3
Ahem, sorry for the really delayed reply, some stuff in life came up (as always seems the case).
The major benefit to this is integration with existing technology. I am a huge proponent of xoops-designed modules, and often take NewBB over their counterparts. I am almost greatly opposed to hacking core files just to integrate XOOPS with other technologies. Some web applications provide, however, an XML-RPC interface that could be called from XOOPS to submit or retrieve information. The API for it rarely changes, meaning that the same interface usually works from revisionn to revision. I use movable type as an example only because I am somewhat familiar with the posting API. The major bonus of the "module driver" idea would be that XOOPS could for example, post a blog to Movable Type through MT's api interface, and retrieve records using the same module driver.
Given my current position (and new job) it turns out that sometimes the biggest drawbacks to using XOOPS in a business environment is the inability to get it to work with existing technology xyz without having to try and migrate everything. Plus, by writing a module that interfaces with the API, there isn't any core code rewriting going on.
(The springboard for this of course is to research XOOPS own RPC interface if it has one, and see how to for example, make posts by email, have a simple application manage your news postings, update modules, etc)