What You "Like" Isn't What You're Like: The Unflattering Reflection of the Facebook Button

In the past two years, the world has adopted a strange and new kind of currency: the Facebook Like. It’s not equivalent to any real amount of money in any typical sense, of course. But on the Internet, every new "like" for something—a website, a company, a product, a personality—is like the Midas touch, giving that content an air of meaning, and perhaps economic value.

This is not so unreasonable: the “like” is a useful tool for determining how something is being seen, appreciated, and shared, and by whom, and provides valuable insight regarding what kind of content creators should be creating. Indeed, our redesign here at Motherboard has made it much easier for you to “Like” our wondrous literary and documentary gems. (You should feel more than welcome to Like this particular story.)

The most interesting thing about the Like phenomenon is that the data behind the Likes can be used to figure out just what the hell people really care about in the endless wasteland of the internet. With those questions in mind, Yury Lifshits, a researcher at Yahoo! Labs, recently put together the “Like Log Study.”

Compiling information from 45 sites, over 100,000 articles, and over 40 million reactions over three months, Lifshits came to some interesting conclusions about what we like, what we read and don't like, and how you can optimize your computer contributions to win the approval of Facebook and make yourself rich in Likes.

Yuri summarizes his project here:

The tables are really the only way to get the whole picture of this information, but here's a short synthesis of what he found.

The top five sites according to Yury's study, based on total Likes between Oct., 2010 and Jan., 2011 are:

• New York Times (6,815,796 total Likes)
• BBC (4,331,367)
• NPR (2,549,613)
• The Guardian (1,883,161)
• Yahoo! Sports (1,774,183)

Pretty wild, really, that Yahoo! Sports beats CNN, The Wall Street Journal, Huffington Post, and Fox News. Sad but not surprising really that it beats The Economist.

More curiously, here are the top five most Liked stories within that time frame, as determined by those who feel the need to use Facebook to tell their friends about something great they found:

• Why Chinese Mothers Are Superior: (342,294 – Wall Street Journal)
• New Zodiac Sign Dates Earth Rotation Changes Horoscope Signs (328,192 – Time Magazine)
• No, Your Zodiac Sign Hasn't Changed (238,596 – CNN)
• A Holiday Message from Ricky Gervais: Why I'm An Atheist (164,166 – Wall Street Journal)
• Tired Gay Succumbs to Dix in 200 Meters (152,780 – Reuters)

First of all, is anyone else perplexed by the fact that the second and third most popular stories directly contradict each other, with the one that is seemingly providing false information in the lead? Second, the fifth awkwardly-titled item, in which a “tired Gay succumbs to Dix,” was substantially more popular than NPR’s reporting of the shooting in Arizona, and BBC’s story about Burma releasing Aung San Suu Kyi.

Strange? Probably not. As Lifshits uncovered, “Among [the top 40] stories there are only four articles about factual political news and three about celebrities. The most common type of hit stories is opinion/analysis. Other common themes include: lifestyle, photo galleries, interactives, humor and odd news.”

My favorite part of Lifshit’s study was his section on “Engagement Trends” (about halfway down the page). When it comes to companies or products, apparently people really like to read and share stories about Facebook, Apple, Skype, and Kinect. On the other hand, while coverage of Amazon, Windows, and Samsung are common topics, they don’t engage the audience very well.

As far as topics go, the hot stories include keywords like “how”, “why”, “best”, and “future”, whereas “search”, “tv”, and “cloud” get little “Like” love from readers. The biggest wild-card in the study was video features: some sites (like yours truly) effectively integrate pretty moving pictures (ie. All Things D, Engadget, Mashable!), whereas others are really struggling to make it work (TechCrunch, CNET, GigaOm).

Funnily, the business side of things doesn’t get a lot of props from the internet-going public: “CEO”, “Funding”, “Sales”, and “Enterprise” are terms that don’t seem to be very well Liked on the virtual playground. That doesn’t mean these stories are not getting a lot of page views, but readers don’t seem to think that sharing them will up their social media street-cred.

Yury’s study is a great tool to help us visualize what Internet traffic looks like, but he’s not the only one doing it. Buzzfeed has made a business out of using Facebook Likes and other indicators to determine if there is any rhyme or reason to what will go viral. The site (whose founder we profiled in this video) has designed a number of tools to help them figure out what we read, such as the Viral Dashboard. Currently in beta testing, it shows how articles/lists/etc. propagate through social media, Google, and StumbleUpon.

Ky Harlin, a data scientist at BuzzFeed, says his experience matches up pretty well with what Yury described:

Ultimately, we see that the most viral articles are the ones that can be universally appealing/entertaining to someone who isn’t necessarily a constant follower of the topic on which it concerns. In other words, content simply needs to be engaging. People’s activity on the web is more sporadic than ever before—it’s not like the old days when people were loyal to a single website/news source. Now they’ve got tons of sources to choose from and they’ll leave your site in a second if they don’t find something that can quickly catch their attention. It also depends on the time when an article was published. Something might do really well on a slow news day while it would hardly be noticed on a day when a different big story broke. Other stuff that matters is the format of an article—for example, we’ve found that lists tend to do particularly well.

Point taken. Here’s Ky’s take on what makes a Like-able article, in list form:

1. Grab people’s tiny attention spans with a headline and story that most people can understand
2. Release your content on a slow news day
3. Make it a list

Look at Buzzfeed’s Viral Dashboard – or search your logical brain – and one thing is clear: stories of the naked celebrity variety are much more popular than news about peace in the Middle East (or lack thereof). Still, as you might expect, while pornography makes up a massive amount of Internet traffic, it’s representation on social media is immensely small by comparison. People seek it out either by going directly to these sites or by finding them through search engines – not by, say, finding them on friends’ Facebook walls.

It isn’t complicated to figure out why this is, but it does suggest, again, a disconnect between what your friends are like on the Internet versus what they might be like in reality. As Ky explains:

In general, we’ve found that there is a disconnect between what people want their friends to know they’re interested in and what types of content they’re actually most interested in. A lot of the time, people are most interested in ‘not safe for work’ content: articles involving sex, pornography, and the like. Since our site is often browsed by people at work, we try to eliminate most of this content from buzzfeed.com. But yes, NSFW content does see some of the highest numbers in terms of seed content while their viral-to-seed traffic ratios are generally very low.

Yury has similar views about this, and points to Cosmopolitan, Sports Illustrated, and Maxim as the kind of sexual content that is accepted on the mainstream Internet, just as it at your supermarket. His next project aims to compile the statistics from 20 woman’s magazines to seek out some trends specific to this field of content. The most shared article about sex he found wasn’t about sex per se, but a 300-word Women’s Day article from December 2009 about the “10 Things Husbands Should Never Do,” posted to Yahoo! Shine. It has well over 5,000 comments.

The point remains that Facebook Likes can tell us a lot about only one specific segment of information consumption on the Internet: the content that makes us likable online, and not necessarily the content that we really like.