Why we created Paper.li
Clay Shirky: “It’s not information overload. It’s filter failure”.
On the web, the amount of content vying for our attention is exploding. Some kind of content curation is thus needed, now more than ever. Such function has typically been provided by the editors of our daily newspapers or magazines. The last couple of years, however, has seen a shift in how we find & consume news. According to a recent Pew Internet study, 75% of news consumed online is through shared news from social networking sites or e-mail. The social network of a reader is becoming their personalized news wire.
Twitter has become a choice platform on which to share content socially. In fact, the service has so successfully simplified this act, that users are still (or once again) faced with the issue of content volume. What’s more, for most of us, reading tweets in a dizzying stream is a sub-optimal experience to say the least. The act of sharing news socially has been solved, its efficient filtering and consumption has not.
This is where Paper.li comes in – filtering.
More than just presenting the news shared with you in an easy to read manner, Paper.li acts as a powerful relevance filter. To do so, Paper.li does 2 things behind the scenes: semantic analysis of shared content and it’s ranking.
In order to provide an efficient content discovery and reading experience to a user, Paper.li needs to know more about the content being shared with that user. Paper.li thus semantically analyzes all URLs shared with the user, and assigns a topic (e.g. Technology, Health, Art, Environment, …) to each. This is how Paper.li can provide topic sections in each paper.
Paper.li currently analyzes over 10 million links a day, in four languages: English, German, French and Spanish.
Contextual relevancy ranking
While semantic analysis of content helps in structuring its presentation (i.e. topic sections), there also needs to be a way to determine the relative importance of content items. This enables the presentation of news not in terms of chronological order, but in terms of relevance to the user. This ranking of content is done using Paper.li’s own set of ranking algorithms, which take into account a mix of variables relating to both the context and the content.
Uses of Paper.li
Since we launched, we see the service being used in all kinds of manners. The obvious one has always been to read your own Twitter stream, i.e. a more personal usage. But over time, there is a clear trend towards using Paper.li in a more publishing mode, creating papers of interest to a wider community of users. While some personal papers of celebrities have a large readership, e.g. The Joe Rogan Daily, here are some other uses we see a lot of:
work groups (based on Twitter lists or hash-tags)
trade shows (generally based on Twitter hash-tags)
news events (generally based on Twitter hash-tags)
brands (generally based on Twitter or Facebook search queries)
causes (generally based on Twitter hash-tags or search queries)
Launched last summer, Paper.li reached 2 million unique users in over 200 countries and publishes 150,000 online newspapers.
Where to from here
Here are some of the things we are working on:
geographical expansion, including additional languages support for Dutch, Italian, Portuguese, Japanese & Chinese
multi-device support, including smart-phones and tablets
paper customization features, e.g. removing/replacing ads, manual content curation
platform API – integrate other content sources, integrate Paper.li technology in other services & media sites, etc.
Paper.li basically wants to become smarter at what it already does as well as provide more publishing options to paper editors.
We are receiving hundreds of comments & feature requests on Twitter @smallrivers, this blog or by email firstname.lastname@example.org – keep it coming! We try our best to respond to all, and knowing more about your appreciation of the service and what you would like to see us doing in the future is key.