This morning Facebook was awarded a patent for the “resource management of social network applications”. The patent describes a method in which resources are allocated to applications based on the affinity toward those applications expressed through various forms of interactions. While components of the patent have been made irrelevant by upgrades to the Facebook Platform, certain sections of the patent may remain relevant.

Measuring A User’s Affinity Toward An Application

One core component of the patent is the measuring of a user’s affinity toward an application. According to the patent “User affinity is based on measuring positive and negative interactions”. This includes various forms of interactions within the “notification channel, invitation channel, mini-feed channel, and wall post channel”. There are three types of feedback (all of which are either implicit or explicit): negative, neutral, and positive.

For example, an application can provide a notification to the member on the member’s web page. If the member ignores the notification when viewing the web page, e.g., does not click on or otherwise interact with the application, this is an implicit feedback that can be interpreted to mean that the member is neutral towards the application. But, if the member clicks on a link in the notification that unsubscribes the member from further notifications from this application, the member has provided explicit negative feedback towards the application. On the other hand, if the member clicks on a link in the notification that provides the member with more information about the application, the member has provided an explicit positive feedback about the application.

While Facebook has been relatively clear about how invitation and notification allocations have been made for applications, this patent provides us with a little more insight. While application notifications will be removed from the Facebook Platform on Monday, there are pieces of this patent which still remain relevant: primarily implicit and explicit feedback through a user’s wall.

Forced Invites Receive Patent Love

An interesting image included in the patent is a flowchart which illustrates the forced invited process, that drove millions of users to applications prior to being banned in February of 2008. While there is no explanation of relevance, aside from a description of the drawings, it’s interesting to see how much this patent pertains to functionality within the Facebook Platform.

Channel Allocations Based On User Affinity

“Buckets”, a term which has become common among Facebook application developers, are a system for essentially controlling the viral channels that developers have access to. For example, some applications are allowed to send 40 application requests per user, per day, whereas others are only granted 20. This number is determined by a number of factors, but as this patent illustrates, a large portion of that is not just application ratings, but other user affinity statistics (highlighted earlier).

An image included in the patent also illustrates how allocations are determined by the “rate of action based on number of impressions in a given time interval” across all applications. In other words: all applications are ranked against each other and then allocations are based on that. While Facebook doesn’t appear to go into more detail about how the “number of messages allowed for each application” is calculated, we would assume that buckets fall into a normal distribution curve.

All Interactions Are Measured

The primary conclusion from reading this patent is that Facebook is measuring all application interactions within the Facebook platform. Whether that’s news feed interactions, application ratings, application request responses, notification responses, and more. While much of the features of this patent have been changed since filing (for example there’s no more “mini-feed” as it has been combined with the “wall”), there is plenty of insight into the inner workings of the Facebook Platform provided by this filing.

Thanks to GoRumors for tipping us off.

Update
As a commenter has noted, this patent has not been approved yet. Instead, the application was only recently published. Whether or not it will be awarded is still unknown.