The Hard Problem: Dynamic Content

It all went downhill when we started saving our progress in games.

Whose brilliant idea was that, anyway? Once we introduced the virtual bookmark, our industry decided we were making virtual books. This was a mistake.

Until that sad day, videogames were arcade games. When you walked up to Pac-Man or Q*Bert and put a quarter in the slot, it didn’t matter how many hundreds of times you’d played that game before. There you were, beginning at the beginning. Sure, maybe you could get to the third level in R-Type. Did it matter that each time you visited the arcade you had to play through the first two levels to get to the “new” stuff? No, it didn’t matter – because those first two levels were still hard as hell and a blast to get through each time.

This is all by way of me explaining why it is that Carnival Games on the Wii has handily outsold touted blockbusters like God of War II and Fable II, among many others. Because it’s not a virtual book, it’s an experience you can pick up anytime and enjoy. You know, like Clue or chess or football. But this also explains why games have gotten too expensive to make, too pricey to buy unless used, and too much of a turn-off to the mainstream: Those giant AAA level-based sacks of linear, use-once content demand enormous resources, require an exorbitant retail price, and are too complicated for ordinary people.

The solution is dynamic content, game content that is generated on the fly from a large library of components. Many games have dipped their toes into this approach. Some have jumped right in. But it’s still considered a rudimentary technology that can’t possibly deliver the quality and immersion of carefully scripted and planned gameplay embodied by the Call of Duty series, among others.

So what the heck is the hard problem here that I want to solve with dynamic content? The problem is levels.

Level 6: The Sewers of Ice World Volcano Mansion Driving Sequence

Level-based games have dominated the market for years. In some form, they’ve been with us since the days of Pac-Man and Galaga. But it’s been the rise of the first-person shooter, dating back to Castle Wolfenstein 3D, that has really led to game content being structured in a linear series of levels, each comprised of unique terrain and encounters and connected by a narrative. And unlike R-Type, you aren’t expected to begin at the beginning every time you sit down. No, we save your game, making the preceding levels obsolete once beaten, and turning your $60 game purchase into a one-way trip down obsolescence street.

Level-based games are a huge problem. They’re why AAA next-gen games cost $20 million and up to make. Every level’s worth of content represents a massive investment of time, talent, and manpower, all for an experience that a typical player might see for twenty minutes or so before moving on to the next level. Forever.

In fact, let’s take a look at two games that used dynamic content and see what we find there.

Fable II: While most of this game was the usual scripted missions, there was some dynamic content going on. This manifested as the way that the characters and world reacted to your choices. To cite one example, a given town could shift between lawful and lawless, prosperity and poverty, based on your actions there. That’s cool on paper. But in reality, that kind of content isn’t dynamic enough. If the player picks one general approach to play and sticks with it, he’ll never even notice the town’s subtle shifts because it keeps evolving in the same direction. Only if the player is deliberately whipsawing between good and evil, prolificacy and penury, will he see the town jump from lawless ghetto to well-guarded suburb and back again, and only then will he realize the scope of what the game might do. What’s needed here is content that waxes and wanes due to factors other than just the player’s journey. Set the baseline behavior as turmoil and churn and then leave it to the player to do the work to dampen the turmoil and push the town in a deliberate direction – and then work to hold it there.

Left 4 Dead: This game is really the poster child for dynamic content because that’s its raison d’etre. Valve wanted to explore dynamic content in the form of their “AI Director”. They built four short “campaigns” but then put the real work into a hugely varied and responsive system of spawns, spawn points, and spawn rates. The enemies ebb and flow in a dramatic fashion, and the same level played ten times plays differently each time. One playthrough may have an epic battle in a parking lot, and the very next time the parking lot is deserted but three medium-sized fights break out in the surrounding buildings. So Valve did great work here, but they were exploring dynamic content within a very narrow definition. The player’s ability to consciously influence the dynamism is about nil. The joy of the game isn’t from interacting with the dynamics; it’s the simple joy of shooting zombies over and over again on the same maps without it getting repetitive.

Okay Mister Smartycontentpants, What’s Your Bright Idea?

Let’s not screw around. Let’s redesign Grand Theft Auto IV without scripted missions. They call it a sandbox game, right? Let’s play in it.

So here’s my intent for this exercise. I want the player to drive the experience by setting out some very clear goals he can pursue and a diversity of methods for achieving those goals. Then I want to put some dynamic systems in place that will respond to his attempts to achieve those goals and add challenge, surprise, and depth.

In my version of GTA IV, Nico Bellic is out to make his fortune in his new country. We show the player a set of progress meters he can try to fill. These are called Wealth, Looks, Infamy, and Influence. Each one goes to 100% and there are thresholds at each 20% increment. Here’s how they work.

Wealth: You fill in this bar by getting cash, stealing cars, taking control of businesses, buying houses, etc. The cash is straightforward. Your property, though, has to stay safe. Your garage full of stolen luxury cars, your six houses on the waterfront, the thirty bars and dry cleaners and shoe stores you’ve muscled into giving you a cut of the take – you’ve gotta keep it all intact to maintain your wealth.

Looks: You fill in this bar by stuffing your wardrobe with more and more expensive clothes; by visiting the gym to keep yourself in shape; by keeping your hairstyle up to date with the latest trends; by ferreting out private trunk shows by fashion designers; and by collecting accessories like cologne, cufflinks, gold necklaces, etc. Your looks have to stay current, though. Your clothes and hair have a decay rate as they fall out of fashion.

