After a year or so of invite-only alpha testing, NewsCred launched as a public beta service this morning. The site, founded by Shafqat Islam and Iraj Islam, is trying to create a kind of outsourced reputation system for news websites and blogs, in which users vote on the credibility and accuracy of specific news stories or blog posts, and those votes are combined with the site’s own algorithms to generate a credibility profile. It’s an interesting effort, and one that I think will likely appeal to many news and blog readers, since we’ve probably all read things and snorted in derision at the unbalanced or inaccurate take someone has taken — both in professional media and on blogs. But is NewsCred the solution?
In a sense, Newscred is trying to take the Digg or Slashdot model a step further. When people Digg a story or link, they are often simply voting on whether they like the topic, or the photo, or in some cases whether they like the person who Dugg the link. Newscred.com wants people to explicitly vote on the credibility of the site itself, (or at least the author of the story or post). As more than one person has already pointed out however, credibility is a difficult thing to measure, and it’s not clear whether it’s the kind of thing that a site like NewsCred is going to be able to outsource or generate through an algorithm. If someone clicks the “discredit” button, is it because they don’t like the author? Or because they simply disagree with them, if it’s a blog?
Although it might complicate things, I’d like to have the ability to see the track record of the individuals who voted to credit or discredit a story, so that I could judge whether they were the type of person whose opinion I valued or not. As a friend of mine has said before about recommendation sites like Tripadvisor.com: What if all of those people like things that I don’t? Then their advice is meaningless. It’s also impossible to know what criteria NewsCred takes into account when it derives its own credibility rating for a specific site. Is it tracking corrections or updates? How much weight do individual rankings carry?
NewsCred isn’t the only site trying to add a layer of credibility to the news. Another service that does something similar is called NewsTrust, a non-profit entity founded by former journalist Fabrice Florin. NewsTrust’s advisory board includes Dan Gillmor and Craigslist founder Craig Newmark (who has also financed the site), and the service has gotten a multi-year grant from the MacArthur Foundation, among others. Josh Catone, formerly of Read/Write Web, has a good look at the differences between NewsCred and NewsTrust.
Update:
Shafqat Islam responded via e-mail to some of the points I raised above. Here are his responses:
1) We will be building out social profiles on our site, and allow everyone to see how people have voted. Note that its not our intention to make this a social site – simply to allow users to set up profiles so others can track their voting or simply their source selection. Of course, you will have to opt-in to share some of the sensitive data, but not your votes. That should be transparent as you mention.
2) We plan on using some of this user data to build a recommendation engine. Very similar to what Findory was doing back in the day.
3) You are spot on about the issue with binary voting (credit/discredit). On one hand its simple and a lower barrier to entry for people who want to participate. At the same time, its not granular enough to make meaningful conclusions so we are going to add (in the next few days) an option when you vote to check a few boxes (‘not fact-checked’, ‘biased’, ‘no disclosures’ etc) or leave a short explanation.
For the final point about whether people will just vote based on if they agree/disagree, that is a real concern. We are hoping that with a critical mass, the wisdom of crowds effect really does take over and the overall, aggregate knowledge of the crowd will be fairy accurate. Some people will always vote based on their own biases, but we can implement technology to prevent gaming or to detect when a user only votes for a certain source or constantly discredits a journalists.