Joystiq interview: Jenova Chen

Xinghan (Jenova) Chen has big dreams. A recent graduate of the University of Southern California's School of Cinema & Television, Chen focused his master's thesis on changing how game designers look at difficulty and redefine the concept of Dynamic Difficulty Adjustment. With two award-winning independent games (flOw and Cloud) under his belt as well as contributions to a multitude of other games, Chen's resume reads like an old professional in the gaming field.

We had a chance to speak with Chen to talk about his thesis, its (mis)interpretations, and his current work with Maxis.

Why did you choose this project as your Master's Thesis?

The Master's Thesis has to have some contribution to the field that you want to be a master. For me, I have created so many games in the past I feel just creating another game is not contributing anything. I'm more interesting in coming up with another solution in better game design that can be used by other game designers to attract more gamers. For other industries, any kind of interactive design, the theories in this project are helpful.

What are some common misconceptions about your thesis?

They basically said the opposite of what I propose. What is DDA? The computer will change difficulty based on how I play. If the player didn't notice it, I'd be fine. But they did know it, so they felt cheated and think, "oh, the game is easier if I play bad, harder if I play well."

This is the wrong way to implement DDA. Traditional DDA includes two elements. First a system, algorithm, estimating how a player plays. They want to do data collecting like how many people you killed, how to survive, etc. [The computer] then says, "okay, you have 200 points, you play really good (or really bad)." This might work on a large scale, but every individual is very different. 200 points may be a very good score for Player A, but it might not be good for Player B.

Here is an example, albeit a somewhat gruesome one. Let's say someone played Grand Theft Auto, and there's an algorithm figuring out how well he is performing. And it's deciding how well the player is surviving? What if it's a little kid? I've seen a little kid play GTA and just kill himself over and over again for fun. What if the computer thinks, "Okay, he's killing himself a lot, he must be really bad." Its not accurately detecting how other people feel.

Traditional DDA is based on data. They will have another message / algorithm to change data. If you change that, how much you want to change? How much difficulty do you want to change to make it feel right? You don't really know. It's purely estimation or purely data based on input from testers.

The DDA I'm proposing -- or EDA as I'd like to call it, to avoid the negative stigma of DDA. Embedded Difficulty Adjustment is the name I chose. It's embedded into game design, as opposed to a separate system. In EDA, the difficulty changes based on player's subconscious will. It does not change by system or designer.

How does EDA determine that?

Let's use, for example, Grand Theft Auto again. It has a good EDA existing but people don't realize it. In GTA you have all kinds of different activities you can do – you can race, kill people, do missions, etc. You have all these different activities with different difficulties and tastes around you. And the computer never tells you, "Oh you need to do this next." It's the player that decides intuitively or subconsciously what to do next. When the player wants to change the gameplay, the options are right in front of him embedded as game elements. And that's the best method.

If I play GTA, I do a little questing and grow tired of it, so I decide to on a random killing spree. After a while, I'd be tired of killing people, too. So maybe then I'd decide what to do. While I think about it, I don't think about "oh there's a system that changes the difficulty of what I do next?"

In psychology, in order to realize the Flow state you have to feel like you're in control. The fictional GTA example we used earlier is passive traditional DDA – you have no control. But with active DDA, you are in control.

I've worked on many games. FlOw is a simple game. It's the simplest test of active DDA.

Explain how flOw works.

Basically each level has different difficulties. It's not a linear progression. I want to see if the player can adjust by himself. I think by level 5 there's this ring monster that is really hard to defeat. Lots of people tell me, "oh it's really hard!" but what they did is skip it, or return to previous level to eat more and grow bigger, then return to fight the ray monster. It's not like in traditional games where you have to beat one level after another.

Take Shadow of the Colossus, for example. You have to beat one giant after another. I reached a point where I couldn't beat this one giant. I have no choice – I can't skip him, I can't work past him, I have to basically give up on the game. If that game uses the flOw system, the player can at least play other giants and when they're more skillful they can come back. This automatically increases the playtime of other players. I think I've been playing games for the last 10 to 15 years, and I still have trouble with Shadow of the Colossus. I have a friend who doesn't play games at all and he couldn't figure out first giant. So he just gave up.

As more and more people start trying video games, the problem is how to make games adapt to different types of players. I think by using active DDA system we can accomplish that.

Switching gears for a moment, what were your favorite games or biggest influences?

This is very hard because I've nearly played all the games so it's hard to say which is better than the other. Recently, the game that makes me feel inspired is Katamari Damacy. I really like ICO and Shadow of the Colossus. I can't say World of Warcraft is influential, but I play it every day. I just cancelled my subscription yesterday, actually. I've been playing that game for so long. Oh, and Final Fantasy VII, which is where my name is from.

You chose your name?

I don't have an English name, so I chose the name when I moved here. I wanted to have a name where I don't have any trouble applying to any account. I never have to add numbers. This is an awesome name. [Laughter.]

What are you currently working on?

Right now I'm working on Spore.

How are you applying your thesis work into the project?

I can't really talk about the project ...

... ?


Damn! Okay, what's it like to work with Will Wright?

I didn't get to work with Will very often. He's very busy as you can imagine, but from the conversations we had, he's always amazingly inspiring. If you consider me as a designer who has a bit knowledge about video games, the charm for Will is that he knows so much more than just video games.

Is there a way to implement DDA in multiplayer?

Multiplayer is a hard situation. Multiplayer is actually born to be DDA. If you have another player and he plays you, he is assessing you. Let's say it's Counter-Strike, 1-on-1. If that player feels that you're weak and gets bored, he will create challenges. He'll only use knife or pistols to fight you, for example, while you use the machine gun. He lowers down difficulty for you and creates challenge for himself.

Humans are lazy but not stupid. If the game is different, we will try to adapt. This is how active DDA works. If player wants to adapt, he will adapt. In a multiplayer environment he is much more accessible. When you look at Counter-Strike, you have two teams – one is good, one is bad. Generally, the good player will switch to the worse team. Psychologically and subconsciously we all want to make a game better.

That's why I think multiplayer is more addicting. I never see anyone play Counter-Strike and go back to simple single-player shooter. World of Warcraft is a great example – the quality of experience is totally different with more people.

There are all sorts of structures designer creates for society. Player must adjust accordingly.

Thank you for your time.

[Image credit: Wired]