Schwanksta :: Fuzzy Journalism

/*I’ve been having extreme issues with my host, so this post has been in draft form for about a while now. I’m just hitting publish and crossing my fingers….*/

Inspired by the conversation occurring in my last post, and spurred by some video I don’t feel like editing, I’m going to flesh out my theory on how newspapers should treat their data.

To borrow from a comment, I can’t stand when people say that we’re being held back by those who “don’t understand X-web-service.” There needs to be a clear shift in news websites and how they handle the information that the organization collects before we worry about getting our readers to Twitter. (I know it might seem like I’m beating up on Twitter, but it’s one of those very useful services that simply takes too long before you understand its capabilities. I like it, I swear.)

Non-news Web services teach us a lot about how conversations and interactivity evolve on the internet, but I don’t necessarily see the future of journalism in them. Part of the draw to newspapers is the ease of use and the idea of a single point for information. Besides reading and page-turning, no special skills go into obtaining the news therein. Information needs to be *that* accessible and simple for our online readers.

Dave Stanton makes a good point, and that’s that newspapers should hire statisticians. This becomes even more relevant when we begin to link our information together and patterns will no doubt begin to emerge in the data

In any case, the type of system I’m proposing would have to be extremely context-aware. This would most likely come about by the combination of human inputs (in-story tags, etc) with what the system knows (links forged within the database, acrobatic algorithms, whathaveyou).

Take this with a grain of salt, as it’s sort’ve spur-of-the-moment, but…:

Imagine if every person a newspaper used in a story got a database entry, Person (could be subclassed to Source, etc. if necessary). You post a story about this person, and tag it [storytag:person role="primary"]Pegeen Hanrahan[/storytag] (it would need probably need to be more nuanced than just saying she’s the “primary person in this story”).

Your system knows already from the general story tag that this is a “local_election” story. This means that as you publish, the CMS would update her database entry with a link to this story, and and Article object would be created for it with the new-found knowledge that it is a “local election story primarily about Pegeen Hanrahan.” Now you, as the editor or reporter, can choose what related information to bring up and display about either Pegeen or local elections.

The same links can be made backwards in the database, too. If you post another story a week later about Pegeen or local elections, the CMS should pull this story from the database as a reference link, among others.

Now, of course, that’s a simplistic explanation and example. What if it’s a story about Pegeen Hanrahan, during a local election, primarily about her promises to give more money to the local animal shelter? Puling up other stories about the election might fit, but what if the animal shelter is in dire need of funding, and the story is heavily about that?

Well, then the tags should dictate: [storytag:focus]Gainesville Pet Rescue[/storytag]. Assuming your reporter has done their job and gotten some financial information for the pet rescue, and hopefully inputted it by now (all input of primary sources should happen before the story is tagged & parsed), the CMS should automatically pull up the record and ask what to include from it in the story. If nothing else, it should automatically link the name in the story to a new page which will display some of the contents of the database record.

I don’t know if you can really get past the human element of it, and I’m not sure if you’d really want to. I’m sure you could write some code that would approximate what I propose reporters/editors/online staff do, but I think the human touch would probably result in more relevant info.

Clearly my example tags are probably not the best way to do this, but it works for a spur-of-the-moment show of what I’m talking about. Perhaps the best way to handle it would be to have the metadata separate from the actual article, or as a part of the database’s Article model (there would of course be models for Video, Business, etc.). But either way, the point remains the same: we tell the computer what the article is about, and it gives us the related information we should include.