We posted the brainstorming notes on the wiki at http://wiki.activitystrea.ms/Cross-Posting
and really want additional input, thought, and feedback.  Hopefully
this can be one of the topics at the next Activity Streams meetup.

--David

I have a few random comments:

  * While I agree that having the site hosting the duplicate publish the
crosspost:target element is the ideal scenario, I think crosspost:source
is necessary as a bootstrapping mechanism since generally this
cross-posting is done by hitting a general "post new content" API on the
target that doesn't have any knowledge of whether it's a cross-post or
not: until these APIs are updated, the source that created the duplicate
is in a better position to document the fact that it did so.

  * The distinction between crosspost:source and crosspost:target was
intended to make it easier for an aggregator to find the "original" (and
therefore probably best) version and drop the duplicates. It's possible
that some publishers will lie about being the original in order to gain
favor in aggregators, but since this is user-hostile users would
hopefully remove feeds that lie in this way from their accounts. It
would be best if aggregators would present a summary of the duplicates
alongside the original so that publishers of duplicates feel less
incentive to lie.

  * Regarding the answer to "why not use rel=via", note that in Atom it
is not valid to combine multiple rels into a single link element as in
that example. This aspect of the HTML link element was not included in
Atom's link.

And by the way, apologies for my server being down this last week. My
host picked a bad time to force me to move to a different server so I've
been unable to get it up again while I've been travelling. However, I
hope to rectify that this week and get the specs back up again.

Thanks,
Martin