Pros and Cons of Today’s Facebook Announcements
Earlier today, Facebook held a developer conference called f8 and took the opportunity to announce a number of new features that impact both developers and average users. I’ve assembled a non-exhaustive list of several important changes the company described, along with a summary of each change and a quick pro/con evaluation from my perspective. I’ll be looking at these and other new features in-depth over the next several days.
The Open Graph
While Facebook has often talked about how its users friend relationships form a “social graph,” the company is now focused on creating a broader “open graph.” This is essentially a map of connections between people, companies, products, websites, and so on. When you list your interests and tastes on your profile, you’re helping build this structured database of links.
- In many ways, this idea echoes the vision of a “Semantic Web” that others have outlined in the past. In fact, World Wide Web creator Tim Berners-Lee has long called for building a similar structure.
- Facebook’s implementation includes simple ways for sites to add usable information about them, and they’ve built a simple interface for accessing data on pieces in the graph.
- While this graph may be “open” for contribution and access, it’s definitely controlled by Facebook alone. That setup has obvious business, political, and philosophical implications, but centralized administration of such a graph has technical trade-offs as well, such as dependence on a single point of failure.
- Facebook’s new version of the Semantic Web still carries many of the same issues as older versions, such as major privacy concerns, data poisoning, and data inconsistencies.
Universal Social Experience
In today’s keynote, Facebook CEO Mark Zuckerberg often talked about the high-level goal of enabling social experiences for users across the entire web. By combining the latest features Facebook offers, any site can bring identity and relationships into its own ecosystem.
- Much of the information that you encounter on sites today is generic and requires that you spend time sorting or searching to make the site more relevant. With data from your part of the open graph, sites could customize and optimize in a way that’s tailor made for you, providing more relevant content right away.
- This approach greatly reduces friction on other sites as well, since you won’t have to go through the tiresome process of setting up a new account, remembering another password, and trying to find people to connect with or useful content.
- One person’s feature is another person’s privacy violation. However well-intentioned other sites may be, their “social experiences” can fail to recognize the value of anonymity or take into account a rightful degree of user control.
- As others have pointed out previously, since this type of optimization often centers around your establish relationships, it can create an echo chamber effect and further isolate socioeconomic or ideological groups from each other.
This is the marketing term for a feature Facebook first earlier this year. The company has partnered with certain “pre-approved” websites that can now automatically identify a Facebook user at their first visit. The sites can also access what Facebook classifies as publicly available information.
- This is a more specific example of Facebook’s vision for social experiences reducing friction. The feature is aptly named “instant,” as it basically sets up a user’s account on another site without any interaction, a behavior some may find very convenient.
- From a privacy standpoint, Facebook has included a global opt-out under users’ application privacy settings, and clearly indicates when this sort of automatic authentication takes place with a banner at the top of the site.
- The feature still raises a number of privacy concerns, and essentially repeats several of Google’s well-documented mistakes with the launch of Buzz. And while a full opt-out does exist, users are opted in by default. This personalization will likely be the source of many surprises and violated expectations.
- Facebook controls who has access to the setup, and currently it’s not entirely clear how sites can become pre-approved or how much the program will expand in the future. The privacy controls also lack some clarity, as the opt-out does not cover information shared by friends who use instantly personalized sites.
Any web site now has access to a range of simple tools that add Facebook features, such as “liking” a page and publishing approved stories to a user’s news feed. These widgets also replace some of the options previously offered to developers under Facebook Connect.
- Facebook has built these plugins with ease of deployment in mind, and they drastically reduce the complexity of integrating with the service. Many developers will be pleased with the simplicity of these functions.
- From a security perspective, Facebook’s approach also sets up a barrier between the external site and Facebook content the users sees. While the like buttons and friend pictures may seem to be simply part of the page, they actually reside in a separate data space from the rest of the page’s content until you choose to authorize access for the other site. This helps protect both the developer and you as a Facebook user.
- In practice, the deceptive appearance just described may mislead many users into thinking that Facebook is exchanging far more data with other websites than they actually are. This will likely lead to some unwarranted panic.
- These plugins do rely in many ways on developers providing accurate data, and it’s likely we’ll see these features abused by scam artists and distributors of malware. Currently, the plugins seem to lack certain authentications that may lead to unintended consequences.
As part of a more streamlined development experience, Facebook has launched a technology called OAuth 2.0 for authenticating applications and websites. This replaces the proprietary model the site had been using and should once again simplify building Facebook-enhanced services.
- This is a major validation for an open standard many companies have helped put together. Many developers will be encouraged to see Facebook choosing OAuth over a proprietary system.
- As already mentioned, this is another way that Facebook has simplified application development. OAuth should reduce confusion over how other sites can access Facebook information.
- While perhaps not a completely fair point, I’ll note that the use of OAuth does not diminish the threat of application-based attacks through vulnerabilities known as XSS and CSRF.
- A number of other sites, such as Twitter, have used OAuth for some time, but this is a major roll-out of a very new version. We may see new security issues related to Facebook’s implementation.
At f8, Facebook expanded on their plans to offer a virtual currency system for application payments. Several applications are already using Facebook Credits, but we’ll likely see far more implementations in the near future.
- Yet again, this system helps reduce friction. For developers, Facebook offers a simple way to include payments without having to worry about a number of implementation details.
- Also, for users, virtual currency can reduce the hassle of worrying about issues such as international currency conversion.
- Since Facebook is already facing widespread criticism over privacy issues, some users may hesitate to add credit card information to their Facebook profiles, even if it can only be accessed by Facebook.
- This service makes Facebook a middleman in potentially millions of dollars of transactions, and could raise liability issues.
Granular Data Access
Though perhaps overlooked, Facebook made good on their promise to include more granular permissions when applications request user information. This feature comes in response to concerns raised by Canada’s Privacy Commissioner last fall. With the new setup, applications will have to individually request private profile fields when a user chooses to authorize.
- This change will immediately provide more transparency and accountability, since users will see listed out exactly what fields an application will want access to when they authorize.
- Many users may simply click through anyway, but the new system may raise awareness for many users who did not previously understand the range of information applications could access. Seeing a greedy list of data fields may give users pause.
- Since announcing granular access last fall, Facebook has radically changed the definition of what constitutes “private” information. Consequently, many of the fields that might have been included in this setup are now considered “public” and thus generally outside access controls.
- While commendable, this change may not lead to any substantial changes in practice. The model relies on developers limiting their requests, and many users will probably still want access to applications that ask for all information.
Persistent Data Storage
Until this week, applications and Facebook-enabled websites could not store most information accessed via the Facebook API beyond 24 hours. Now, Facebook has removed this time limit, meaning developers can save user data for as long as they want.
- This change will significantly reduce overhead for both developers and Facebook, since applications will no longer have to exchange data with the service each day a user connects.
- Users will likely see some performance gains from applications, since they can cache data locally rather than constantly checking with Facebook before rendering content.
- Facebook applications will now be far more valuable targets for attackers. If a popular application suffers a database compromise, millions of users’ private information could be put at risk. Hacking Facebook directly tends to be difficult, but many applications lack the same level of security.
- This increases opportunities for behavioral targeting and visitor tracking, since third-party developers will now be able to maintain complete archives of profile information.