... They're very popular, these open world games. But I think it's worth considering their strengths and weaknesses rather than simply screaming "open world!" and spending ten million dollars on it.
The core idea of an open world game is that you can interact with the world in any way you see fit, rather than being stuck in some kind of linear mission progression. The truth is usually that you're stuck in a linear mission progression, but you can ignore it and go diddle around.
I don't like open world games.
Okay, that's really wrong. I love open world games. I love them so much that I hate them for falling short so much.
For example, Crackdown is an "open world game". You have the whole city. You can go anywhere you want. You can even take your starter pistol and baby-fat starting character over into ninjas-kill-you land. If you're good enough, you might even win. And the game can handle that.
The issue here is that this is not a simulationist open world. You can tackle the quests in whatever order you like, you can collect orbs, you can even fight the cops, but none of this stuff is terribly emergent or adaptive. Even the cops get tired of chasing you after a bit.
This is true of every modern open world game I've played, except maybe some of the roguelikes. GTA3 and Mass Effect are the same way, as simple examples: you can do the missions, you can explore the city/universe, and maybe you can play with the cops for a bit until you get bored.
There are things to do in the city - steal cars, do races, etc - but these are simply side missions scattered around the city just for kicks. They are scripted in, preprogrammed pieces that change nothing except, perhaps, your XP meter.
I find that these are a disappointment. I feel that an open world game should maybe be about THE WORLD. Hence, you know, "open WORLD".
Fable II offers a very basic glimpse into this kind of idea, although I only bring it up because it's recent: half the space-adventure games since 1993 have done the same thing. It's relatively easy to model the economy of a system if you ignore realism, so these games all have an economic system for you to open-worldly abuse.
The problem with these kinds of games is that I find them a bit unsatisfying. Once you've bought the city (or learned that buying the city will take all year and isn't worth it), what's the point? All of those stupid prescripted missions do have a purpose: they give the world texture and flavor. They make THIS planet different from the last one you landed on more than superficially.
But the missions are boring! Not only are they only minimally adaptive, they're not "open": the mission goes no deeper or shallower than scripted. For example, if I save a group of slaves from Plorbax the Grundarian, I'm generally given a good option (let them go, even though we're in the middle of a jungle full of fifty foot high monsters) or an evil option (usually, kill them. Occasionally, sell them).
I can't actually interact with these newly rescued people. I can't offer to schlep them back to their homeworlds, can't offer to make them my crew, can't try to date one of them, can't try to settle down in the jungle like the Robinsons, can't do anything.
Obviously, simulation of PEOPLE is a bit more difficult than simulating an economy, not least because we can't easily simplify people without losing what makes them interesting. It's a bit more difficult, yeah, like 'nobody's ever done it' difficult.
But what, I thought, if we're coming at this from the wrong angle. What if instead of trying to simulate the people, we just try to simulate the illusion of emergent behavior?
In my mind, the point of an open world game is that I am permitted to explore the universe as quickly or slowly as I see fit, in as much or as little detail as I wish.
Let's say that we build a typical open world with all its kajillion little side quests. But, instead of placing those side quests in the world, we leave them floating free in the database.
As the player explores our world, we can assign these side quests. So, he's trying to chat up someone on the street? Now she's the main character in the "I'm being chased by the mafia!" side quest. Going down a dark alley? Now it's the "out of control car!" alley side quest. Looking more closely at a corpse? Now it's the "mysterious letter in my pocket!" side quest.
Furthermore, using this method it would be easy to release supplements or mods, free or for a price, to instantly integrate into the world. You would keep the density down, obviously: you don't want every random passerby to hit up the player with a side quest, you don't want every new building to be a weird new situation: just the ones the player seems to show an interest in. It's also quite possible to string them together: while rescuing the slaves, you can take special interest in one and they will have a suitable side quest assigned, just as if they were an actual character with an actual, meaningful existence.
In addition I think this would permit a whole new set of "flavor" subquests. For example, if you're on top of a building, admiring the view, it could spawn the "side quest" for someone else admiring the view with you. There's no competition, no challenge, it just adds some targeted flavor to the world.
What do you think?