Modeling the "reshare" or "retweet" activity

[Chris Messina]

I spent some time this week discussing the "reshare" activity with
Will and documented our thinking on the wiki:

http://wiki.activitystrea.ms/Reshare

Resharing is a weird kind of activity — whether it occurs as a retweet
on Twitter, a reshare on Google Buzz, or a "repeat" on Status.net. The
point of the reshare activity is analogous to forwarding an email —
where you're passing on a message that you received and maintaining
attribution to the source or original author. In blogging, we could
use blockquotes and trackbacks... in stream-based mediums where the
intended target is the stream itself, the proper model is unclear,
especially when you factor in the ability to add your own annotation.

In cases where you reshare an activity and provide an annotation, are
you *posting* a note with an attachment?

In the case where there is no annotation, do you simply leave the note
out?

Or, would it make more sense to do something different? — modeling the
activity as "share an activity" — copying the entire original activity
as the object of your activity, and providing an activity:annotation
element to capture any additional commentary?

Our conversation lead us to favor the latter approach, making it
possible to treat an activity itself as an object (which we've
heretofore not done).

Thus, I'm interested in other people's thoughts and ideas here. I hope
I was clear in presenting the issue — and what we're trying to model.
Again, the wiki page captures more detail:

http://wiki.activitystrea.ms/Reshare

Thanks,

Chris

[Martin Atkins]

On 06/10/2010 05:59 PM, Chris Messina wrote:
>
> Our conversation lead us to favor the latter approach, making it
> possible to treat an activity itself as an object (which we've
> heretofore not done).
>

I'd prefer to avoid this as long as we can. This should be a last
resort, as it'll make our data model even more confusing than it is already.

Also, I'm not convinced that this action is really applied to an
activity anyway. On Twitter and TypePad it's the note/article that gets
re-whatevered, and I'd bet that for most other systems that is true too.

In TypePad, we model a "reblog" similarly to a comment: it's a new
article object that happens to have a pointer to another article.

Much like Trackbacks before it, this re-whatever mechanism is really
just replying in a different context. We should consider just modelling
it as posting a new whatever that's in reply to another whatever, where
"whatever" would be a note on Twitter and an article on TypePad. I can't
speak with any certainty for anyone else's data model, but that would
map exactly to TypePad's internal data model. I'm sure we're not alone
in modelling it this way.

[Rabbit]

Conceptually I think of it as writing "You'll like this. Great read! --
Aaron" on a sticky note, slapping it on the cover of a book, and
putting the book on your friends desk. The "content" is an annotation
and intended to promote, publicize or otherwise elevate the attention
of the recipient to the target object.

What about copying the content but then providing another element for
wrapping the annotation (if it exists) so that they aren't separate
but separable?

=Rabbit

> --
> You received this message because you are subscribed to the Google
> Groups "Activity Streams" group.
> To post to this group, send email to activity-
> str...@googlegroups.com.
> To unsubscribe from this group, send email to activity-strea...@googlegroups.com
> .
> For more options, visit this group at http://groups.google.com/group/activity-streams?hl=en
> .
>

[nitya]

Hi Chris, all,

I remember reading danah boyd's paper on how Twitter users were using
this feature (www.danah.org/papers/TweetTweetRetweet.pdf) and
subsequently also seeing the furor which erupted when Twitter created
its own "Retweet" mechanism to counter the organic "RT" annotation. I
mention this for context for my comments below. Basically, the
"reshare" or "retweet" possesses three attributes of interest
(a) the content itself
(b) the chain of users through which this progressed
(c) the commentary added by one or more of that chain, to that
content

If the intent is to support the entire richness of this ability, then
the one part I didn't see in your description was the capture of this
'chain of influence'.

Having said that, with just (a) and (c) alone, I am inclined to see a
Reshare or a Retweet as a combination of a Share activity and a
Comment activity. If I just RT, that would be sharing, but if I modify
the tweet with annotations, am I not just simply posting a "comment"
about that content. Given tweets (and probably buzz shares) have
unique URLs, can this basically be formulated as a complex double-
activity (share + comment) where the activity object is still a URL?

Finally, to get back to my earlier point:
When I read danah's paper I got the sense that whom you RT-ed played
as crucial a part as what you RT-ed. So,
I think the critical aspect of such reshares is the capture of
influence that they bring (I know many folks who feel short-changed if
they are not given sufficient attribution in RTs, and who monitor
those mentions). Should this be modeled? (i.e., is it part of an
'activity stream') If so, how?

