Saturday, January 12, 2013

The Central Problem of Computer Roleplaying Games

Roleplaying games on a tabletop, where you and friends play characters having adventures in a gameworld, are a lot of creative fun. You can invent stories as you go.

Computer games, which can actually show you what created worlds look and sound like, are also fun to experience.

But roleplaying games on a computer? There's a problem. And it's a big one. And not many game developers seem much interested in trying to fix that problem, with serious consequences for gamers who want to play interactive stories where their choices matter.

So what's this problem I somehow think I can see that others can't?


Consider the following experience of playing a table-top roleplaying game (RPG).

You and your friend are roleplaying as a warrior and a thief, respectively. The character you're playing is a barbarian warrior who has been able to learn some magic spells. Your friend is playing as a thief who's remarkably good with a short bow.

Your two characters enter a dungeon that your other friend, the Game Master (GM), has prepared for you. Before you got together, the Game Master spent time mapping out all the rooms and secrets and filling them appropriately with enemies and puzzles and traps and wonderful loot. And the whole dungeon is just one segment of an intricately designed campaign that will involve you and your friends in a deep and engaging story.

As the barbarian and thief characters of you and your friend enter the first room of the dungeon, your Game Master has noticed that you (the barbarian character) are walking a few steps ahead of your nimble thief friend. He rolls a die, and tells you that you've inadvertently stepped on a trap -- and the room is now filling with smoke.

Being a relatively clever warrior, you announce to the GM that you are lying on the floor to see if the air is clear there. The GM wasn't really expecting that, but he rolls a die and tells you that, yes, you can see the mail-shod feet of what appear to be five enemies striding toward you.

This removes the element of surprise from the bad guys, and you and your friend are able to prepare for battle.

Out of the smoke, enemies attack the two of you. Despite your preparations, the dice are against you tonight and you both take a lot of damage very quickly. Your barbarian character, and the thief character of your friend, are probably about to die. It won't be the end of the world; you'll just roll up new characters. But it would be less fun than keeping your current characters.

You look at your friend, he looks at you... and both your characters turn as one, bolt from the smoky room, run out of the dungeon, and flee at maximum speed to the village down the road.


The GM looks a little puzzled that you've completely ducked the adventure he had designed for you. But he shrugs and lets you keep running. You fall over yourselves getting into the tavern, where the GM tells you that the locals and traveling patrons stare at you for a moment, then shake their heads and mutter "some heroes" into their mugs.

After a bit of quick dice-rolling, your Game Master friend tells you that you stick around the village for a day or so, with nothing much happening. The next night, though, while you're back in the tavern trying to drink away the memory of your brush with non-existence, a well-dressed stranger introduces himself politely, mentions that he couldn't help but notice that you both must have come from the nearby ruins -- "We get a lot of that here," he chuckles -- and suggests that he might have some easier work for you... if you're interested. Nothing dangerous, just a bit of guard duty for some merchants.

You and your friend accept. It turns out to be a set-up; you were hired because word of your flight from the dungeon got around and the stranger needed a couple of patsies who would run from a fight. Then he'd steal the merchants' goods for himself and blame it on you. Your characters are no longer welcome here, and you'll need to choose whether to fight to restore your good names or seek adventure elsewhere.

What's important to notice here is that the Game Master was able to adapt quickly and effectively to your surprising choice to bail on the dungeon adventure that he had set up for you. Your choices mattered, and had consequences, but because of the GM's creativity your actions in that individual story could still be worked into the overall narrative.

In a later adventure, you would learn that the GM used your unexpected actions (and the consequences) to deepen the emotional theme of the overarching story. That would never have happened without your choice of actions, even if they weren't what had been plotted out to start with.


Now consider an example of what it's like to play a Computer RolePlaying Game (CRPG), and in particular one of the variants of CRPGs known as a Massively Multiplayer Online RolePlaying Game (MMORPG) that are played with lots of other people over the Internet.

You and your friend are roleplaying as a warrior and a thief, respectively. You're a level 38 Warrior, so you have the exact same warrior-specific skills defined by the developers for every person in the game who is playing a level 38 Warrior. Your friend is a level 33 Thief, with the same abilities as everyone else playing a level 33 Thief.

Your characters enter a dungeon that the developers have coded for the many thousands of people playing the game. You can't all play at the same time, of course; this dungeon like all the others is "instanced" so that you and your friend are the only players there. The dungeon is laid out as rooms filled with enemies and loot, and it's in a part of the map that's pre-defined as having a degree of challenge appropriate for mid-40-level characters.

