The Untapped Potential of the MMO

This is the script from my lecture at GDC Europe 2010, on the 17th of August. Instead of repeating myself, I’ll use the summary from GDC to describe it.

“MMOs relying on the player to weave her own story becomes an increasingly distant idea as developers borrow narrative traditions from singleplayer games. The gains of the designers’ sophistication in storytelling are easy to see, but what are the drawbacks of this path? Are there alternatives to handcrafted content, and what would the benefits of such alternatives be?

We look at the losing battle of creating the never-ending game, the limitations of static content in multiplayer environments, and the dynamic systems that could be the core of our worlds.”

I’ve added a few of the slides to the script when I felt that it was necessary to convey something. For bigger versions, just click on the images. If you want the slideshow itself, it can be downloaded from my website as a Powerpoint document or a PDF document.

Take a deep breath. It’s a long read.

The Untapped Potential of the MMO

(Or Why Han Solo Returned to Finish the Fight)


Good afternoon, everyone. Thank you for attending this session after a long day. It is a pleasure and a privilege to speak to you here at the GDC tonight.

About the speaker

A few quick words about me to begin with. My name is Andreas Öjerfors. I’ve worked with the design and production of MMORPGs since 2004, having worked on Age of Conan and Age of Conan: Rise of the Godslayer. I was hired to do content design and to write story, but have since moved on as a senior designer and as a lead of design- and production teams.

Before becoming a designer, I was a games journalist and editor-in-chief of the Scandinavian gaming magazine PC Gaming World.

I am currently working with a start-up studio doing design work, while also lecturing about game design at the University of Hedmark, in Norway.

I have the perspective of a storyteller. While this is not really a talk about what you normally would call storytelling – it’s not about plot, dialog or characters – it is about the thousand little things that build up the player’s experience. Those things, everything from combat animations to the balance in a tradeskill system, is to me, among other things, storytelling. But then I am talking about the so called “player’s story”.

The session’s title

I also want to say a few words about the session itself. The session is called “The untapped potential of the MMO”. With MMO I largely mean MMORPG, but I believe quite a bit of what I have to say applies to other massive genres as well. The examples I will use are all MMORPGs, but the lessons are more general in nature. The important part is the two first letters of the acronym, MM – Massive Multiplayer.

One important point to note is that it’s mainly about western games and the western market.

After this session you will hopefully have the answer to two important questions; how do we reach the inherent potential of the MMO, and why did Han Solo return to finish the fight, in the end of Star Wars IV: A New hope?

Ok, let’s begin.

As rats in a lab

Do you ever feel like rats?

I’m not suggesting you’re all disease carriers covered in hair. Neither am I making an analogy to rats in a maze, hunting for the ever so allusive cheese – even though it might have been fitting here at the GDC. But I am making an analogy.

No, everyone in this room is human. More or less. But we share more traits with rats than perhaps most of us are willing to admit or even consider. One such trait is the neurological mechanics for reward and punishment.

This is a Skinner box. I’m sure quite a few of you are aware of it even if you aren’t interested in behaviorism, having been discussed before in regards to MMORPGs.

So what is the Skinner box? It’s a classic experiment. A rat is put into a box with two levers. Pushing one of the levers will generate a food pellet, while the other will give the rat an electrical shock.

The researchers teach the rat to push the levers by introducing rewards for each step to actually pushing the correct lever. Eventually, the rat will learn to push the food pellet lever.

The test showed Doctor Skinner that the frequency of performing a certain behavior depends on if the behavior is rewarded or punished. The action of pushing the lever has in itself no intrinsic value to the rat, but because of the associated reward he will continue to push and push again.

Different methods of rewarding the rat gives different results. A fixed interval schedule, simply giving a reward over a time period, say every 5 minutes, regardless of lever-pushing, is not a very effective method of making the rat push the lever.

A fixed ratio schedule, giving out a pellet after a fixed number of lever pushes, is much more effective. Give the rat a pellet every fourth push, and he will become a fairly dedicated lever pusher.

The most effective method, though, is giving out a pellet after a random number of pushes. Depending on a set value of how probable the delivery of a reward is, let’s call it droprate, each push has the potential to give the rat a food pellet. The rat will never know how far away or how close the reward is, and will keep pushing even if the chance of a reward is low.

That should sound familiar.

It’s true that games in general have and should have built-in reward systems – it’s a central tenet of game design – but MMORPGs have taken it to another level. Carefully constructed, multi-layered reward systems that makes sure that there is always experience or a quest update, an item, level, skill or some gold within reach, makes sure to keep you playing and to keep you paying.

