Popularity and Sharing: The News you Should be Reading.
You’re interested in a lot of different things: coffee, baseball, geopolitics… probably cats. The world produces a huge amount of “content” relevant to these topics every day. A new roaster opened in New York. The Seattle Mariners won the World Series. Kim Jong Un stared longingly at a piece of cake. Some of these things might genuinely interest you, whereas other things might not. Not only are you picky, but you probably prefer hearing about these things sooner rather than later. News that a forest fire is raging through your neighborhood is significantly less useful after you can already smell smoke.
Let’s frame this in another way: if you could watch the newest Game of Thrones episode a week before it came out, would you? I would guess for most people the answer would almost certainly be yes. Knowing, and knowing earlier, gives you a certain social power. You’d be a veritable god compared to your poor friends! They’d be afraid to check their text messages, afraid that one might be from you. You might callously let it slip that Ser Jorah and Danaerys hook up in the next episode! This information has a time-value. Having it sooner (and having it sooner than your friends) is generally better than having it later (and after your friends). The later you get it the less socially valuable it is. Trigger for iOS is the only news reader built from the ground up with this in mind.
When it comes to news or content these are the two things that stick out as particularly important: relevance and timeliness. In an ideal world you instantly know all the things that matter to you, as if somebody is cross-referencing the database of things you care about with the database of things that are happening and giving you the hits in real time.
This world doesn’t exist… yet. Maybe we don’t even want this world to exist. Google’s coming close, but still can’t lay claim to knowing everything about your interests. If you want to get philosophical about it, maybe privacy requires a third party to never know quite this much about you. Regardless of where you draw that line, we here at Trigger believe that we’re all better off living in a world where we’re a little bit closer to the information that we care about.
I’ll write a bit later about how we do relevance, but for now let’s focus on timeliness. At Trigger we claim to be able to deliver you the news “49 minutes before it starts trending”. What does this mean and how do we do it?
The majority of mainstream news readers, even the ones we personally use and love (read about them here), use popularity as a major factor in determining which articles you actually end up seeing. “Popularity is good” is the broad assumption that powers most modern news readers. If you’re tasked with serving a user up with one article from a dozen different sources, which articles would you choose for them without any additional information about their preferences? The safe bet is just to choose the article that the largest number of other people have already read and enjoyed. While this generally isn’t such a bad strategy (and gets even better with some additional personalization on top) it fails in one major way: you always get the news after everybody else who made it popular.
If you want to make money in the stock market you need to beat the crowd. You need to predict where a stock is going and buy (or sell) it before it gets there. If you want to get the most of the news you need to do the same and our “social momentum” algorithm attempts to do just that.
We’ve went back through all the available sharing data we could find and carefully examined the sharing behaviour of popular and unpopular articles alike. Just like people try to find patterns in the trading behaviour of stocks on the market, we tried to find patterns in the behaviour of articles being shared. What we were looking for were, in essence, trending rules; sharing behaviour quirks common to popular articles (or unpopular articles).
What we found seriously surprised us. There were some powerful predictors of an article’s success or failure. Even more surprising was that, in some cases, a few of these predictors arose very shortly after an article’s publication. While we’re still fine tuning the algorithm, we found ourselves able to predict the popularity of more than a few types of articles quite accurately.
One of the biggest indicators of an article’s future popularity (and our algorithm’s namesake) is its sharing acceleration. Once an article is published it spreads outwards via shares on social media, direct e-mail/message, and discussion on forums and other blogs. We found that articles that experience large bursts of sharing early on in their life almost inevitably became more popular than their more slowly accelerating counterparts. For a variety of complicated and interconnected reasons, slow and steady DOES NOT win the race in the battle for online popularity.
We’d love to tell you more, but why not just come experience it for yourself? Trigger’s brand new version 3.1 is out now for iOS, featuring a completely redesigned (and in my opinion significantly more awesome) main feed. Our V1.0 for Android will be coming out this week as well so keep an eye out for that too.