This may sound like a loser, but I believe "Let players choose their own goals and paths".
In my opinion, it's potentially a loser because players are normally not rewarded for taking such actions. Now you can do it the new way and create metagame mechanics that tell the players "You are doing it right!" with bennies and powers and narrative rewards or you can go old school and generate a mechanically-derived game world. The world is the game in this case and rewards are based upon its design.
Playing to this latter approach is called powergaming. Designing for powergaming is like designing for PacMan or Zelda. You assign gains and loses based upon actions taken in the game. Some actions gain XP accumulating into more class abilities. Some are race related and augment the character. Some are environmental, like a dragon's egg, magical sword, or castle. Some are merely procedural like a defensive strategy made up by the player, but accounted for in the game world, (think marching order).
My other bit of advice is "Don't write plots for the PCs"
There is no game I now of that says "This is the choice the player makes here, here, here, here, and here..." You're not making a game if there are no choices for the players just as you aren't making a game if those choices aren't meaningful in their consequences after they are made. Games require continuity, stories don't. So you need to track the actions taken by the Players/Characters over the course of the game. You don't want to outline those actions beforehand.
Devise the situation, devise the world, devise every single aspect of the game universe the players are not going to be doing themselves, but rather playing with. Then let your players do that. Let them be the actual actor in terms how they play the actual game. If you have to think in terms of plots, which I strongly advise not to do, think of the multiverse's behavior as the plot. That is what you need to script ahead of time as mechanics and the results of mechanics.
In my games I tell them now it's a sandbox, but I begin with a shared goal for them all to be aiming for. They don't have to follow it, but it's needed to retain the game's core choice between cooperation and competition. Don't start the party together, let the players always choose their own course, and, if they every group up wholly, it will likely take much longer than otherwise.
The games become proactive because there is an imagined world actually there for them to explore. Through their actions they discover and from those discoveries they innovate their own goals because the multiverse is stimulating their desire. Those might be fear, want, greed, friendship, love, or even simple bloodlust. I let them understand these in their own way.
The greatest challenge for me of being a DM is "Never, ever, ever-ever show or tell the players what is behind the DM's screen." Allow them to explore it. To decipher it on their own. They'll never be absolutely accurate, but they are grounding themselves like roots in the soil by doing so. The other challenge and most of what I do all session long is track everything the game has me track that the PCs do. And tell the players the consequences of their actions throughout of course.
My last bit of advice, is DMs typically don't design NPCs well. They don't design them as game constructs players can interact with. Rather DMs tend to treat NPCs as character personalities to be portrayed improvisationally. "Don't improvise if you are DMing." Keep those rules behind the screen unchanging, so the players can remain in a game. NPC behavior is generated before play just like the rest of the world. They have plots and positions and actions and knowledge and all sorts of other elements. But you will need to design all the aspects of play for NPC, which the Players perform for their PCs. So players log and remember what they've done, you do this for the NPCs. Players can have goals and plans, so can NPCs. There's tons more to this, but it's absent in most games and therefore games usually end once players start talking to someone.
