Learning from the Facebook Mini-Feed Disaster

by Jared M. Spool

Imagine this scenario: Your team works hard on a new feature—something that you think will bring your design to a completely new level. Launch day arrives and you wait with anticipation to hear the rave reviews of the millions of existing users.

However, instead of kudos, you immediately hear complaints. The users are upset. They are so upset that they band together and start a petition to have you remove the new feature. The petition itself gains attention as the list of names grows and, suddenly, major newspapers are reporting how users are complaining about this feature you thought was so great. All of this happens in just a few days.

This is exactly the scenario that Facebook, the social networking site, found itself in a few months ago. They created a grand feature: the mini feed. The feature is neat: instead of requiring users to visit the pages of all their friends, the feed reports on changes immediately. It’s an immediate way to keep track of what’s up with your friends.

Unfortunately, this cool feature wasn’t well received at all. Upon its initial rollout, users suddenly found their familiar home page replaced with a listing of random people doing random things. Many people initially assumed their own actions were now transmitted to people they didn’t know.

As is common on Facebook, several users started groups to protest the new feature. Ironically, it was the mini feed itself that announced that friends were signing up for the protest groups. As more people joined the groups, an increasing number of users became aware of the protests. Within 24 hours, one protest group had grown to 750,000 members.

The Facebook team was caught by surprise by the sudden, massive backlash to their design. They posted a blog entry that said, “Calm down, breathe. We hear you.” Unfortunately, their reaction added flames to the fire, making the protesters only angrier about the sudden changes to the site.

Bloggers picked up the story, and then the mainstream media started reporting it. Within days, the New York Times and the Wall Street Journal both had prominently placed stories talking about how hundreds of thousands of Facebook users were revolting against the new features.

What can we learn from what happened? What did the Facebook team do to upset everyone so radically? What could they have done to prevent the backlash? By reconstructing the sequence of events, we can try to understand what happens when things like this go wrong and learn how we could avoid similar results with our own designs.

The Shock of Sudden Change

In the old days, users could control when their software changed. An upgrade was suddenly available, but the acquisition and installation of that upgrade was under their control. (Or under the control of a local representative, such as a company system administrator.)

However, in the world of web-delivered content, the content provider now controls when the change occurs. Incremental changes to a design are now commonplace, with small changes happening frequently and often without warning.

The developers of Facebook have changed the Facebook design dozens of times in the past. Nobody got terribly upset. Nobody formed protest groups. Why should this time be any different?

Unbeknownst to the development team, they had walked into a minefield. With one new feature, they had changed a fundamental element of the site: what it means to be a friend. In addition, they hadn’t carefully explained the implications of the change, causing users to extrapolate incorrectly the design’s behaviors. It was these two things that formed a ‘perfect storm’ condition, making the system ripe for the problems to follow.

Keep Your Friends Close and Everyone Else Closer

From its inception, a core feature of Facebook was adding friends. When you add a friend, Facebook recognizes your relationship. When you ‘friend’ another user, you have access to their photos, status messages, and other capabilities of the system, and they have access to yours. The power of Facebook is in the friend network each user sets up—the more friends they have, the more powerful their network.

This increasing power led many users to freely add people as friends, even if they weren’t friends in the offline world. They’d meet someone at a party or bump into them at Starbucks, discover they were both Facebook users and agree to add each other into their network.

Soon, many users were competing on the number of friends in their networks. People had hundreds of friends, many of whom they didn’t really know. Because it was easy to ignore people you didn’t know well—you never visited their pages—the cost of having a huge friend network was minimal, while the reward—claiming large popularity—was great.

The introduction of the mini feed changed these costs and rewards. All of a sudden, anything a friend did was presented in the feed. Users with large networks saw listings from people whose names they didn’t recognize. Their feeds were full of “news” about people they didn’t care about.

Users saw this as superfluous information, cluttering up what used to be a perfectly well organized home page. This was the first problem that started the backlash.

