I think I've learned a lot from every project I've been on, but I guess F:NV was the most illuminating. It had the longest development cycle of any game I've shipped and, by far, sold the most units on the most platforms. I think that wider-than-usual net helped show how much people "got" or didn't get certain types of gameplay mechanics, narrative structures, etc.
I also had more freedom to mess with the rules on F:NV than I did on any of the D&D games I contributed to, and seeing how people responded to the minor and major shifts between F3 and F:NV was very informative.
I think the biggest thing I took away, more definitively than ever, is that how things actually work matters more than how people react to the idea of how they will work. I.e., there are really two levels of response to something in the game: the idea of what it is (often interpreted outside of the game) and the reality of what it is. The idea is often more upsetting or disconcerting to people than the reality. But the bottom line is that the reality actually has to be enjoyable in the context of the game, regardless of where the idea started or what the intent was.
When we eliminated Big Guns and spread the weapons around to other skills, there was a lot of head-shaking. After the game came out, not many people complained about or or really even seemed to care. That isn't to say that NO ONE cared -- some people cared, and still care, a lot about it. But the end result didn't generate a lot of negativity and most people responded positively to it or just didn't care.
On the other hand, the way the map was illustrated was logical and map-like but confused people because they thought it was literally at the same scale as the F3 map. In F3, the map border *is* the border of the world. In F:NV, the map border is the extreme outer extent that encapsulates the irregular border of the world. Essentially it was like forcing Colorado and Nevada to be fit into an identical square frame map that's 10" by 10". Nevada is larger than Colorado in reality, but it is always going to take up less space if pushed to the edges of a 10" by 10" map because Colorado has a rectilinear shape and Nevada doesn't. Long story short: it makes sense, but it confused a huge number of people who thought that we were wasting portions of the map. We addressed this in Honest Hearts by using an irregular border instead of a square one.
And speaking of Honest Hearts, I also learned that between the free-wheeling nature of F:NV's content implementation and the strict, low-risk implementation of Honest Hearts content, OEI content usually needs to fall somewhere in-between. A quest that is completely cut-and-dry bog-standard will usually come across that way even if it takes place in a new setting. A quest that is a tangled skein of nightmare scripting will probably ship as a broken mess of half-fulfilled dreams. So when it comes to working with designers, it's good to start with a really solid, stable core of gameplay but leave time for (and encourage) more risky secondary elements after the core has been developed.
Having content that's just "in" and works isn't enough -- both for designers and for people who are playing an Obsidian RPG. People enjoy weird and wacky stuff in quests; it just has to work properly. Our concepting, design, and review processes need to account for the basics but also ensure there's time for the cool and unusual stuff.