You walk into the first room of the dungeon. You enter the range (hard-coded at 10 meters) at which one of the enemies standing there is programmed to detect a player character. All of the enemies in the room immediately attack you. You die. Then they attack your friend, and he dies.

You both respawn outside the entrance to the dungeon. You go back in. The dungeon is exactly the same, with the same enemies in the same room. You both die again.

You both respawn outside the entrance to the dungeon. You go back in, and get a little further in, then both die again. You repeat this several times, getting better at it with each run-through, until you finally reach the end of the dungeon and collect the nice loot.


Several days later, after you've both improved your characters a few times so that they're level 40-ish, you go back to this dungeon.

Everything is still the same. The same enemies are standing in roughly the same locations; the same kind of loot drops when you kill them (though more quickly this time).

You go through this dungeon a few more times, grinding without much risk through the now-easy enemies to collect loot and and earn experience points for leveling up your characters again.

The next night you go into a different prebuilt dungeon with somewhat more powerful enemies and slightly better loot. You grind through this dungeon, too. If there's a "story" to the elements of either dungeon, they aren't connected to each other in any way. Nor are the stories of either dungeon related thematically to the challenges that deliver the main story line, which you can only do in order. Also, you can't advance your character abilities past certain levels until until you've successfully completed these developer-designed core challenges.

Other people you know played the same dungeons and challenges, and had the same kind of stories to tell about them. You like that this common experience gives you something to share with all those other players. But you also find yourself wishing you had some stories to tell that were uniquely your own.

You've made thousands of choices. But they all resulted in your character experiencing the same story as everyone else who played your kind of character. Despite playing an interactive game, where the computer is programmed to detect your actions and change the world of the game according to encoded rules, your choices ultimately didn't seem to matter. And you wonder why.

If you're playing a unique character in an interesting world -- which is the promise a roleplaying game makes -- shouldn't what your character does have more effect on your personal story as it helps define the story of that world?


The obvious difference between these two scenarios is that there is no human Game Master in roleplaying games that are presented completely by a computer.

I believe this is the central problem in all computer-based roleplaying games developed so far. Every feature in a computer RPG is an attempt to program a computer to do some of what a human Gamemaster (GM) can do. And they're all unsatisfying compared to tabletop RPGs because there is no computer program yet written that can do what a human being can do.

There is no computer-based roleplaying game that can perceive what each individual player chooses to do and why they make that choice, that can understand the kind of game that each player wants to play, and that can respond to those unique desires by using their choices to help tell a good story for each player.

The computer game has not been written that can adapt any or all of its existing content elements -- or create new features on the fly -- to satisfy the play interests of one or more human gamers while dynamically weaving their choices and consequences into a larger story unique to that gameworld.

People still enjoy roleplaying games, though. And it would be great to be able to play these games even when a human Game Master's not around. Computers are the obvious solution to this wish. Let them handle the generation and presentation of gameplay features and events, and especially let computers do the tedious calculation chores for table-driven outcomes.

So computers have been programmed to do those things. They can generate pseudorandom numbers, and render pictures of places and objects. They can process inputs according to simple rules and select outcomes from lists or calculations. They can even cause objects within the gameworld to perform specific behaviors in response to a small set of player-activated trigger events.

As a result, there have been computer roleplaying games for years now. Some of them are a lot of fun. But none of them would claim to be as perceptive and understanding and creatively effective as a human Game Master, or as deeply satisfying as a game whose character-focused content is tailored on the fly to what a player enjoys by a responsive human GM.

Computers remain bad at portraying interesting characters who demonstrate human-like behaviors and communications. Despite the welcome recent interest in procedural generation, computers are still not good at dynamically generating new content that responds plausibly to a wide range of human player inputs. And they are almost completely hopeless at figuring out how to weave individual events into a logically and emotionally engaging high-level story.

The result is that for all the effort spent thus far, something wonderful has been lost in the conversion of roleplaying games from the tabletop to the computer. From the expressive richness of worlds that adapt to all your unique actions, we get play experiences that are simplified to one-size-fits-all movies with limited interactivity so that a computer program can deliver a story that has been predetermined by the developer.


There is reason for hope that this can improve. Some efforts have been made to allow more player choice, and to support consequences for those choices.

