Overworlds are cool and amazing and you could probably be paying more attention to them than you are right now. The reason: Context.
At any given point in any given game you can stop in your tracks and think (well, in some games you can't literally stop in your tracks, but you can probably mentally stop in your tracks and get a game over or whatever, but the point is still made and you've still stopped in your tracks to think) about what you're doing, and why you're doing it. I mean, there's the obvious "because this is a game and I'm having a good time or at least being engaged" but it goes deeper than that. In The Legend of Zelda (pictured above), for example, if you stop in the middle of--oh, anywhere, but let's say a forest--and wonder "why am I trying to pass through this area, again?" your answer is probably "I'm trying to get to what's on the other side." Look at that overworld. Check it out! There's so much stuff to see! And the greater context of why you're trying to get to the other side of the forest is even more varied:
Overworlds lend all this different context to the simple activity of 'traveling'. You and someone else could be doing the exact same thing on one screen but be thinking about entirely different things. So for this next bit, we're going to have to cut up the experience of a game, crudely, into two parts. It'll be like a magician-saws-assistant-in-half kind of thing. Overworlds aren't everything, either. There's a lot more at stake here.
No time for subtlety. I'll call these bits ACTION and, predictably, CONTEXT. So in the order written, I'll talk about action first -- and though I'd like for context to take center stage I do want to touch on the importance of action. 'Action' as I talk about it here is the part of the game that directly engages you; it's the activity you're currently participating in. The first part of making a game is making the action, because it's the entirety of what cannot be left to the imagination. It's moving; it's jumping.
Context, now, that's a trickier beast to define. The problem is, you can sort of leave context up to the player... so sometimes that's what happens, to the detriment of a game! As a game designer, you can ignore context! I, as a player, can simply enjoy the action of a game if it's really good action, and come up with my own reasons for solving puzzles, killing monsters, collecting trinkets, talking to people, or placing blocks.
That's not to say I'm implying every game needs a narrative, a hand-crafted raison d'être. Quite the opposite, really! It's easy to state a character's motivation, and it's just as easy to tell a player what should be motivating them; the hard part is getting them to believe you.
Let's talk about Mario.
ACTION: I'm going right (->), the Bullet Bill is going towards me (
CONTEXT: Everything I've described above should be obvious to anyone who's played Super Mario World, but all those activity assumptions came from context, from the rest of the game. Why should I be going right? Why is a Bullet Bill a threat (it looks kind of round and -- okay, not friendly, but not harmful, exactly)? Why is being hit bad? There is complex context behind that simple action. Let's break it down.
You go right because that's how you get to the end of the level;
you want to get to the end of the level because that opens up the road to the next area on the overworld.
You duck because being hit is bad;
being hit is bad because it makes you shrink down to little mario;
being little mario is (usually) bad because if you get hit again, you die;
dying is bad because you lose a life, and all progress in the level--so you don't make any progress on the overworld.
You don't think about things in these terms, though! You come up with rules to follow and you stick to them. When you avoid an enemy, it's not because you're thinking about all of the consequences. You have distilled the context of the game down to what's relevant at a local scale: while controlling a running and jumping plumber, all you really need to remember is "GO RIGHT; DON'T DIE."
You don't need to confront the consequences of either victory or defeat until you're there.
Interaction is a give and take. Actions demand feedback. You can hold a controller in your hands and hammer away at the buttons while you're watching a movie, but the best you can do is memorize what happens and then press buttons, pretending what you're doing is having any effect on what you're watching (that actually sounds kind of embarrassingly fun to do with familiar material. I'm going to give that a try sometime).
Many games can be split cleanly into an action mode and a context mode. Super Mario World, to return to an earlier example, can be divided into its overworld (hey, it's that word again) and its levels. Many games ape this style of overworld, missing the point entirely of its power (including later Mario games!). I won't pick out examples, but it's terribly disappointing to me that basically nothing has utilized that style of overworld in a more interesting way since SUPER MARIO WORLD, a game made BEFORE I WAS BORN.
