Monday, March 17, 2008

Haven't Met a Comment - a FriendFeed drawback regarding lack of relations between similar metaentries

I just discovered something about FriendFeed that could be a potential drawback of the service, although it's also a possible opportunity for improvement.

Remember when I said that I was going to type some stuff at the FriendFeed entry for Duncan Riley's post? And remember when I subsequently noted that hardly anyone else was talking about Duncan Riley's post at the FriendFeed metalevel?

As Jim Bakker would say, I was wrong.

Turns out that there was a lot of discussion on FriendFeed about Duncan Riley's post - but it wasn't at the original FriendFeed entry for the post, but at the FriendFeed entry that resulted when Louis Gray SHARED the post. (There's a ton of discussion there.)

So, in essence, FriendFeed has two areas to discuss the Duncan Riley post, neither of which is connected to each other.

Actually, more than two, since other people may have put Riley's post in their Google Reader feeds, or someone may have linked to it in Twitter. Or in Jaiku. (It should be noted that Riley has not linked his Jaiku account to his FriendFeed account.)

So, because of the way FriendFeed works, there may be multiple unconnected FriendFeed items that link to the same artifact. In other words, the metacomments aren't meeting.

So how do you link them, other than linking them manually (as Louis Gray has done to relate the original Duncan Riley post entry to the shared item entry)?

Perhaps FriendFeed needs the capability to detect when two or more of its artifacts are linking to the same thing, then show "related items" information - in other words, someone who looks at Duncan Riley's post can also see that Louis Gray put it in his Google Reader shared feed. Alternatively, if they're looking at the FriendFeed entry for the Google Reader shared item, they can trace back to the FriendFeed entry for the original post. (That's what the Reed College students would do; we like the original sources.)

While this capability sounds like a good idea, I'm not sure how desirable it really is, or how much computing overhead is required to detect the related nature of various FriendFeed entries. For example, what happens if 2,000 FriendFeed users happen to place (insert popular story here) into their Google Reader shared items? Would the calculation of related link items bring the FriendFeed servers to a screeching halt?

Anyway, it's worth thinking about.

[mrontemp business] | [mrontemp politics] | [mrontemp technology] | [mrontemp tags]

Sphere: Related Content