Since FairSpin’s public launch yesterday we’ve received lots of positive comments, constructive feedback… and votes! Thanks to all of you who have tried it out so far.
We’ve also seen a few really good recurring questions that we wanted to take a moment to answer here.
“How are the bias ratings on FairSpin determined?”
They are calculated automatically and entirely based on user votes. Bias is calculated when either of two different events occurs: a user voting on a story, or a new story entering the system.
When a user votes, the vote count for the story in question is incremented and the specific type of vote (e.g. left, right, or middle) is recorded. The system then looks at the entire set of votes tendered for that story to date and applies our ranking algorithm. This algorithm looks at where the balance of the votes fall from left to right, and from that determines the overall bias rating. This means that the bias for a story is recomputed every time a user votes on it, so it can change in real-time.
It’s also important to note that every vote for a story is also tabulated against that story’s author and source. This enables FairSpin to build up an increasingly accurate picture of the historical bias of both the authors and the sources they write for. Why does that matter? Read on…
When a new story arrives, FairSpin checks if it has seen this particular author or source before. If it has, then chances are that they each have their own historical bias rating (as explained above). The authors’s bias rating is applied to the new story, or if there is no author specified then the source’s rating is applied. This serves as an “educated guess” and enables FairSpin to place the story on the homepage spectrum and draw your attention to it. But as soon as human users start voting on the story, the role of this “educated guess” is reduced.
So what does all this really mean? It means that all the bias ratings on FairSpin come from you, the users. It also means that FairSpin is constantly learning and refining its picture of the political spectrum, and your votes play a direct role in that process.
“FairSpin says The New York Times is ‘fair’? But that’s just silly!”
What you’re actually seeing in this case is a story by The New York Times that is rated fair. As explained above, initial story ratings are based on the historical bias ratings of the story’s author and source. In the case of The New York Times, this means that some stories show up as fair, while others show up as left. Sometimes they even show up on the right (former NYTimes columnist Bill Kristol’s work is a good example). Once users vote on a story, those votes take precedence. So the best way to help us make FairSpin more accurate is to vote on the stories you read!
“Isn’t this hopeless, since the site going to be overrun by [liberals|conservatives|libertarians|sea monsters]?”
We’ve of course seen other online social communities become dominated by one group to the detriment of another, so we know this risk is real. But we think the concept of FairSpin is inherently nonpartisan and will appeal to both sides of the aisle. We’ve also made some design and feature decisions with this issue in mind and we will be doing our best to ensure FairSpin grows in a healthy and balanced way.
And by the way, the fact that you asked this question means you care about the problem. We’re glad to have you on the team.
“Aren’t you guys biased yourselves?”
Well… yeah. Of course we are. But everyone is. Show us the person who is 100% unbiased and we’ll show you the positronic brain we’ve been working on. :)
To be serious, we all bring our own internal biases to the table. The question is whether or not we recognize them and what we choose to do about them. By building FairSpin, we’re hoping that people (including ourselves) will find it a useful way not only to reveal bias in the media, but also to explore their own biases.
We’ve tried very hard to avoid designing any unnecessary bias into FairSpin. In some cases this has led us to omit features. A good example: unlike many social media sites, FairSpin does not allow users to submit news stories. This is because people may have different — and not always alturisic — motivations for submitting a particular story. Instead, all stories are gathered automatically by the respected news aggregator memeorandum. Since memeorandum has been around for a while and is almost entirely automated, it tends to be less biased and covers a broad and diverse set of sources. Using memeorandum as our data source allows FairSpin to focus on the key challenge of detecting bias in the news instead of fighting two problems at once.
“Isn’t left vs. right a simplistic and unrealistic representation of bias in the real world?”
We generally agree, and we have some ideas on better ways of representing this, but we started with what we have because it is something that everyone is familiar with and can relate to in some way. We also wanted to keep things simple so we could launch sooner and act on your feedback sooner. We hope that if FairSpin gets enough momentum it will provide an opportunity to try some different approaches in this regard.
Thanks again for the questions and comments, everyone.
—Stephen and Dave
10 months ago