Loading…
0:00
15:52

The Accidental Invention of the Like

In 2005, Rob Manuel had a problem. Four years earlier, after the dizzying rise and crash of the dotcom bubble, he’d started a site called B3ta.com with his friends Denise Wilton and Cal Henderson. Like many post-crash projects, it was an attempt to build something more human and, well, funnier than the pompous, shiny-suited fantasies of dotcom-era startups.

At its heart, B3ta was a newsletter and a message board, two technologies that harked back to the pre-web communities of the early 1990s. B3ta was a place to share jokes, often in response to weekly image challenges that encouraged the kind of mashup amateur meme aesthetic that is now commonplace. It quickly generated a dedicated community of creative talent, sharing everything from simple image collages to fantastically surreal animations and parodies of art history.

By 2005, the B3ta community had grown so large that the simple message board–style front page was getting difficult to navigate. Manuel needed a way to curate the content so it was easier to find the good stuff.

It turns out this was a problem that a lot of early-2000s web communities were trying to solve. The mysterious FilePile, launched in 2000 and a big influence on B3ta, had plus and minus buttons for up- or downvoting content. Digg.com launched in December 2004 with a simple “Digg” button to rate links. But even before FilePile and Digg, sites like Hot or Not (the inspiration for Mark Zuckerberg’s pre-Facebook project FaceMash) used buttons to measure users’ approval. Manuel was still hand-curating B3ta and saw these voting buttons as a potential solution:

“I needed something to cope with the large amount of images and stories being submitted. It was exhausting to try and read it all, to moderate the site and find content for the front page and newsletter — so it seemed an obvious enough move to allow users to vote. My thought was that I wanted a halfway house of editorial control. I get the final say, but user suggestions could guide me to the better content.”
Email from Rob Manuel to his developer on B3ta; November 2005.

Manuel emailed his developer to suggest adding the feature and initially wanted to call it a “wooyay” button, fitting in with the kind of informal language the community used on the site. It’s not clear now why B3ta decided to go with “like” instead of wooyay, but regardless, by November 2005, B3ta had launched its version: a button reading “I like this!”

Example of the ‘I like this!’ button on b3ta.com shortly after launch in 2005. Image: Rob Manuel

Building an Engine for Likes: The Social Graph and the News Feed

The Facebook we use today is nothing like the online student directory Mark Zuckerberg launched in 2004. Over the past 15 years, Facebook reached global dominance through a combination of three core concepts — the social graph, the news feed, and the “like” button. Facebook didn’t invent any of these concepts — the company’s success has come from its ability to mix these three elements into an increasingly sophisticated cocktail while managing the competing needs of users, advertisers, and — more recently — regulators.

It was this heady cocktail that differentiated Facebook from the other web community sites of the early to mid-2000s. Whereas FilePile, B3ta, and others were happy to remain as relatively obscure creative communities, Zuckerberg’s ambition and venture capital funding required exponential growth. In the years between 2007 and 2009, Facebook took these three elements — all of which came out of the web 2.0 scene of the mid-2000s — and used them to build an engine that could deliver this growth.

The first element, the social graph, was explained in a 2002 academic paper by economics student Philippe Bouzaglou, a Harvard contemporary of Mark Zuckerberg. This described how to map the interwoven connections between groups of people, in a similar way to the Six Degrees of Kevin Bacon game that was popular on internet message boards in the 1990s. Most social media startups in the 2000s used a social graph of some kind, from photo-sharing site Flickr to the pioneering but short-lived SixDegrees.com.

For these sites, the social graph was implemented mainly for users’ benefit —to help you find a connection with another user or to see what your friends were doing on the site. When the social web was still young, just showing a user how they were connected to others was thrilling enough.

At the F8 Facebook developers’ conference in May 2007, Zuckerberg announced a far more ambitious plan for the social graph: “Until now, social networks have been closed platforms. Today, we’re going to end that.” Rather than being limited to a group of connections on one site, Zuckerberg saw the social graph as a new way of connecting everyone on the internet. “The social graph is changing the way the world works,” Zuckerberg said. “We are at a time in history when more information is available and people are more connected than they ever have been before, and the social graph is at the center of that.”