Apologies if this response was a bit off-course. Am still relatively
new to this space and may have interpreted or misunderstood aspects of
the original post :-)

Best
Nitya

[Darren Bounds]

Cliqset also has a 'share' concept. Shared activities use the 'share'
verb and create a new activity duplicating the object but replacing
the actor with user issuing the share. We then populate the
atom:contributor with the original author name and profile URL and
populate a link with the 'via' rel which references the original
activity shared.

Example:

<contributor>
<name>Ryan Penagos</name>
<uri>http://cliqset.com/user/agentm</uri>
</contributor>
<link href="http://cliqset.com/user/agentm/MTU4NjE5MDg0MTYe"
type="text/xhtml" rel="via" title="agentm posted a note on
Twitter"></link>

Darren

> --
> You received this message because you are subscribed to the Google Groups "Activity Streams" group.

> To post to this group, send email to activity...@googlegroups.com.

> To unsubscribe from this group, send email to activity-strea...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/activity-streams?hl=en.
>
>

--
darren bounds
dar...@cliqset.com

[Ivan Pulleyn]

On Thu, Jun 10, 2010 at 6:26 PM, Rabbit <rab...@cyberpunkrock.com> wrote:

Conceptually I think of it as writing "You'll like this. Great read! --Aaron" on a sticky note, slapping it on the cover of a book, and putting the book on your friends desk. The "content" is an annotation and intended to promote, publicize or otherwise elevate the attention of the recipient to the target object.

What about copying the content but then providing another element for wrapping the annotation (if it exists) so that they aren't separate but separable?

It's seems this should be approached carefully. RTs are often edited, with some content removed. Also, how is it different (if at all) from "via", or even "like"?

Ivan...

[John Panzer]

Should RT editing be thought of as summarization (conceptually like
the atom:summary vs atom:content distinction$?

> --
> You received this message because you are subscribed to the Google Groups "Activity Streams" group.

> To post to this group, send email to activity...@googlegroups.com.

> To unsubscribe from this group, send email to activity-strea...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/activity-streams?hl=en.
>

--
--
John Panzer / Google
jpa...@google.com / abstractioneer.org / @jpanzer

[Monica Keller]

I am not sure we need to make a big distinction between Reshare from Share which we already have. 
Share is specifically designed for content which is not owned by the actor.

Seems like we only need to add the attribution to the person


If someone [person 1] shares a video on Facebook and another person [person 2] re shares it then 
we model it the same as if person 2 had shared the video directly. On the UI (which I have added to the wiki) we show person 1's name 

so I like Darren's idea of the contributor to carry this information.

In our case person 1's annotation gets overwritten by person 2's looks like Buzz did the same. I think only Twitter offers the choice

[Anthony Leonard]

Retweeting is crucial concept. To borrow Kevin Marks's language, it is the personal recommendation that is at the heart of Twitter's success, even more than hashtagging, since it's richer to follow a person's idea of what's interesting (factual and phatic) than your typical topic-based RSS feed. Retweeting is therefore the way that news spreads across separate "publics". The switch from organic RT to official "retweeting" was also a precursor to Twitter's own nascent advertising model, which is a model you can be sure commercial users of activity streams will want to piggy back on (!)

It's crucial for the end-user (and of course the author/advertiser) to know who authored what. Therefore should be a clear separation between sharing an activity untouched, sharing it with annotation, and sharing with edits. Only the first two of these can be viewed like a concrete web link to be mined later.

I also agree with Ivan in the similarity of "resharing" with "like". It seems daft to me to see Buzz with a like counter, but no way of knowing who liked it? Better to stick with knowing how many times it has been reshared, and by whom.

Best wishes,
Anthony.

--
Dr Anthony Leonard
System Integrator
Computing Service
University of York
Heslington, York
YO10 5DD
+44 1904 434350

http://antleonard.blogspot.com/

EMAIL DISCLAIMER http://www.york.ac.uk/docs/disclaimer/email.htm

> --
> You received this message because you are subscribed to the Google Groups "Activity Streams" group.

> To post to this group, send email to activity...@googlegroups.com.

[Geoffrey McCaleb]

+2 on the like vs. reshare similarities. 

It seems we have overlapping use cases here. 

Both likes and reshares (and even favorite to a certain extent) can push an object onto the downstream graph with no physical change to the object itself (ie, no annotation or contextual change). With regards to a share, while the vast majority of RT's/reshares add little in the form of commentary/context, there is the possibility that the annotation that is added fundamentally reframes the entire object itself, thus making the "re-shared" content unique in it's own right. 

Or, commentary could be annotated onto the source content that offers a differing view to the original source, thus in a way branching it fundamentally from the original parent object. 

Geoffrey