I've been thinking about intrinsic goals, and how to create them.
Most games have goals that are stated. Kill the monster, shoot the alien, save the princess. But these games also have goals that aren't stated.
For example: you have some time, gold, and XP before you get to the monster's castle - how will you spend them? The fundamental rules about how your character interacts with the world determine the kinds of things you'll consider doing.
Grinding a bit to get the super-sword. Learning to use the fire-ice combo to kill hordes. Searching for the hidden nook that might have a treasure in it...
These are unstated goals. The game doesn't simply say "you should grind!" Instead, it simply allows you to fight while traveling from one place to another. The player quickly realizes that he can travel as much as he likes in order to get into those fights. The fundamental construction of the world allows the player to figure this out and do as much or as little as she likes.
However, that's a baby's toy - the tiniest of potentials, just enough to let the player feel like they aren't bored. An example of taking this to an extreme would be - obviously - Kerbal. Yes, I'm still talking about Kerbal.
Kerbal has no stated goals, and you really don't gain anything like experience points or such. However, there are a lot of goals built right into the world. Or, rather, the world exists to allow the player to create their own goals.
The most obvious things are the planets and moons. Each of them is a target you may want to reach. None of them have any particular reward for reaching them - they're just various sizes and atmospheres of rock. However, they have very unique physical characteristics, so reaching each is a distinct challenge.
The planets and moons are just the quick in, though - the obvious foothold. It takes almost no time to discover that the real power behind Kerbal's longevity is the physics of the universe. The planets do follow those physics, so you could say that the planets are simply specific instances, almost simply samples of how the physics can work. The core draw is the physics.
More specifically, movement.
The movement rules in Kerbal, along with the several kinds of visualizations used to help you plan your movement out, make moving an interesting challenge.
You can talk about splitting it up - launching, achieving orbit, transferring orbit, landing, docking, flying in atmosphere, supersonic flying, and so on. But they all use the same few core mechanics: position, velocity, gravity, drag. The existence of planets, especially your starting planet, provide a fun set of bumps and gradients to use, mixing the mechanics up and also prodding you into filling in your own instances. While you can't launch objects that create gravitic or atmospheric bumps, you can launch objects that have cyclic position and velocity, meaning you can use them as waypoints. Dock with them.
Anyway, this gives a very powerful set of implicit goals. Not simply because there are obvious implicit goals like in an RPG, but because the act of aiming for the obvious implicit goals makes you realize that your actions can create more implicit goals in a constructive manner. It's less like an RPG, and more like if every RPG party you sent out into the world remained there, ready and waiting for you to team up with them using another party you create later, or can be slotted into towns to make the town reshape the landscape...
Thinking about this kind of constructive implicit goals, I've created a variety of theoretical options. One such example is the RPG system briefly mentioned above, but nearly any kind of genre or theme could work. The key is that you need to give up on imposing a structured progression. Instead, you need to provide hooks to get them started, and the ability to work off of the things they build while chasing the hooks.
So if you want to make a fantasy RPG where that's the case, you would create a vast and inhospitable series of planes, each more dangerous than the last and filled with a spotty spattering of cities, ruins, and monster-gods.
Your actions would be to create adventuring parties, then guide them around - fighting monsters or avoiding them or whatever. Moving uses up some adventurer spirit - the more adventurers in your party, the less you move per point of adventurer spirit. You can stop wherever you like, whether you're out of spirit or not... but once you run out of spirit, you can't move any more. There's various kinds of long-term things you can do while stopped, depending on what adventurers are in your party. These generally change the map slightly in your favor in some manner.
Similarly, certain kinds of adventurers are vastly less mobile on certain kinds of terrain. If you're okay with sticking to roads, then horse-mounted knights will be ideal. If you want to move through the jungle, you'd be better off with rangerly elves.
If another party reaches you, then the two parties can exchange whatever members they want, and transfer whatever adventure spirit quantities remain between the two parties as they like. Some adventurers can generate spirit once they settle down permanently, so this is a valuable contribution: stop by and get your spirit recharged.
Anyway, the game would need to be somewhat carefully crafted such that your party mechanics are a bit more complex than "N members walking along". You'd probably construct a party out of social links - this guy is the leader, those three are sworn companions, etc. These links would provide a structure to help respond to challenges in a manner well-suited to your needs. A lot more work needs to be done on the design of it before you can say "okay! That's it!"
Anyway, there's lots of other genres you could start to think about, but since it involves iterative building, every genre would have to be re-cast such that there was some kind of iteration. So a first person shooter - you'd probably have to play a series of different marines or something.
It's a fun thought experiment.