The architecture of content and comments

The posts below about GoogleNews’ comments and this Fred Wilson post about centralized comments on distributed content coincidentally raise the same challenge about the strings that tie together content over time — and not just comments about it but also corrections, updates, and further reporting. Is there a better way to do this in the ever more distributed and atomized architecture of content online?

Fred says, and I agree, that content will become more distributed — see Edgeio’s content widget — and Howard Owens talks below about the necessity of reporters following up on their readers’ and subjects’ reaction to their articles wherever that reaction occurs. What can do that? I think that was the promise of Technorati and the wish of Cocomment. I prefer the emergent strategy: rather than trying to organize everyone to do the same thing or centralize at the same place, which isn’t going to happen, it’s best to try to discover the organization that already exists. But Technorati won’t deal with, for example, YouTube videos that are embedded with comments on them in blog posts everywhere. So the strings start unraveling.

What I want is something that maps all the relationships among pieces of content. This reminds me of a map Stuart Butterfield of Flickr once showed that visualized every interaction within the service: Person A comments on a photo from Person B and a line is drawn; person C tags that photo of Person B’s and a line is drawn, and so on; soon, you see collecting points around content and people. This is one of the elements that allows Flickr to find its influencers and to see what content they’re interacting with and that adds into the service’s algorithm of interestingness. This is possible inside Flickr because it is a contained and controlled content environment. It’s much more difficult in the open environment of the internet.

The problem is, in part, the architecture and use of the URL, the assumption that every piece of content has its home and everyone goes to that home to interact with it and that’s the way to organize the discussion around it. Not in a distributed world. I was goint to enter into a discussion of URN (uniform resource names) but quickly found myself over my head. So I leave that discussion to more knowledgeable folks than me, I hope.

Besides that map, I also want some qualitative view of what kind of content each piece is: an article, a comment, a correction, and so on. This would enable me to subscribe to just one sort of follow-on content: Show me when and if a correction is posted, by the original article’s author or by anyone else (which now means that I also want to map authorship and identity on this, making this more complex). If I blogged about that original piece of content, I could also automatically publish any corrections that are made. I wished for this in a Guardian column last year:

The internet can be better at corrections than old media. A fix can be attached to an error where it occurs, and many online denizens pride themselves on confessing missteps faster than their print and broadcast counterparts. But the internet can also be worse – online, errors can spread wider faster and take on a longer half-life. I wish we had a technical solution – that everyone who linked to an incorrect article could receive an alert and correction.

Similarly, I’d want to map the discussion that is occurring here, among many other places, around a controversial LA Times editorial about GoogleNews’ plans to add to the conversation with comments there. The LA Times piece appeared there and, for all I know, it could have been syndicated elsewhere and certainly was quoted in lots of blogs and sites. Then the author of that piece left a comment in the discussion of that article here at Buzzmachine and someone else left a reaction in turn. At the LA Times, I want be able to see those comments — and Technorati would enable that so long as each comment and post linked to the original article at its home URL. But it gets more complex if we were discussing something in a widget or video; we need to see not only links to original content but embeds of it. And it gets more complex again if we add this desire of mine for more qualitative identifiers: author or content type and also timing. On top of all that, I not only want lists and feeds, I probably need visualizations so I can see where the hot discussions are or where the original author is.

All of this enables the conversation and, as Howard Owens points out, can also improve the journalism.