That appears to definitely be a link to the wrong thread - it's about issues with Emma's companions. Anyway, the advantage of that over extending OSG is that we don't have to meddle with components of other projects, just our own.
This one?
viewtopic.php?f=2&t=3724
Anyway, your assumption means that people will want to develop yet another format and that people will want to use it. If it worked so great with 'components of other projects', then why would it be any different for us? Until someone does this, it is still a bunch of opinion and no action which means openmw is still at the same place with the 'components of other projects'.
Put your money where your mouth is and actually develop said format.
This also includes tools to import/export in other software such as Blender. Going at it alone is an option, I just fear that it will be too much for anyone considering the the lack of interest (or time or knowledge) from other developers.
Scrawl had (has?) plans for extending the .OSG format to include animations. However this requires additions in the osgexport plugin for blender support. This also means updating the OSG plugins for Maya3d and 3DSMax as well.
viewtopic.php?f=20&t=3336&p=37125#p37125
I have been working behind the scenes with the Niftool's crew to bring the nif_blender_plugin up to speed for Blender 2.77 and together with Greatness7 for animation support. At this point NIF is pretty much mapped out up to FO3 and Skyrim, the extra features just need to be implemented in OpenMW (easier said than done but cc9cii has made progress with NIF features from Oblivion in OpenMW) .
So using an open standard (at this point NIF can be considered open with the amount of time and reverse engineering put into it) that has more than one project using it is to our advantage since it will likely be better supported in other bits of software.
So the options are:
NIF (works now in OpenMW)
OSGextended (needs work)
OpenGEX (needs work)
Collada (needs work)
self-rolled (start-from-scratch)