Even the best PIM infrastructure is quite useless without any data. It becomes viable once it has access to the usual data sources, e.g. vcard files, maildir directories, etc. It becomes exciting when in gains access to source previously not or not easily available to the user.
Various PIM frameworks try to address this through some form of extension mechanism, e.g. plugins. Akonadi addresses it through the use of agents, autonomous helper processes, called Akonadi resources.
Writing such a resource can be considerably easier than writing a plugin, mainly because a fault in a resource under development won't take down the whole system, can be debugged without the symbol overhead of other backend implementations.
The complexity of dealing with the actual data source is basically the determining factor of the overall complexity of the resources. In fact, creating one for a simple to access and/or limited source is dead easy.
This will be demonstrated live and uncensored!