automatic newspaperism (some clues to paper.li)

The idea of automatic newspapers
There's more and more content on the web - everyday, new links float through twitter and facebook, in an ongoing stream of data. The idea of automatic newspapers like paper.li is to help to curate this content:  "We believe that people (and not machines) are the ones qualified to curate the content that matters most. We also think that these same people can greatly help their own communities to find their way through this “massive content world” we live in."

a blueprint of a newspaper
Since this summer, there is a BluePrintReview newspaper, which is still in the trial/error/adjust phase, but comes with daily editions that often look like they have been pieced together by an editor. (Well, there in fact is an editor at work: behind the settings, and to some degree, with the content). And added to that, the generator often pulls interesting articles that otherwise would have gone unnotice, especially as Twitter is short-text only, and Facebook has the tendency of pre-selecting content by cutting down the newsfeed to a core circle of connection. (more about that, here: facebook cuts newsfeed)

Here some clues to creating / optimizing paper.li papers:

• the newspaper is created automatically, based on suggested keywords for twitter and facebook. twitter settings allow to be more specific and seem to work better than facebook settings/keywords
• it's basically trial and error: experiment with settings and see what works best
• paper.li seems to work especially well with events that have an own hashtag in twitter, like #storysunday or the 100days group project: #100days2011
• it also works well with very specific keywords. for a freelance project, i tried a trial newspaper, and with only one keyword: "webcomics", it creates interesting editions.
• format: to add a single twitter user, use the format: @twittername. for a twitter list, the format is: twittername/listname.
• test suggestion: to get an idea of how the various keywords / settings work, i suggest to try each as single setting in a test newsletter, and look how it shapes out, then erase the test again, and try the next setting in a new test (that way, you can work ahead instead of having to wait 12 or 24 hours for a new edition after changing settings). some things i noticed there:  it seems rss-feeds sometimes create no content, and keywords like "book" raise japanese content, which can be fixed by setting the content language to english
• i tried "photo" as keyword in the settings, which then produced a photo and media field (not sure if those would have come up anyway at some point)

adjust content

• once the edition is published, you can adjust / control the content on the newspaper page by hovering over the right lower corner of each newsbit – hovering will make a field come up that allows you to move up into the headline, move down from headlines, or to erase the newsbit.

more clues

• some more clues can be found in the paper.li blog