Unfortunately, there's some evidence that if you choose to make this kind of roleplaying game, you have to be prepared to give up the need to tightly control the full story. If the choices of the players don't really determine how the story ends, players will recognize that the limited interactive freedom you gave them is just an illusion.

I think this is precisely what we saw in the respose of many gamers to the end of the last game in the Mass Effect trilogy.

Players who were unhappy that the ending(s) of the final game didn't reflect the roleplaying choices of their character were both right and wrong. They were right that the player agency EA/BioWare implicitly promised -- by letting them develop the character of their Shepard through their gameplay choices -- wasn't delivered in the epic ending of the story. They were right to want and expect their choices to be respected, because EA/BioWare designed Mass Effect to solicit choices almost continuously throughout all three games.

But players were wrong to think that Mass Effect, which was designed from Day One as a world and a story created entirely by EA/BioWare, would ever permit an epic story to have a less-than-epic ending. What happens to the galaxy, and to Shepard, were built into the Mass Effect story from the very start because it was conceived as a developer-directed story.

The presence of choice-permitting gameplay mechanics (some of which were severely curbed in Mass Effect 2's RPG system, which should have been a heads-up) was never going to shift any of the core storytelling to the player. Mass Effect was EA/BioWare's story, and the conclusion to it had to be -- was always going to be -- dictated by the game's developer, not the player.

The relative absence of some form of adaptive/creative human storyteller in Mass Effect, either a real person or a computer simulacrum, meant that players were going to get a single predetermined core story designed for a mass audience. Perversely, it was EA/BioWare's attempt to allow a small amount of human storytelling adaptability in these games that caused the ruckus: having only the appearance of meaningful choice turned out to be insufficient.

The convention in computer-based roleplaying games has been that these games must tell an epic core story. This renders these games into books or movies that happen to offer some moment-to-moment interactivity. Mass Effect seemed to promise more; it broke with the games-as-interactive-novels convention by implying that the player's choices of how Shepard behaved in emotionally-weighted moments would change the story.

And ultimately the story of the world was affected. Multiple endings were keyed to Shepard's overall nature. But for most players, the story of their Shepard failed no matter what choices they had made within the Mass Effect CRPGs... and that was the story that mattered.

Players who care about story in a roleplaying game want their choices to matter for the character into which they've invested so many roleplaying hours. Without that personal responsiveness, computer-based roleplaying games will remain interactive novels. And even if they can't explain why, players will continue to feel puzzled and disappointed that that some vital and unique benefit of roleplaying games is missing from the computerized version, and that some part of the promise of interactivity is being wasted.

I think they are exactly right to feel that way. The central problem of computer roleplaying games so far is that they don't have the ability to respond creatively to the vast range of possibilities of what characters ought to be able to do in a gameworld.


So let's say you're willing to accept my theory that this lack of a responsive, interactive storyteller is the real source of the unsatisfyingness of today's computer-based roleplaying games. If that's the problem, what's the answer?

I don't know exactly. (If I did, I'd be doing that.) But I do feel pretty confident that there are a couple of design approaches that could improve CRPGs. They might be separate, and surely will be at first, or they might work together, but there are two paths I can see that lead toward better CRPGs:

1. Catalog and evaluate what human GMs do, encode many more of those abilities as rules that computers can process, and design CRPGs that can apply those rules.

2. Accept that sentient human beings can be really good at providing creative and adaptive storyplay opportunities, and design CRPGs that give some players powerful tools to act as GMs for other players.

Let's consider that second possibility first: how can a CRPG be designed so that real people can act as GMs for other players?


Obviously this isn't really optimal for a single-player CRPG.

Players may create and share content as objects, as Spore demonstrated. (And why hasn't any game developer followed up on the massive success of this aspect of Spore, which saw millions of creatures created before the game itself was even released?) But a single-player game has no one outside the player actively guiding the dynamic generation of immediate action or strategic story progression.

Integrating a person as a kind of GM is a more natural possibility for games that are designed from the ground up to be multiplayer games. Instead of spending so much time and effort trying (unsuccessfully) to replicate various human abilities to create and organize events, why not design the game to support human GMs doing what they do so well?

A hint that this is possible can already be seen in the Artemis Starship Bridge Simulator.

