Despite having a large amount of design documentation, I didn't feel I had a clear picture of how we were going to get to a finished game. Arguably of more importance, the game didn't feel fun. There's a lot of abstraction in D&D RPGs, so it's never really going to feel "viscerally" fun, but the controls and interface were frustrating. Of course, the framerate was also very low and that was a barrier to understanding other underlying problems (e.g. companion behavior).
...
Beware those who come bearing high-minded ideals of what they would do in retrospect. But, since you asked...
In terms of designing the OC, I would have gone for something a lot more open-ended, more about exploration and optional content/companions than critical path game length. I also think I would have asked for engineering to focus on revising game logic roadblocks that prevent modders from extending the ruleset. For example, revising the custom spell list data. Revisions to the resource system and the addition of a manager also would have helped a lot.
And though I know it would have resulted in a lot of resistance and probably less attention from the media, I would have pushed back as much as possible against replacing the renderer. I just don't think we had the time to do it completely and well. It also really screwed up all of our GUI code and changed all of the art asset pipelines throughout production.
But again, this is all in retrospect.
...
I think the game as released is a high 7, low 8 title. To be honest, the major issues are due to a lack of polish. There's certainly a lot of stuff in the game, but none of it really looks or feels great. At best, the controls and features feel good, at worst they feel terrible.
For example, the camera. Programming got the camera in and fixed a number of its glaring problems, but for all the various camera modes you could run in, it was hard to find one that felt good. The toolset is also an appropriate example: highly functional, very powerful, not enjoyable or fast to use for many tasks.
The biggest problems during development were an unrealistic scope and a lack of focus on quality/fun from the beginning. It's arguable that the former resulted in the latter. With D&D games, it's easy to become consumed by the idea of adding every feat, class, and race you can find in various books.
