[The design director of Volition/THQ's expansive open world action title Red Faction: Guerrilla talks about the transition from being scared of player freedom to truly embracing it, including several case studies of specific missions in the game and they evolved to support meaningful player interaction.]

Designers create experiences for the player. That's a loose enough definition that there's little room for debate. But how much control is implied by that short phrase "create experiences"? Does that mean micromanaging the moment to moment play? There's a school of thought that says that's what designers do, and many of us believe it. And for some types of games it is true.

Then there are open world games. Crafted from the dust of early computer role playing games -- the original open world designs -- the goal isn't to tightly control the player experience, but rather to build a world and turn the player loose in it.

Give the player the freedom to choose his or her own path. Provide meaningful options. Encourage experimentation. Carrying over the hand-holding approach of linear storytelling games doesn't work; an open world is more than just a lobby for starting linear missions.

To truly fit into the open world model, missions have to provide the same sense of freedom that the world itself provides. And to make that work takes a change of mindset. It means letting go of being a control freak and instead embracing the chaos that's inherent in open world design.

There's a certain fear here. Much like some paranoid graphics programmers thought that their worlds were crashing down when texture mapping moved to hardware, some designers feel that they're being outsourced to code-driven systems. That fear is unfounded, even in the games that go to extremes to maximize openness.

I was the design director at Volition for Red Faction: Guerrilla. If ever there was a game that struck terror into the heart of a design team, that was it. Not only was it open world, but every single wall and fence, every door, every building -- including the ceiling and structural frame -- could be damaged and completely destroyed in arbitrary ways.

A tower could fall sideways onto a two story building, tearing through the roof and drilling straight down to the ground floor. A vehicle could explode on a bridge, making the bridge unusable for other traffic. Rubble from a building could fall in the road, preventing reinforcement personnel carriers from getting where they needed to go.

And none of this was perfectly predictable, being at the mercy of dozens variables going through the destruction engine. That tower could have just sheared off the outside wall, depending on the exact forces that caused it to topple. That vehicle could have exploded off to the side of the bridge, still creating a large hole, but one that can be navigated around. That rubble might have blocked doorways instead of the road, or even killed your attackers.

How can you even begin to control the player when all bets are off, when the traditional ploys of locking a door or blocking an alternate path with a chain-link fence don't work?

But that's the extreme case. Before considering that, let's go back to the first problem: How to build an open world mission that emphasizes player freedom?

Open World Mission Basics

Linear missions may be a comfortable design technique, because there's a long history of them in games, but they're not easy to build. The more we try to control what the player sees and does, the more we try to be cinematic, the more effort it takes.

It doesn't take much thought to poke holes in a hyper-linear mission. What if the player is looking the other way when an important event occurs? What if the player is still fighting bad guys when he crosses an important dialogue trigger? Should that dialogue play during heavy combat? Go too far down this road and you might get frustrated: "All that work I put into that scripted event, and half the people playing didn't even see it!"

You can fix that by taking away player control and locking the camera, but all this micro-directing of the mission gets to be expensive: every minute of gameplay has a fixed content creation cost associated with it.

As harsh as it may seem, to embrace the freedom inherent in open world games you need to stop caring. You can't control what the player experiences every moment. That's not a failure; it's what comes with letting the player do what he wants.

The key to designing an open world mission is to identify -- and avoid -- places where the player thinking for himself could break things, and that means looking places where time is a dependency.

Usually these show up as the word "then" in the spec. "Collect three sandwiches" is dependency free. "Collect three sandwiches, then find a picnic basket" isn't. You can't collect the picnic basket first? The worst case is "find the ham sandwich, then the egg salad sandwich, then the grilled cheese sandwich." Now there's no player choice, and there also more questions. Does the grilled cheese sandwich even exist in the world before you've picked up the first two? These are the kinds of questions that any good quality assurance team is going to press you for answers on.

Most time dependencies involve a player gating mechanism. "First find the key to the museum, then get the mammoth tusk from the natural history wing." There's no worry about collecting the tusk before the key, because the door to the museum is the gating mechanism. (Unless your game lets you knock the door off its hinges or smash right through the outside wall; more on that in a bit.) Gating works, but there's no denying that most methods fall back on some of the most spectacularly trite elements of game design: keys, force fields, drawbridges, radiation suits, etc.

