I just managed to get Tom Francis' latest video watched. What I found very interesting was the end, where he discusses that the best way to handle a "main objective" is to simply pepper what the player is already doing with things that get the main objective done. He specifically brings up how cool of an idea it'd be in Oblivion and Skyrim -- rather than having a main storyline that distracts you from the fun things you can do, the fun things you can do give you bits and pieces of hints as to how to complete the main quest. Needless to say, this was particularly interesting to me because it was someone who really knows his stuff saying what a good idea one of my goals for TD is: make the actions the player chooses to do propel the narrative, rather than giving the player a narrative that they must follow, regardless of what they'd otherwise choose to do.
Don't get me wrong, I'm sure player agency (or an authentic feeling illusion thereof) is very hard. It's almost downright impossible if you live in the frankly dangerous place that game developers tend to exist in right now. At present, game developers are chasing cinematic experiences. They want to make movies that you play (the recent Metal Gear and Kingdom Hearts games are the exemplars of this). Compelling narratives with complex characters and beautiful cut scenes. These are not bad things. But there is an inherent danger in cinematic experiences. Ultimately, cinema (movies) is about the "director's vision." Games are not supposed to be about the director's vision, they are supposed to be about the player's experience. That's a wholly different goal. It's why people who are much smarter and more experienced than I always say that unlike movies, game makers should concentrate on "do, not show." This fascination with cinematic experiences has other side effects. Since the main story is the same, wonderfully told thrill ride every time the player plays, developers have to add achievements, unlockables, and New Game+'s to get replayability. If, instead, the narrative derives from the gameplay, then the player will replay the game on their own to experience how the narrative changes in response to their choices. Even the most simplistic of these systems; the good/evil metrics of Fable or paragon/renegade choices of Mass Effect or the Light side/Dark side of Knights of the Old Republic, will generate in many players the desire to beat the game one way and then play it again to see how the other side lives. If it wasn't so personally difficult for me to play evil characters, I'd certainly be one of those individuals. By giving the player agency, you've effectively increased replayability.
Of course, to give players agency requires putting in a bunch of extra work. You have to build out all the different options that a player might choose, knowing full well that most players won't see most of this content. Building content that players won't see (or might not, which is pretty much the same thing as "won't" to a creator) is creatively exhausting and expensive in terms of time and budget. It's easy to see why a developer might rather spend that time and money working on cooler cutscenes that a player will definitely get (read: "have") to watch. But I feel that the effort is worth the work to make an experience that's really special for that particular player. The best stories that players tell each other are the ones that they shaped. Players can be impressed by a narrative invented by someone else, but the narratives that they helped shape themselves live on in their imaginations forever.