In April 2007, Facebook reached 20 million monthly users and the global internet user base was just over 1 billion. Zuckerberg’s vision of connecting everyone was ambitious, even idealistic, in that context. It was impossible to imagine what a world connected by Facebook would look like. Over the next decade, we’d all find out.


The second idea, the news feed, was a fundamental shift in how social media sites were structured. Early social media sites, like Blogger, assumed that its users were creators, and so they focused on making it easy for users to publish their own content. Reading other people’s content meant visiting their individual blogs or using a service like an RSS reader to aggregate posts from the people you wanted to follow.

In 2006, Twitter launched as a micro-blogging service, with a design focused as much on reading your friends’ updates as posting your own. The very first version of the homepage, in July 2006, introduced the timeline — a list of updates sent by people you followed. At that time, posting to Twitter was via SMS, so the website was mainly focused on reading your friends’ updates, rather than posting content.

Twitter’s first homepage design, in July 2006.

Facebook launched its version of the timeline — the news feed — a few months later, in September 2006. As with the social graph, Facebook’s adoption of the news feed radically changed the way social media sites had worked up until then. Instead of publishing to a site that you had complete control over, your content was aggregated into a never-ending stream that was personalized for the reader, not the creator.

This shift from viewing users as producers to consumers was critical to Facebook and Twitter’s growth. Both sites gradually realized that they could be the hub of a new kind of media service, mixing updates from friends, celebrities, and brands into real-time streams of text, images, and video. Then, in 2007, the launch of the iPhone gave users the perfect tool for browsing those streams. Within just a few years, the idealistic vision of web 2.0 empowering everyone as media creators was replaced by a reality of consumers holding glass screens in their hands, scrolling through streams with their thumbs.


How the Like Gave Facebook Meaning

The history of attention metrics has been a long procession of (often accidental) innovations that end up serving the same purpose: turning audience behavior into valuable data.

Over the episodes in this series, we’ve seen how applause is not a natural response, but a culturally conditioned act that has been used in different ways to create money and power over the centuries, from Roman emperors to television sitcoms. We’ve seen how Arthur C. Nielsen pioneered big data by capturing broadcast ratings using mechanical devices and analyzing audiences using spreadsheets. And we’ve seen how the music chart was invented to sell advertising in a music magazine, but ended up driving the growth of both the record industry and postwar teenage pop culture.

The story of the like is a fascinating combination of all these stories. As with applause, the like reduces our complex responses to culture into a simple gesture. Similar to broadcast ratings, the like aggregates invisible audiences’ behaviors into analyzable data. And, like the music charts, it creates markets out of our febrile, emotional responses to popular culture.

Facebook’s introduction of the like button on February 10, 2009, was a historic day in the history of attention metrics. Not because Facebook invented the like — in fact, after launching its like button, Facebook bought FriendFeed, a service that had introduced its own like button just a few years earlier—but because, unlike other sites who had invented earlier versions of the like, Facebook added it to an already powerful engine, built on the two concepts of the social graph and the news feed.

In doing so, Facebook created the conditions for generating an unimaginably huge amount of personal data by an unimaginably huge number of people. When Facebook launched its Graph API in 2010, it gave app developers the ability to request data about not just one app user’s activities, but also the activities of all their friends. In fact, if someone wasn’t on Facebook at all, you could piece together a reasonable profile of them just by extrapolating data from people they were connected to outside of Facebook. Just by getting a few thousand people to give your app access to their Facebook data, developers could gather data on the millions of people in those users’ networks.

Facebook limited these features in an Graph API update in April 2014, but as Jonathan Albright, research director at the Tow Center for Digital Journalism, has shown, the company replaced it with something even more powerful:

Zuckerberg and Facebook’s April 2014 news release refer to the company’s change of heart about their industrial PII [Personally Identifiable Information] access feature that exposed users’ sensitive personal info and that of their extended friend and family network as a move towards “putting people first.
The exact same day in April, in bullet points underneath, the company announced their biggest tracking and ad targeting initiative to date: the Facebook Audience Network. In simple terms, this extended the company’s data profiling and ad-targeting juggernaut from its own apps and services to the rest of the internet.”
Jonathan Albright, “The Graph API: Key Points in the Facebook and Cambridge Analytica Debacle

Zuckerberg’s vision of a world connected by Facebook, announced barely two years earlier, was now about more than just connection — it was about collecting and analyzing billions of users’ emotions, intentions, and beliefs, and then selling access to this data to anyone who wanted it.

