I stumbled across a new site called paper.li that allows anyone to create a "newspaper-style" digital publication from a variety of Twitter feeds. The process takes a few minutes, and offers some interesting prospects.

The tools are in alpha, so things are raw and can change at a moMy customized paper.li "newspaper"ment's notice, so my descriptions are based on what I did today.

There are three options for your publication by:
1) Pulling your feed plus those of the people you follow.
2) Pulling a feed from a designated #hashtag
3) Pulling a feed from a Twitter list

I created two: "The Logan Molen Daily" based on my @lmolen handle and the 253 people I follow and "The journoinnovation Daily," based on Steve Buttry's @journoinnovation list, which pulls in my Tweets.

The paper.li "pagination" process takes a few minutes and creates a publication pulling in posts sorted by topic categories.

For Logan Molen Daily, paper.li gave priority to "Media" posts, followed by my Tweets, "Media" (aka video), Politics, #media posts, Arts and Entertainment, Photos, Business, Education and Technology. Ads were served by Google, and included a geotargeted Groupon banner (cripes, those things are like weeds), as well as a banner pitching Joan Jett-branded items. I'm guessing that's targeted based on my frequent posts about rock music; instead of Joan Jett, the journoinnovation mix included an ad for ClassesUSA. Paper.li is also pulling in information from local businesses using Constant Contact. This is the first I've seen of Constant Contact as a content-exporting tool, and media companies should keep an eye on that kind of usage in their local markets.

Example of a newspaper culling Tweets from a topical list"The journoinnovation Daily" had a slightly different mix, which lower play for "Politics" and "Arts and Entertainment" but higher play for  "Stories." and "Business." I'll have to play with some other feeds to get a better handle on the paper.li algorithm.  But here's what its FAQs say:

Publishing a newspaper is no small task! Here is an overview of all the things paper.li does in the background for you:

• extract all tweets that include URLs

• extract the content found on these URLs:

  • text, e.g. blog post, newspaper article

  • photo, e.g. Flickr, yfrog, Twitpic, ...

  • video, e.g. YouTube, Vimeo, Dailymotion, ...

• analyze the extracted text for topic, e.g. Politics, Technology, ...
• surface the day's most relevant articles (using paper.li magic)
• construct a newspaper frontpage using the filtered articles, photos and videos

Publications are updated and published in one of four languages every 24 hours by default, but the frequency can be customized. ,You can promote your publication manually or automatically using Twitter. There's also a prominant Facebook "Like" button.
As I noted earlier, Paper.li is in alpha, so there are some rough spots. The automated pagination leaves some overly large white space, the videos sometimes don't load in the first frame, and the ads somethings don't pull in graphic images.

But for alpha, paper.li offers a lot of promise. I'm reminded of FlipBoard, which does a nice job of creating a social-media publication on the fly. Paper.li doesn't deliver the same kind of serendipity but I see it being a great way to dive into a sea of Tweets once or twice a day, whether it's my feed or something more narrow. The differentiator will be in the "paper.li magic" mentioned above, which must be some kind of relevancy and influence measurement that determines ranking. Like Google algorithms, the devil is in the details.