The Evolution of Privacy on Facebook

Switch to the animated version (won't work in IE).

About

Facebook is a great service. I have a profile, and so does nearly everyone I know under the age of 60.

However, Facebook hasn't always managed its users' data well. In the beginning, it restricted the visibility of a user's personal information to just their friends and their "network" (college or school). Over the past couple of years, the default privacy settings for a Facebook user's personal information have become more and more permissive. They've also changed how your personal information is classified several times, sometimes in a manner that has been confusing for their users. This has largely been part of Facebook's effort to correlate, publish, and monetize their social graph: a massive database of entities and links that covers everything from where you live to the movies you like and the people you trust.

This blog post by Kurt Opsahl at the the EFF gives a brief timeline of Facebook's Terms of Service changes through April of 2010. It's a great overview, but I was a little disappointed it wasn't an actualtimeline: hence my initial inspiration for this infographic.

Data

The data for this chart was derived from my interpretation of the Facebook Terms of Service over the years, along with my personal memories of the default privacy settings for different classes of personal data. The population sizes are statistics from Google, the Facebook Data Team, and wild guesses based on what seemed reasonable to me.

I welcome data corrections, so please leave a comment below if you have better numbers to share.

Types of Personal Data

Facebook's classification system for personal data has changed significantly over the years. I tried to capture what I thought were broad topics that have remained relatively consistent. But they might need some explanation.

Likes: a person, band, movie, web page, or any other entity represented in Facebook's social graph that has a "like" button. "Likes" started with status updates, but have now grown to encompass pretty much everything. In Facebook Newspeak, they're a "Connection".
Name, Picture: self-explanatory
Demographics: Your age, sex, birthday, and other basics
Extended Profile Data: Your family members, city, place of birth, religious views, favorite authors, schools attended -- anything that is an entity you can list a relationship to in your profile.
Friends: The people you've friended
Networks: The personal networks you've set up on Facebook (e.g. colleges & universities or companies).
Wall posts & Photos: Self-explanatory.

Audiences

Audience sizes are based on averages and my own extrapolations from those averages.

The meaning of "network" has changed a lot on Facebook. For this infographic, I've interpreted to be an average network size, plus (later in the timeline) all "friends of friends"

Implementation

The audience scale is logarithmic, so that we can compare audience sizes of 100 and 1 billion. I also did a big no-no and mapped the audience size to the length of the slice, not its area. I don't feel too terrible about this, because the area comparison is already distorted by the log scale. Plus, frankly, the linear scale just looks better.

I built this sketch using Processing.js. You're welcome to download the source. Sorry, no Internet Explorer.