For users, this meant that Facebook became an increasingly personalized window to the world. The data produced by all of our likes were mapped across our social graphs and used to feed algorithms that decided what Facebook would show in our news feeds. The social graph and news feed created the core engine of Facebook’s growth, but the like button gave Facebook meaning and changed it from something that didn’t just connect people, but framed the way they saw the world around them.


How the Like Went Bad

Facebook’s growth over the past few years has been so fast, and so complex, that it’s almost impossible to comprehend. Right now, although the problems caused by this rapid growth are plain to see, Facebook’s potential decline is equally hard to predict.

It’s easy to blame Mark Zuckerberg for having too simplistic a vision of his creation, but as we’ve seen through this series, methods of measuring attention are palimpsests, built not in one blindingly clear moment of intent, but changing and adapting over time. The global industries that are built around these metrics are not created by one person, but by the competing needs of content creators, advertisers, investors — and audiences.

If we want to point to where Facebook went wrong, the first accusation would be that it didn’t — and probably couldn’t — have predicted the consequences of adding something so seemingly simple as a like button to a platform that already combined two exponentially powerful ideas — the social graph and the news feed.

And having built this, Facebook assumed that algorithms alone would be good enough to manage and control a platform that would end up with billions of users. Unlike Rob Manuel, who wanted a like button to make his job of curating content from his community a bit easier, Facebook has continued to insist that the company can exist only as an algorithmically curated technical platform without human curators.

Finally, Facebook broke with traditional media market models in how it sold this data to advertisers. When new media distribution networks have emerged in the past, audience measurement and monetization have been split between different companies so no single organization can control the market. In the television industry, broadcasters use ratings gathered independently by companies like Nielsen or BARB, which are jointly used and funded by their competitors.

In fact, most advertising markets base their prices on audited reach figures produced not by media platform owners but by independent research companies. With social media, Facebook (and Google, to a similar extent) combine both roles in one company—a dominating role that we now see is unhealthy for both industry and audiences.


What We Can Learn from the Like

If we want to imagine how this domination could unravel, we should not start from now, in 2018, as Facebook stands accused of ruthlessly monetizing our personal data, amplifying fake news, and even influencing government elections. We need to go back to 2007, when Facebook was still one of many web 2.0 startups. Over the next two years, it would mesh together three ideas — the social graph, the news feed, and the like — but it was almost impossible to imagine how powerful this combination would become.

We could retrospectively try to unpick that moment and break up our new digital monopolies in the same way the oil and telephone monopolies were broken up in the 20th century. We could create new public institutions to responsibly manage personal data, and we could create limits on any single company’s control and monetization of our attention. Perhaps we could even insist that no single platform should be allowed to scale beyond the point where human curation is no longer economically or logistically possible.

But even then, we wouldn’t be able to spot the next idea — the next like button — that has the potential to create a future attention monopoly. The ideas that shape our world never start big, but are created by people, like Rob Manuel, with a smaller, specific, problem. Monopolies are created through the combination of these smaller ideas, and that’s a much harder process to predict, let alone regulate.

Nearly two decades into the 21st century, we are starting to see the impact on culture and society of a number of technologies that could be called exponential—ideas that, when combined, create unimaginably fast-growing, complex systems. We can’t predict the way these technologies will change our world by focusing on specific companies or the visions of their founders.

The ideas behind these technologies are developing in too many places, and by too many people, to be able to pick specific examples that we might need to regulate. Instead, we should be looking at some of the shared concepts involved in the development of these exponential technologies and seeing if technology developers can take more responsibility for the social effects of their creations.

For example, we could have demanded commonly agreed standards for all social media companies using technologies like the social graph, news feed, and like button back in 2007. These standards could have been agreed by the companies themselves, rather than governments. This would have affected their growth, but it would also have given us more time to understand and predict the consequences — both positive and negative — of their exponential growth.

The popular saying (accredited to everyone from Voltaire to Spider-Man) goes, “With great power comes great responsibility.” The problem now is that great power can be accrued with alarming speed, but great responsibility seems to take much, much longer.