Infamy: You fill in this bar by committing crimes: killing, carjacking, bank robbing, all that GTA stuff. The more your Infamy fills up, the more recognizable you are on the street. You’ll intimidate some enemies into surrendering on the spot, but others will make you a target just to make a name for themselves.

Influence: You fill in this bar by making deals with important people. Early on you’re shepherding two gangs into joining forces to expand their joint territory. Later on you’re getting a city councilman to cut a deal with the mafia. As your address book fills up with more and more powerful people, your Influence spreads. But those powerful people have powerful enemies who may come after you, and if your powerful friends end up dead, then so does your Influence.

I mentioned that each of these meters has thresholds every 20%. These meters are dynamic, shifting up and down depending on how the game is going. But whenever you get one to the next 20% threshold, that becomes the new floor for that meter. Get your Looks to 43% and it’ll never drop below 40% again. This lets you experience the dynamism of the system without being forced into constant maintenance on everything all the time. You can push to get Infamy from 45% to 60% while coasting on Looks at 20%, then let Infamy sit at 60% while you get Looks to 40%.

Those thresholds are also used to unlock more content. In the case of Influence, I described how the kinds of important people you deal with changes over time. That’s explicitly broken up by these tiers. When Looks crosses 20%, fashions change and now you need a new wardrobe, hair, etc., to get to 40%. When Infamy crosses 80%, random street punks no longer take potshots at you because you’re too notorious – but skilled assassins start showing up to blow up your cars, set traps in your houses, and murder your Influence partners. So as you get each meter to a threshold, that aspect of the game changes and new content and challenges come your way.

These meters are pretty cool. Let’s not just restrict them to the player. Let’s introduce the concept of rivals.

A rival is an NPC who is pursuing the same set of meters you are. Let’s say that whenever you reach a meter’s threshold, a new rival is introduced. He’s at the same threshold as you and his goal is to beat you to the next threshold on that meter. So when you hit 40 on Infamy, we introduce Johnny Stumpkin, rising thug, who is also at 40 Infamy. Let’s say we can introduce a rival within any Bank Robbery, New Partner Deal, or Business Extortion action you take. Those encounters have a slot where a rival can be slipped in, triggering a short non-fatal confrontation where he robs the bank just as you arrive, or kills the new partner, or firebombs the business just after you extort it. In these encounters the rival screws you over, announces himself, and splits with his own personal thugs. Until you reach 60 Infamy, Johnny Stumpkin will keep causing you trouble.

We’ll customize the standard content to include this rival’s agenda: For example, when a street thug takes a shot at whacking you, there’s a good chance we’ll make that thug one of Johnny’s goons. Meanwhile, Johnny is pursuing his own goal of reaching 60 Infamy and he has the same options you do – only your holdings are part of his list of valid targets. If the Johnny NPC decides to steal a car, there’s a chance he steals one from you. If he wants to rob a bank, we’ll make it a bank in the neighborhood where you currently are. Johnny will haunt you and dog your trail until you manage to kill him – which is really, really hard – or until you beat him to 60 Infamy. If you do that, Johnny’s thugs desert him and you can take him out easy. But if he beats you, he becomes substantially more powerful and continues striving to beat you all the way to the top.

One of the concepts I’m using above is the notion that any of the standard encounter types, such as a bank robbery or a carjacking, can have a wild card. I’d really want to run with this idea. The standard templates for all these things should have a couple of random points where things may or may not happen differently. If you’re doing an Influence deal, your new partner may give you a task such as firebombing somebody’s business before he’ll complete the deal. That’s a possible twist on the Influence encounter. But that firebombing encounter may have its own twist: The target belongs to your current Wealth rival and his beefed-up thugs will make the job tougher. So these wild cards aren’t just random events; they’re slots into which other currently active dynamic systems can plug into. As these connections get made, the world comes alive. Everyone has an agenda and every action has consequences.

I Could Go On

There are plenty of problems with taking this approach, but we aren’t going to solve them by sticking to our static, level-by-level content.

My GTA example is a big, sprawling project meant to show how you could do a giant AAA project with this approach. But scale down those ambitions and the presentation and you could turn that into a Harvest Moon-style game about running a happy little farm, or a game where you’re a circus ringmaster who needs to keep the audience on the edge of their seats by deftly summoning the right acts at the right moments and responding to the energy of the crowd. You can make games a lot cheaper, a lot more pick-up-and-play, and a lot more responsive to the player’s will.
I have two final points here.

One is that we don’t need to make virtual books and be auteurs. We can make experiences. That’s what Wii Sports and Carnival Games and Settlers of Catan are: simple, repeatable experiences that can be picked up, enjoyed, and put down until later. They don’t need the plodding tread of level content weighing them down. They’re games made by game designers, not auteurs who can’t get their fantasy novel published.

My other point is that the way to make dynamic content work is to make dynamic content work. We advance by building on what the last guy did. We need more studios trying this approach so that the next round of games can learn from their successes and mistakes and take this further and make it better.

Valve just threw down a gauntlet with Left 4 Dead. Who’s next?

John Scott Tynes is a game designer who wishes he had the time to be a modder, because that’s where the action is.