All of a sudden, it mattered who was in your network. Having dozens of people who you could previously quickly forget about now took up your attention. However, that wasn’t all.

An Invasion of Privacy

Given incomplete information, people sometimes jump to incorrect conclusions. This is exactly what happened when users found dozens of folks in their feed, many of whom they didn’t recognize.

Many users assumed the feed was informing them of random users of the site. Given that assumption, they jumped to the next obvious thought: their own activity on the site was being broadcast to random other users. With these two assumptions in hand, many users decided the feed was an invasion in privacy.

Of course, these assumptions weren’t true. The feed only reported activity of people in the user’s network. In fact, there wasn’t any new information being presented in the feed. The only difference was, before the feed, you had to visit each friend’s own page to discover the information now being summarized in your feed. No invasion of privacy was occurring—just an increase in the visibility of already available information.

The designers assumed most users only added people into their network who were truly friends and people they’d recognize. It surprised them to discover people had added dozens of mild acquaintances and didn’t recognize everyone they’d added.

Users in the protest groups immediately demanded options to control the privacy of the system. Interestingly, the privacy settings already existed in Facebook’s core functionality. However, the default settings were to share all information. To make things more challenging, the privacy settings weren’t easy to find.

Preventing the Perfect Storm

What could the Facebook Team have done differently? Was there anything they could do to prevent the sequence of events that forced them into damage-control mode after the national coverage in the mainstream press?

Of course, it’s easy to play Monday-morning quarterback and second-guess the decisions made by others. However, we do know some things that likely would have worked better.

First, while the team had an existing deep relationship with their users, it seems clear they weren’t aware of how these users were using the system, particularly when it came to building out their friend network.

Spending more time watching how people added friends into the network would probably have helped. Friends are a core element of the Facebook experience. The more the team knows about how users build up their network, the more they’ll embed that knowledge into their designs.

Second, a staged rollout of the new functionality would likely have helped gauge the users’ initial reactions. The site has millions of users, yet a rollout to a few hundred, and the appropriate feedback collection mechanisms, could have easily predicted users would have misconceptions with privacy functions.

Third, it’s clear, when dealing with social dynamics, users need a sense of control over how the system disseminates any personal information. While there is always a tension between keeping an interface simple and allowing for control over functionality, social systems seem to demand more on the control side. The defaults can still tend to be open, but users need easily accessible methods to restrict the visibility of content they want to remain private.

Finally, designing for embraceable change, in addition to designing the functionality itself, would likely have helped with the adoption of the new functionality. Through an iterative rollout process, the team could have learned the information users needed to correctly understand the mini feed and its implications.

Just designing new features isn’t enough. We need to design the process by which users will learn to adopt those features. One strategy is to use an opt-in preview, where the site offers the new functionality as a sneak peak, allowing users to start using it on their own schedule. (Currently, Yahoo! Finance is using this approach for their new chart capabilities.)

Another strategy is to make sure the users are clear on the implications behind the change. Had the Facebook team explained at the outset how the people in the feed were only established friends, that personal information was not broadcast beyond the existing friend network, and how it was easy to change the privacy settings, much of the brouhaha could have been avoided. Clear messaging or a guided tour of the new feature is a way to communicate these concepts.

Epilogue: All is Well That Ends Well, Right?

In the months since the mini feed was introduced, the users’ initial reactions have since gone the other way. The mini feed is now a critical part of the site’s functionality, allowing users an easy way to keep tabs on their network.

One could say the initial ruckus was just an over-reaction from users resistant to change and, now that they have become accustomed to the power of the new features, they are happy for the new capabilities. In the long run, the controversy wasn’t really an issue.

However, I’m betting the Facebook team would prefer to never have to live through that period again. And I’d be willing to bet most other development teams would prefer to avoid massive petitions and nationwide mainstream press coverage in the same way. Nobody wants a backlash. So, while things have worked out well for the Facebook team, learning from what they went through is important for all of us involved in designing excellent user experiences.

Published here on July 16, 2007.