We have built the virtual skinnerbox.

What was originally only a meta-game of the RPG has come to define the game experience. It fulfills something within us, but it is really fun? Or is it only there to keep us playing, and to cover for the somewhat poor gameplay and content of most traditional MMO’s?

It seems to me we really haven’t figured out what to do with these types of games yet. Many of us sense a tremendous potential in the massively multiplayer, but then we mostly just try to shoehorn previously proven singleplayer ideas into the MMO environment, stretching them over the Skinner box. It gets dressed in a thin layer of uninteresting, static content.

It works, although not very well. Yet we trust in the Skinner box, and we keep improving it. We got it down to a fine art now. I don’t dismiss the importance of good reward systems, but that we have come to be so reliant on them to give meaning to our games is in my opinion a problem.

So what we end up with is an ever-increasing reliance on the Skinner box, dressed in static content based on singleplayer traditions. This, I argue, is missing the point of the MMO. The untapped potential that the session’s title refers to is the possibilities of the massively multiplayer for players to express themselves in a shared world, if we provide them with the necessary tools. Tools powered by dynamic systems that could allow the players to co-author the narrative of their gaming sessions, together with the designers, their friends and their enemies.

To realize how we can make sure our players return time after time, is to realize why Han Solo returned to finish the fight! But one step at a time.

History of narrative

To better understand just what we are missing out on in the massively multiplayer, we first need to look at the history, the evolution, of narrative in MMO’s. We must investigate how we played when the graphical MMORPG was young, and how we play today. What has been gained and what has been lost. How we got from the “player’s story” to the “designer’s story”. I will use some defining games as examples to describe this change.

Let me first define “player story” to make sure we mean the same thing. It is the resulting story from emergent situations and player choices, as experienced by the player. The idea that what the player experiences is a story in its own right. Put this dynamic storytelling in contrast to authorative, static storytelling that we see in movies, literature and, in my perspective, far too many games.

I will not include MUD’s in this overview. A good starting point is what is generally known as the first generation of MMORPGs.

Generation one

Ultima Online, launched 1997, represents a school of design that differs from what has become the standard MMORPG model. It did a lot of things differently, partially because it was largely based on Richard Garriot’s RPG games of the Ultima series, rather than trying to make a graphical MUD.

The game was not quest-driven. It had dynamic systems allowing the players to affect the economy, and to change and expand on the contents of the gameworld. These were tools for expressing “player stories”. The player was expected to take responsibility for his own play experience, for good and bad.

So many new, unproven ideas ultimately lead to some problems. One of them was that the designers didn’t anticipate griefers – players destroying the game experience of others.

Both belonging to the Everquest-school of design, Asheron’s Call and Everquest of 1999 focused on static PvE content where the necessity of group play and a scarcity of quests forced players to create their own adventure, to tell their own story. The lack of an ever-present narrative made the players feel in control, as if they owned the play experience.

Without the expressive tools of Ultima Online, it was a bit like children not having any toys, yet playing around in the forest making their own with just cones and sticks, having a blast.

These three games defined the first generation of MMORPGs.

Generation two

There were plenty of interesting titles and features in the second generation of MMORPGs.

Dark Age of Camelot of 2001 had Realm versus Realm, RvR, where a 3-way PvP conflict between factions allowed players to battle for castles and land through combat mechanics and siege weaponry. Each faction’s victory or defeat affected a dynamic front line, rewarding the conquerors. It used other players as content in a meaningful context.

Dark Age of Camelot had a big PvE world with quests, but RvR was a real evolutionary leap in player-driven dynamics and in empowering the player.

Two years later the second generation generated two other influential games – Star Wars: Galaxies and EVE Online. Both relied on player-driven economies and conflict, evolving player-driven dynamics even further. You could say that they inherited quite a few genes from the school of Ultima Online.

While Galaxies has more or less collapsed, EVE goes from strength to strength to this day. Even though it is something of a niche title, it’s very interesting and I use it as an example later when I talk about dynamic systems.

Generation three

Things changed with the third generation of MMORPGs. Enter the period of the “designer’s story”.

In 2004 we saw two influential games. One would become much more popular than the other – World of Warcraft. The other one was Everquest 2. Both games were heavily quest-driven. The quests provided an abundance of objectives, and a constant designer-written narrative. Mostly everything of gameplay value was the target of one or several quests, and if you just followed the quest-chains you would be taken to each point of interest at the right time.

The room for, and the necessity of, the “player’s story”, shrunk dramatically.