The first step is designing an open world mission is to remove as many time dependencies as possible. Behind each time dependency is the need to gate the player, and there are only so many bulletproof gating mechanisms. An easy option is to see if "do A, then B, then C" can be rephrased as "do A and B and C in any order." Next, see if "A, B, and C" can be cleverly massaged into a single "A."

Case Study: Death by Committee

In Red Faction: Guerrilla, "Death by Committee" is a late-game mission where you discover that a group of businessmen who've been working with the enemy are having a secret meeting. In the original spec, you arrived at the meeting location before anyone else, so you could plan your attack and set traps. Then at some point the businessmen arrive, and you ambush them.

This sounded simple enough, but there was a key time dependency causing problems. There were two distinct phases: the set-up (before the businessmen arrive) and the attack. Key to the mission was that the businessmen would run for it once you started shooting or blowing things up, so what if you were causing explosions before they even arrived? Should that be an instant failure, because no one is going to drive up to a building that's on fire, expecting a PowerPoint presentation?

There are also some troublesome player feedback issues: knowing how much time until the participants arrive, knowing when they're there, knowing when they realize it's a trap and try to get away.

Most of these problems were solved by flipping the situation around. Instead of you getting to the meeting location first, the participants are already there. You show up and ambush them. Now the flow is much harder to break: get to the building, go inside, attack the conspirators, and if any try to escape in vehicles then chase them down.

Even that description is too complicated and controlling. Really, the goal is simply "kill all the businessmen." They're at the building, so you obviously need to get there first. And naturally when you show up with a gun and explosives, they're going to make a run for it.

"Go inside" is too much information, because you may not need to do that. You could use a sniper rifle from an adjacent building. You could ambush them as they run out of the building. You could destroy all the escape vehicles and chase people on foot. (Perhaps the cleverest option is to get into the building, climb up one story, then destroy the stairs leading down to the ground floor so no one can escape.) One simple goal, lots of options.

Open World + Destruction: All Hope is Lost?

A rite of passage for new designers upon joining the Red Faction: Guerrilla project was to declare that arbitrary destruction was too difficult to work with and we should abandon it. Heck, even I said that.

Destruction initially scares the hell out of designers, because it removes so many options from the usual bag of tricks. A sledgehammer serves as the key to any locked door.

Beautiful level design where the world looks expansive, but you're actually funneled through choke points (see just about any Left 4 Dead 2 map) is much harder to do. Let that take a little while to soak in.

Walls, rickety fences, wrecked cars, piles of rubble, furniture stacked in front of a door... those can't be used to direct the flow of movement, because any of those obstacles can be blasted through with a variety of explosives and weapons and even moving vehicles. All those things that sound scary are really player options, and just that they exist at all is a source of wonder.

The original Red Faction, released in 2001, was a fairly linear game in the same vein as Half-Life, except it allowed arbitrary "geometry modification," meaning that you could dig tunnels through dirt and walls. You'd be fighting through a mine and come to a steel door that's sealing off the area. Fire a couple of rockets at it... nothing. It's too thick.

But aha! Start shooting rockets into the dirt surrounding the door, and get a tunnel around it going. Until you run out of ammo, that is. To combat that, there were some unfortunately lame fail-safes built in, such as your associate Hendrix announcing that he hacked into the security system and the door would open shortly. Open world plus destruction is a much finer combination, one that enables enough interactivity that it's worth making headway on the design headaches it causes.

Full Guerrilla-style destruction isn't as crazy as it first may seem, in terms of building missions. First, destruction takes time and ammo. Early on in the project, one of my great worries was that players would level the world within the first twenty minutes.

That turned out to be unfounded. Destruction also destroys cover, opens up the battlefield, and makes buildings structurally unsound so going inside them runs the risk of being caught in a collapse. It's also a conscious decision for the player to focus his weapons on inanimate objects while he's being fired upon by enemy soldiers.

And most important of all, unlike the original Red Faction, terrain in Guerrilla is impervious to explosives. So while it may look like the player is ridiculously powerful and all hope of pacing and balance is lost, don't panic. There's more than enough for designers to cling to without stomping all over the fun that players have blowing things up.

