Heya! Well, our architecture is very Flask-like, but yes it does not use Flask. There's kind of "historical reasons" for this; at the time that I started MediaGoblin Flask was fairly new, and we started out using MongoDB (so we couldn't make use of some of the nicer Django features anyway). I also really strongly prefer Jinja2 templates over Django's because of the ability to pass in arguments to functions, so eventually Django didn't make sense. But the real root of it though is that it's not actually as much work as people think to hook together a WSGI application that uses the libraries you already know you want... not really that much harder than using Flask anyway! http://docs.webob.org/en/latest/do-it-yourself.html is a good read in that vein.

Will MediaGoblin stick with its own direction, or move over to using Django or become a Flask app? It kind of depends... at the moment, things seem to be running well, and aside for waiting for some of our libraries to catch up to Python 3, we don't seem very hampered by it, and we're also very nimble... changing core parts of MediaGoblin to fit our needs is presently very easy. It also might be interesting to see what happens if eventually we end up adopting some XUDD technologies: https://xudd.readthedocs.org/en/latest/ but admittedly that's a ways out (though MediaGoblin already can run on XUDD, as of yesterday ;)) if that will happen at all.

Sorry, kind of a rambly response! Hope it was interesting.