This trend continues to this day, a foundation of most, if not all, of the AAA MMORPGs released the last few years. While this has provided us with accessible games with pretty good content, we seem to have left the road of player dynamics that we were exploring. But the road is still there, and wouldn’t be exciting to see where it leads?

I think it can breathe new life into our games – and judging from the lack of success of recent MMORPG’s, I think they need it. This is a post-Warcraft world, after all. Many player’s simply aren’t that keen on another 60 levels of limited content and then the formulaic “endgame” anymore. Warcraft did it really well – why leave your friends and community for another, very similar game?

This idea of how to present a game-world and its content has been cemented into the mindsets of large studios and publishers. I hope to see a change, because not only does it continue away from the potential this session is about – it is also filled with design- and production problems.

We’ll look at those now.

Singleplayer traditions

Ever since Everquest we have seen a steady improvement of quests and content. The problem is that the improvements have largely been achieved by borrowing old tried and proven ideas from singleplayer games. There are a few fundamental flaws in this line of thinking. The first major issue is the idea of “the lone hero”.

The lone hero

Most narrative-driven games, and literature and movies, circles around one character – the protagonist. He has to go through the hero’s journey, with its revelations and trials. The protagonist’s actions change not only himself but the world, and the people in it.

Frodo finally reaches Mount Doom and destroys the ring, freeing the world of Sauron’s evil eye, in Return of the King. Blade-for-hire samurais defends a village, and saves it from succumbing to the brutality of bandits, in Seven Samurai. Luke Skywalker blows up the death star, forcing the Empire to retreat in Star Wars IV: A New Hope.

They cause change. Without the change, these stories would not be nearly as meaningful, interesting or entertaining.

This works very well in singleplayer games. Here we can change the world based on the player’s actions. Support the lowlifes and see the city fall into decay, in Fable 2. Detonate a bomb and see a city vanish forever, in Fallout 3. But add a thousand other players into the same environment, and you’ll quickly notice it is impossible to have meaningful change in the same way.

MMORPGs treat the player as a single agent of change, despite the inability to provide lasting change in a world where thousands of players are on the same narrative path.

So we either just pretend there’s been a lasting change even though a new Death Star will arrive in 5 minutes, or we accept that the motivation for our quests are mundane and meaningless, and resign to ask the player to kill another 30 TIE-fighters and please bring their pelts back.

This also leads us to a rather bizarre paradox. Inside the gameworld, the players are all on the same point on its timeline. But the world is filled with quests, and the players are on different points on those quest timelines. This while still existing on the same point on that world timeline.

Imagine a chat between two players whom both are playing the Attack on the Deathstar quest.

Luke Skywalker: I was asked to attack the Deathstar. That is so cool! I need to buy an X-Wing.
Luk3 Skyw4lk3r: What are you talking about? I destroyed it last month. Haven’t you noticed? WE’RE FREE! HOORAY!
Luke Skywalker:  You are seriously confused. Perhaps you should speak with Yoda.
Luk3 Skyw4lk3r: Yoda’s dead, man. Yoda’s dead.
Luke Skywalker: That’s it! I’m logging off.

This obviously hasn’t stopped MMORPG developers to try to improve their content by adapting singleplayer traditions. The issue of lasting change is handled in a number of different ways.

One is to reset consequence targets. If you kill the antagonist, he will reappear as if nothing ever happened within a few minutes. If you burned down his house, the burned down state will reset into its original state. This gives a very fragile illusion of accomplishment. Even a complete MMORPG beginner will figure it out in a few hours. This is not only jarring for the person who’s burning the house down, but for those around him that will see this house burn down, reset, and burn down and reset again.

We have all come to accept this as the way things are. This, in a sense, is the natural state of the MMO. But I think that we have just become blind to how poor it really is.

Another strategy is instancing. Create a copy of a limited amount of the gameworld, and allow a group of players to have what seems as lasting change as long as they are inside that unique instance. This, of course, is only a different way of hiding that you have no lasting effect on the world.

World of Warcraft’s phasing technology is better, but yields similar results and problems.

The big problem with using instances as a solution to using singleplayer traditions in the massively multiplayer environment is that by doing so we abandon the massively multiplayer, trading it for something closing in on singleplayer itself.

This is not a solution. This is defeat, clad in the alluring colors of success.

Without lasting change, our actions are meaningless beyond the numbers game – the Skinner box. The answer is to abandon the idea of the “lone hero”. In a game where your actions are observed and experienced by real living people, you don’t need to singlehandedly save the kingdom every other hour. Trust in the meaningfulness of being part of something bigger than the individual. Community. A faction. A guild.

