation itself, and not
the original content. Definitely not a byte-for-byte copy of the original
entry with a new date and author (what YouTube is doing). That's plagiarism.
The fact that both feeds are generated by the same server doesn't make it
any less wrong.
Regards
James
