I wrote a widely shared post in June titled “Facebook can save itself with two simple measurements it already collects.” Some three months later, Facebook has implemented one of the two changes I advocated for. I was originally inspired to write that post during the Kim Kardashian / Kanye West wedding, when Facebook user feeds were taken over by clickbait-y articles, photos and gossip pages that were provided in such duplication that it may as well have been a Zappos retargeting campaign. (Go put anything in your cart at Zappos.com, but don’t check out and then browse the web as you normally would. You’ll be seeing Zappos ads with the shoes you were looking at on nearly every site you visit.) When they implemented one of those recommendations I was as excited as I was surprised. So what is this silver bullet?
Time on Site
Facebook has decided to add “Time on Site” as a metric to their surfacing algorithm. I pointed out previously that “Time on Site” is not the best solution, but it is one step FB could take right now, with data they were already collecting. They should integrate it into the surfacing algorithm (news feed) and users would see an incremental improvement in the quality of posts being surfaced. In the same post I proposed the addition of “bounce rate” to augment the “Time on Site” metric, but even without the additional metric, we are seeing improvements. I could spend the remainder of this post doling out other algorithm changes that would improve the user experience but instead, I want to take a step back and talk about algorithms in news feeds in general.
Algorithmic news curation is a lot like the atomic bomb
Don’t get me wrong. I love the idea of algorithms. I love the idea that with a large enough data set you can predict what someone is most likely to be interested in and then make money from it. Heck, almost every dollar I’ve made in my adult life has been derived from algorithms in some fashion. But even I, the hard-driving technologist and an unabashed believer in the freedom of the Internet, have a problem with algorithmic NEWS and NEWSfeeds. NEWSfeeds should not be curated by computers. There. I said it. I’ll say it one more time for for you, Twitter. NEWSFEEDS SHOULD NOT BE ALGORITHMICALLY CURATED, at least not with the motivation with which we do it today…which is, for the sake of monetization.
Algorithmic news curation is a lot like the atomic bomb. Just because we have the technology does not mean we should use it. Yes, we can see what you’re clicking on and each one of those clicks adds a few more cents to the company coffer. So you like photos of burgers and top ten lists that point you in the right direction while you wander downtown Manhattan looking for the best rooftop brunch. Is that the only news that you should be shown? The algorithm thinks so. The algorithm thinks that ‘Top 10 Photoshop Fails’ is more important to you than the complete evaporation of independent management in foundations of the Internet. You probably won’t click on the link right before this sentence even though it’s a lot more important than the cutest cat photos in the world. There have been hundreds if not thousands of profoundly meaningful technological breakthroughs in the last 20 years. Think: Stem-cell Therapy, Genetically Modified Organisms, and now the Algorithmic News Feed. All of these have potentially devastating consequences, but there is only one that isn’t saving lives in the mean time: The Kardasian Powered Newsfeed!
This isn’t news to you though. You already know that Fox News (and basically all other cable news channels) exists not for the distribution of news, but for profit, and when their analytics tell them that more people watch vitriol and conflict, they put vitriol and conflict front and center. Facebook does the same thing and now the even playing field that was Twitter, is following suit. My point here is that if these services that take up 60%+ of your daily attention only show you what you will click on and thus what will earn them advertiser dollars, you only get news that is popular; popular in your circle of friends, popular in your demographic or you geography or heck, even your psychographic. You end up with a very homogenized product and a narrow-sighted view of the world. It’s the equivalent of hiring a “yes man” who only gives you the good news all day (even if the “good news” is actually bad.) Twitter was once the alternative to Facebook, where the community spread the content; where a credible retweet could send an article around the globe in a minute and a half. With Twitter jumping into the algorithmic feed fray, ostensibly for the sake of their investors, there is no longer a heterogeneous meritocracy for global news. So, yeah, this is sad news, but it’s also an opportunity. Each time a giant media property moves closer toward automated consumption The New York Times and other traditionally specialist-curated properties are given another opportunity to differentiate themselves. Eventually the information we find on social media will be so homogenized that we will again begin to seek those specialist-curated sources that in decades prior were the bread & butter of passive information delivery.
Originally published at www.nationalmetrics.com on September 16, 2014.