Make content where you’re not Luke Skywalker, the great hero, but you do fight together with the Rebels and you do make a difference. Han Solo was not the protagonist of A New Hope, but his return in the end of the movie helped in changing a defeat into a victory. Without him, Luke would have been killed by Darth Vader. Han Solo and the many other rebel pilots that sacrificed themselves that day were necessary to destroy the Death Star – it wasn’t just Luke.

In Realm vs Realm you can have change, because it’s not about a lone hero. That tells us something we’ll explore later on.

I hope I haven’t spoiled the ending of Star Wars for anyone.

The never-ending game

MMO´s are considered games without end. Developers are unable to keep up the pace with how quickly players are consuming content, unfailingly leading to a shortage of content.  Trust me, your customers will breeze through your content in much less time than you have estimated, and if you trust static content too much it’s probably not going to be a very fun experience when they’re leveling their third character.

The common solution to this is to produce enormous amounts of content. Not only is this very expensive, it is also risky. The more you plan to produce, the greater the risk becomes when content production takes a little longer than anticipated.

This leads to ridiculously expensive, over-budget and unfinished titles. Few studios have the time or the money to make something like World of Warcraft, yet so many of us try. It’s a disastrous pattern we’ve seen for years now.

It’s just not a good idea.

Content through systems

So it’s clear there are serious issues with the current mainstream design philosophy of MMORPGs. But what’s the alternative? And what of that potential of the MMO?

One answer, as hinted at in the beginning of the session, is dynamic systems – or player dynamics. It’s possible to build systems that generate content as feedback to player interaction, based not on “the lone hero”, but on the power of communities. Such systems provide a never-ending supply of content. In addition, they allow “choice and consequence”, as choices result from the consensus of a community while the consequences are generated content.

Quality requirements

There are a number of exciting and theoretical approaches to these kinds of systems, but before we start looking into some of those, I want to define a set of quality requirements that each viable system should meet. Requirements that I believe will direct us towards systems that lead to lasting, entertaining content.

The first of three requirements is challenge. The resulting content must be challenging, in two ways. Will the difficulty level be appropriate to the situation and the players involved? And will the challenge be an interesting one, from a gameplay perspective?

The second is consistency. Does the generated content both fit in and contribute to the story and the setting of the gameworld?

The third is repeatability. The system generated content must remain both challenging and entertaining as the player interacts with it repeatedly, time after time.

Direct conflict

Player conflict, like player versus player, is a powerful tool for player dynamics. Other players are unpredictable, and engaging them in combat (or other kinds of gameplay) creates new layers of meaningfulness. There is a number of existing successful PvP systems fulfilling the quality requirements, like DAoC’s Realm vs. Realm. It provided good content lasting years and years, while building community. The meaningfulness players felt added an additional reason to return to the game.

EVE Online does a fantastic job at using their PvP conflict to create meaning for their players. Not only do you belong to the tight community of your corporation, but both the rewards and the risks are so great – and truly persistent. What you do actually matters in the long run.

Beyond the fact that both examples are PvP and dynamic to some degree, they share one important trait – they are based on opposing factions, or guild or corporations. Instead of basing the conflicts on one man versus another, they put communities against each other. The result tells us that faction-based conflicts are very powerful.

There are numerous reasons why they are so powerful. They expertly exploit the motivational factor of communities, the social aspect which is the core appeal of MMOs. These are some of the methods:

By increasing the number of people working towards a larger shared goal, they make the stakes higher for all of them. They allow players to belong to something greater than themselves, charging the player’s actions with meaning. They strengthen communities, by increasing the amount of player interaction in a structured way. They build on communities instead of the “lone hero”, allowing for real change in the gameworld.

Beyond PvP

So PvP is a good way to tap into the great potential of MMO’s. Research, however,  shows what many of us already know – mainstream communities generally have a bias against PvP, preferring PvE.

Here’s the result of research done by research scientist Nick Yee of the Daedalus Project.

First some statistics showing choices of server type, sorted by gender. There is a very large preference towards PvE compared to PvP, for both genders. Now, PvE servers often contain some PvP gameplay like PvP minigames nowadays, so this only gives us a rough picture of preference.

“Appeal of content types by gender”. This shows us gameplay preference in more detail. The scale goes from 1-5, where 1 means “not appealing at all”, and 5 means “extremely appealing”. Here we can see that the distinction is less extreme, but the preference towards PvE is still obvious. We can also see that female players are generally more hostile towards PvP than male players.

So what alternatives to PvP do we have, what do we do?