** (please send counterexamples or refute this point! nothing would make me happier!)
Rogue Legacy has its action mode (playing the game where you fight stuff and collect gold and things) and its context mode (spending your gold on castle upgrades, and equipment). Those catapult launch-your-Thing-as-far-as-you-can-and-then-upgrade-them games have an action mode (watching your Thing bounce off of stuff), and a context mode (spending the resulting cash). There are a ton of games built on this failure-mitigating structure, in which context is sacrificed to feed back into the action. You're collecting money so that when you ultimately fail, you can spend that money on making it easier to do the actions that help you collect more money, so that... etc etc etc.
That exposes a weird quirk in my action-context model. I -- personally -- think roguelikes are ruined when inter-run progression is added. You'd think that creates context where there was none between runs, but that's not true: it replaces a very personal context which is present in roguelikes as well as arcade games. Playing a game that treats you the same way every time means getting further is reflective of your skill. A game that allows you to and EXPECTS you to improve upon your character means getting further is reflective of some combination of your skill and your time invested.
tiny rant time
If I complete a challenging game, I'd like to feel -- in my heart -- "Congratulations! You're good at this game, and so you beat it!" without even the faintest hint of "Congratulations! You played this game for X hours, and so you beat it!"-taste in my mouth. Not every game has to be about overcoming a challenge, but... well, generally the ones with character progression are.
end tiny rant
... and up.
It's about time, way down here near the end of my piece, to get into muddling the lines between action and context. First off, remember when I broke down the encounter with a Bullet Bill. Although I've separated action from context in Super Mario World, tons of things blur that line. Both kinds of coin are useless in action, but add up to extra lives in the grand scheme of things (which are, in turn, useless in action terms but useful on the bigger picture). The huge colour switches? Same thing! They make a really big difference in how other levels are experienced but sort of... only a trivial expository difference in the levels they're in.
But that's kid stuff.
Overworlds are rad. The Legend of Zelda, bless its bones, has a beautiful map that happens to not be a separate contextual mode. The connection between action and context is kicked up a notch because of how many resources you have to manage, and how many nooks there are to explore. Later Zelda games really locked stuff down: speedrun-quality exploits aside, dungeons must be completed in a strictly set order. Nothing can be skipped, and you can rarely even enter areas that aren't either pertinent to reaching your next (single, solitary) goal or already unlocked.
In The Legend of Zelda, you have these scarce resource to manage: rupees, hearts, arrows, bombs, keys. Spending any of these in the moment will affect you in the future to come. Don't have any arrows? Well, that's because you shot them all fighting stalfoses or that crazy-ass unicorn thing guarding the master sword. No keys? Go find one in the dungeon, or spend your hard-earned rupees and buy a key. When you bomb cracks in walls, you need to be aware of the greater context: that's one of your three bombs, and you might want to hold off and wait, to bomb some dodongos you know are coming up.
Man, those are some weird names.
There's one little problem. In The Legend of Zelda, because of the persistence between action and context, anytime you enter a dungeon with fewer than maximum arrows, bombs, hearts, and anything else, you're acting sub-optimally. The only global time pressure that exists is your patience, and so if you're driven to act with the greatest efficiency, you'll be forced to spend a bunch of time doing potentially not-fun grinding to rustle up to maximum materials... and then, if you're like me, you'll be incredibly stingy with them anyway, to avoid having to do that again.
So, there are a basic triad of solutions:
1. Allow for grinding. See above.
2. Don't make depletable resources necessary for success. In Quake II (along with many other games that have limited ammo), you have at least one weapon with infinite ammo. It sucks, but at least you do have a fallback.
3. Allow for a permanent failure state, almost ubiquitously 'permadeath'. See: roguelikes, Spelunky, probably other stuff that isn't roguelike at all.
Shit. I forgot to even mention Pikmin. As a kid, the long-term time limit meant I quit before my time was even half up, because I felt hopeless. I have thoughts about the feelings I got from this, and how maybe to solve implicit problems with fail states for long-term goals... oh well.