The Google Reader Debate: What is a “friend”? What is “public”? What is “privacy”?

Over the holidays, a big debate erupted over a small change made to Google Reader, its web-based feed aggregator. For a good while now, Reader has included two different options for sharing the content that you read with others. First, the user can enable sharing features for tags that he or she has applied to different posts or subscriptions, creating a public webpage and a public RSS feed for those items. The second sharing option is to click “Share” under individual posts. Similarly, this sends the post to a public webpage and a public RSS feed. For both of these options, the user generally has to choose to share the URL of the public webpage or feed with their friends.

Several weeks ago, Google made the next step towards a more social integration of their various communication products. After determining which of the contacts in a user’s Gmail contact list had been contacted via Google Talk (in Gmail), Google would insert all of the items a user had marked as “shared” into those contacts’ Readers. Essentially, Google was inferring that a contact with whom a user had spoken to via Google Talk was one degree more “intimate” of a contact… and they decided that this degree of intimacy was enough to apply the term “friendship” (as understood in the context of social networking, at least).

In the opinion of this blogger, the outcry expressed by users of Google Reader over this new feature of the product is simultaneously warranted and absurd. It is warranted because Google did not present its users with an opt-in option, because it does not give users enough control over the updated sharing feature, and because it misconstrues frequently-contacted-contacts with “friends.” It is absurd because critics and naysayers are saying that Google’s move amounts to a breach of privacy because it renders public what was once private.

A user’s Gmail account saves every single email address from which email is received, and every single email address to which email is sent. There is an option within the settings of Gmail to automatically make frequently contacted email addresses available for chatting via Google Talk. Speak to one of those contacts once on Google Talk, and that person is exposed to the shared items from one’s Google Reader. Let’s say that it was the contact, not the user-in-question, who initiated the chat. If that were the case, then not once in this process has the user-in-question explicitly asked that Google increase the intimacy with that contact, nor been asked by Google whether it approves of an increased level of intimacy.

Indeed, this is where Google’s confidence in its strictly computer- and algorithm-based way of doing everything meets a wall. Computers are not (yet) able to create a social graph. Facebook claims to understand this. Facebook claims that it only organizes what already exists in the real world. That’s why it is based around networks like schools, businesses, and regions. Further, Facebook requires confirmation of friendship even if its algorithms are able to determine associations. For example, Facebook surely has the capability of seeing that two students at Vanderbilt University are both in the same four classes this semester, that they live in the same dorm, and are both from Little Rock, Arkansas. But Facebook does NOT automatically associate these two people on its network as friends. Yes, Facebook could feasibly message those users and say “Hey, we noticed you and X have A, B, C, and D in common… wanna connect as friends on Facebook?” But it still would need to give them a choice. It goes against the spirit of digitizing the social graph (and not socially graphinh the digital) to automatically formalize that relationship without user input on either end.

If Google is going to compete with Facebook, it needs to understand the source of Facebook’s success. And that success has largely been based on trust. Users trust Facebook to give them a choice when changing the dynamic of their service, both a choice to participate and a choice in the level of participation. Facebook stumbled in this regard with the rolling out of both its News Feed and Beacon features, and retroactively introduced controls that gave the power back to the user. Google needs to follow suit if its going to proceed a single step further in the social integration of its products.

While Google definitely needs to take some responsibility, so do users in the language they use in this debate. Google’s recent move has next to nothing to do with privacy. If users bothered to read the documentation behind the sharing feature pre-upgrade, they would not see the word “private” anywhere. Indeed, Google made clear that while the Shared Items webpage and feed were given convoluted URLs in order to protect the user’s identity from being publicly associated with their shared items, they were not encrypted or otherwise hidden. Sharing an item made it public, in the sense that if someone were to very randomly type in the address of the Shared Items page, they could find it

So there was never anything truly “private” about Shared Items in Google Reader. No, it was not easy for strangers to find or stumble upon them… but neither did Google explicitly say that they were going to protect that information. They gave users a little protection with the unusual URL, but they never guaranteed that this information would be given the same protections as their email, documents, or search history. Sharing has always been considered a public act by Google.

Just as Spokeo, and just as Facebook did with Beacon and with News Feed, Google’s double response has been 1) “we thought you would love this” and 2) “we are just making things that were already publicly available easier to find”. Google needs to remember that not everyone that uses its products are nerds or fanatics of integration. For myself, I think what Google is doing is great. But you have to prepare people for this sort of thing. That is not usually how Google operates – they plan in secret and roll out in a shock-and-awe manner. This may work for some things, but as far as the dynamics of their social features go, it will not. People have to tell Google whether they love a feature or not… Google can’t assume it, and its computers can’t compute it. The term “social” implies intercourse, exchange, and consent. That’s a feature of human nature that computers will never — or not very soon, anyway — be able to read. In turn, users must realize that unless explicitly protected in the Terms of Use or explicitly guaranteed with the use of the words “private”, “protection”, and “never” — what they put online is able to be found and publicized.

However it turns out, this debate is a great example of the realizations and concessions that both businesses and users are going to have to make in this nascent digital age.

Image used under a Creative Commons license from Flickr user dannysullivan.