Case Study: The Dogs of War

During development, this Guerrilla mission was called "Sniper Hunter," and that's a good overall description. There are four apartment buildings, divided into two groups of two, and there are a total of eight snipers on top of them. The goal is to kill all the snipers.

There are no sub-objectives, and therefore no time dependencies, so this is a nice and clean open world mission. Except that the apartment buildings are fully destructible and by this point in the game that player has access to various explosive weapons, including a rocket launcher -- plus the Nano Rifle which can dissolve anything it hits, including walls and ceilings.

The interesting question is, why don't all players just level the two buildings and be done with it? And while that's possible, it's not something that happened often in playtests. Here's why:

Is it still possible to destroy the buildings at a distance, completing the mission in a fraction of the time that a "normal" attempt would take? Sure! This is especially true if players have bought the highest-end weapon, the Thermobaric Rocket Launcher. The existence of that possibility is a wonderful thing. It proves that players aren't forced to bow to the whims of the mission designers. It means that all that money spent buying the big guns pays off, and the game respects that.

Missions That Aren't Missions

Trying to mix perfect predictability into situations like this... well, it's a recipe for failure. Destruction not only encourages thinking about gameplay at a higher level, it's a requirement in most cases.

Upon first hearing about the Guerrilla destruction engine, a common brain flash was to want to topple a tall tower across a chasm creating a makeshift bridge. What a perfect use of destruction! But it's also missing the point.

With damage being driven by a complex simulation system that can make the tower break apart in thousands of ways, and fall any direction in any number of pieces, the only way to guarantee that the tower will make a bridge every time is to bypass the whole system and fall back on a hand-scripted event. And that's a bad trade-off: taking away player freedom because a designer wants to control what happens.

At this point you may ask, if time dependencies are removed from mission objectives, and level designers build missions as situations in which player movement cannot be perfectly controlled or predicted, then should the result even be framed as a modal mission?

One of the great realizations during the development of Guerrilla was that a "mission" can simply exist in the world if the objective is core gameplay that players already understand. In the end, the world map was dotted with "destruction targets": buildings that were important to the enemy that could be destroyed for a reward.

The high importance targets were carefully constructed and were consistently praised as one of the best parts of the game. Yet a target could be built for a fraction of the cost of a mission, because there wasn't all the custom dialogue, there wasn't an introductory briefing, and there wasn't the worry about edge cases, like the player starting a mission and then driving to the other side of the world.

Case Study: Guerrillas at the Gates

"Guerrillas at the Gates" is the second to last mission in Guerrilla. It's not the end of the game, but it's where you strike the critical blow to the enemy by attacking the central command center. Initially it wasn't a mission at all, but a series of key buildings to destroy.

An enemy compound is a great example of an objective that can be built in a non-modal way. There are distinct entrances, protected by checkpoints. The highest points are occupied by guard towers containing long-range turrets. Indestructible terrain funnels the player through a chokepoint that divides the outer section of the compound from the inner sanctum. Just as with a military base in real life, getting inside is a challenge in itself.

The entire central command complex is still destructible, but there's a natural flow to it which means you can't just find a camping spot and snipe all the critical buildings with rockets. There's a layering, a feeling of successive targets being deeper inside. Clever use of terrain keeps the whole base from being visible at once, so there's the feeling of discovery and progress.

As a non-modal destruction target, it got such good feedback, that we decided to turn it into a full-blown mission. The objective was still the same -- destroy all the marked buildings -- but by turning it into a modal mission we could add some more scripted events to better give the impression that you were part of a coordinated assault, such as guerrilla reinforcements arriving. Scripting-wise it was one of the quickest missions to build.

As with "The Dogs of War," the chaos of open world destruction meant that occasionally there was a short-circuiting of objectives, such as shooting down a gunship and having it crash into one of the target buildings. An exploit? Only if you're paranoid about controlling what happens. To players who saw it happen, and knowing full well that they caused it, it was brilliant.

I would like to thank the entire Red Faction: Guerrilla design team, as it was their experimentation with open world mission and level design which resulted in most of the points discussed in this article.