This is a game that allows several players, using personal computers linked in a local network, to play as the different roles (such as Chief Engineer, Weapons Officer, Communications Officer) aboard a Star Trek-like starship. The player in the Captain's role doesn't drive a computer, but instead gives direction to the other players, suggesting where to go and how to respond to the various pre-written encounters.

Currently there's little opportunity for the Captain in an Artemis simulation to generate playable content on the fly as a Game Master does in a tabletop game. Perhaps future versions will include some story management features that the Captain can select as an encounter progresses. Even now, though, this is still a useful step toward allowing an imaginative individual to dynamically guide gameplay for other players.


There's no reason why something like this couldn't also be done in an online roleplaying game like a MMORPG. Several online games already offer players the ability to script missions for other players as pre-constructed stories. Why not extend this with features that allow the mission creator to select and insert choices and consequences while a mission is being performed?

In a game like that, something closer to the example given at the top of this article might be possible. The story creator, faced with an unexpected choice by the players of his story, could pop up a screen allowing pregenerated NPCs to be selected. A slightly randomized Generic Person could be dropped into the bar, and the storyteller could either write text or actually speak to the players in character over voice chat.

While doing this, the storyteller could bring up a drop-down list of planned alternative adventures, and activate one that seemed like something the players would find more fun. He could then guide the players to this new play opportunity. To the players, all this would appear narratively seamless -- they would not know (nor would they need to know unless they wanted to) that this flow of events wasn't exactly what the storyteller had in mind for them from the start... just as a good GM in a tabletop RPG can do.

By designing a MMORPG to have real-time features for paging story elements (including NPCs, objects, and text/voice information) into and out of the local gameworld, players would enjoy an entertainment experience that is much more personalized to their interests.

Naturally there are practical questions to such a design. There would have to be enough people willing to be "virtual GMs" for all the players who just want to be entertained. The numeric progression mechanics would have to be controlled so that standings couldn't be manipulated by storytellers building "Monty Haul" missions for their friends or customers. The experience with current MMORPGs that allow static mission building suggests that these concerns can be overcome... but someone will have to try extending this to dynamic storytelling for us to find out if that scales.


But maybe it's not completely necessary to require a live human Game Master dreaming up content on the fly for a computer-based RPG to enjoy better stories. What about that first option -- what if computer code could be written that does a much better job of allowing stories to modify themselves to adapt in an enjoyable way to the unpredictable choices of one or many players?

The Storybricks team has talked about this possibility in a developer diary at [Note: I'm part of the Storybricks team and helped edit the source of this article.] To summarize it: why not design NPCs to have emotional states, and allow them access to actions that can detect and alter the emotional states of other characters?

The notion here is that "story" is about what people do for emotional reasons. By building emotion-altering mechanics that the characters of a game can use, the game itself can invent new stories for players to experience. As NPCs interact with player characters and with each other, player actions begin to have ripple effects -- friendships and antagonisms, alliances and emnities, come into existence and then change because of what players choose to do.

This effect is different than a game with a GM. It's not under one person's direct control. No one knows where the story will go. That's both scary and liberating!

This approach is also limited strategically for now. Although the current technology (like that of Storybricks) is getting better at the low-level story beats, weaving those individual moments into a thematically coherent saga still requires a human storyteller. Dynamically changing a story to achieve a unified artistic vision from the sum of many individual parts still requires the human touch... but there's no reason to think we can't start to come closer to that capability in computer RPGs as well.


Computer-based roleplaying games have reached a capability plateau. The graphics are excellent. The worlds are detailed. The core mechanics have been refined and polished. Now it's time to allow all that in-game activity to mean something, to let player choices really matter by making the story more of a collaboration between the developer and each player and the gameworld itself.

Giving creative players tools to adapt stories on the fly for other players is one way to get there. Building emotional intelligence into NPCs and letting them generate satisfying story opportunities is another way. There may be (probably are) other ways. What's important is to start.

It won't always go smoothly. The first steps are likely to be hard, as Mass Effect demonstrated. But it's the right direction to go. Computer-based roleplaying games need more emotional plausibility.

Computer RPGs are good enough now at letting characters fight each other. For this genre to survive and thrive, though, it needs to mature by letting gamers, acting through their characters, express more of the kinds of things we can do that make us human.

When game designers finally acknowledge and start to directly solve the central problem of computer-based roleplaying games -- the absence of some form of creative, adaptive mind interactively creating stories with individual players -- these games will begin to express their real potential as a distinctive and worthwhile entertainment experience.