The idea of ​​a modular solution for the social network

On Habré there were many discussions of various engines for creating their own social network. However, many social networks began to support OpenSocial applications that extend the functionality of networks.
There is a logical thought - why not run social networks that perform the functions of the OS container only, and bring the rest of the functionality into the OS applications?

What will it give?

1. The development of social network functionality can be shared among many independent developers.
2. Development can be carried out in different languages ​​\ means.
3. Different applications can be run on different servers - easy functional clustering is obtained.
4. Developed OS applications can be run in other social networks that support the OS protocol.
5. It will be possible to use third-party OS applications in your network.
6. Finally, there will be an opportunity for the implementation of the universal universe of users' karma (if the rules are agreed between different networks).
7. Already there are a number of implementations of OS containers:

incubator.apache.org/shindig
drupal.org/project/opensocial-container
www.elctech.com/blog/ruby-on-rails-opensocial-container-plugin-0-0-2
')
In the case of a more efficient OS container will just go to it.

Some basic functions of the social network in the container will still have to be implemented. At a minimum, this is support for the OS protocol, registration and authorization. But all other components can be designed plug-in modules, unified among themselves, from different manufacturers.

And now I want to understand - how bad is this implementation?