PvP support roles

Looking at titles heavily dependent on PvP, like EVE online, we find interesting PvP support roles. These are the roles based on PvE gameplay, supporting the PvP conflict. The interconnected systems of EVE allow a simple miner, trader or crafter to affect the outcome of the many different wars between corporations or guilds, providing their side with necessary resources and services. The conflict between humans makes the PvE experience more relevant.

However, similar features often fail to meet several quality requirements, making them less than ideal candidates to stand on their own as a good alternative to PvP. You could question the repeatability aspect of mining in EVE, as an example.

An even more problematic aspect is that these PvP support systems usually also leave out important gameplay like combat, which in almost all cases is the central (and most enjoyable) gameplay system of the game. Nevertheless, these kinds of game systems are important if you wish to build a game-world that feels rich and alive.

Let’s take a step back to see what we’re looking for.  The system needs to:

  • Generate content dynamically through interaction with players.
  • Fulfill the quality requirements of challenge, consistency and repeatability.
  • Utilize combat or the main gameplay system of the game.
  • Be based on opposing factions.

Player conflict through proxy

Although difficult and largely unexplored, it is possible to create PvE gameplay that exploit the aspects of unpredictability and meaningfulness of PvP, through what I call player conflict by proxy. Here a conflict between factions remain, with all its power and meaningfulness, but is expressed indirectly through PvE.

The proxy is a feature or a system that acts as a target for the conflict, feeding the effort and choices of each side back to the enemy faction.

Whatever the proxy is, it should have a number of different variables that takes player actions of one faction as input, and use those actions as data that determines the shape of the produced result for the opposite faction. This is how the players affect the outcome for the enemy, providing the necessary unpredictability.

Let’s use an example. We can draw inspiration from Dark Age of Camelot’s Realm versus Realm, which has been a common example in this session.

Imagine a fantasy MMORPG, action-heavy, 3rd person 3D. Imagine two opposing factions, fighting for power in the world. The player has chosen to belong to one of the factions.

Now imagine that the game world is littered with castles. Whomever the castles belong to, have access to unique resources, bonuses and perhaps other content. Perhaps, to take down Dragon X, you might need some resource from Castle X, and to take down Dragon Y, you might need some resource from Castle Y.

Both factions can lay claim to any of the castles, but while the players of the faction that owns the castle must defend it, the attackers from the enemy faction is the faction’s NPC army. The army is the proxy of the player conflict. The players have a different role in the faction and are not part of the army, but they can shape it.

Let’s say, for the sake of simplicity, that each castle is attacked by its enemies once an hour. Normally, the attacking army is very weak and can be fended off by fewer than a handful of players.

This is where the players of the opposing faction come in. During the hour before the attack, they have the opportunity to affect the strength and shape and behavior of the army. They can attack fortified weapon caches, and bring back the weapons to equip and strengthen the army. They can take boss mobs prisoners, train them as pets with skills of their choosing, and have them join the army for the attack.

When the army attacks, the defending players must adapt to its unique configuration. Perhaps they spotted their enemies activity, and called in guild mates to defend the castle because they badly need the resources from this castle to defeat Dragon X tonight.

The army attacks. The number of waves, the kind of mobs and their equipment and behavior are ruled by the choices done and the effort put in by the enemy players in the previous hour.

So the defenders win, and keep the castle, or they lose and the castle ends up in the hands of the enemy faction. If the attackers lose, they have a war torn army to deal with.

Here you have a meaningful and unpredictable system, fulfilling the three quality requirements. It’s a challenge alright, and a self-balancing one because fewer players means a weaker army – but also a weaker defense. It’s consistent with story, as each battle ties into the over-arching narrative about the war. It’s repeatable, because the configuration changes each time.

The most important point is that the actions of the players of each side will affect the situation of the players on the other side. Player conflict through proxy.

It’s useful to make the target of the conflict something that can move between 2 or more states, so that it can be reverted and repeated. The movement between states can be binary or use a scale. In this case, a castle belongs to faction 1 or faction 2 –completely binary states.

In a different example using a scale, the castle could be anywhere from in good shape to completely burned down, with many intermediate states in between. Then you give different factions different agendas of what shape it should be, create a proxy that changes the state, and leave it to the players.

The proxy army is just an example, and there are endless ways of doing this. What I hope to have shown is that these kinds of systems can be very potent.

The fear of change

Contrary to anything dynamic, a static game world is a world of control, balance and predictability. With this in mind, the appeal of static content to designers and producers is clear. Static worlds give a sense of financial security to those in charge. Producers know how to predict the development, and designers know how to balance the systems and structure the content. We have built these static worlds for a while now.

So what are the risks of dynamic content systems – or player dynamics – and how can those risks be anticipated and isolated?

This could be the topic of an entire session itself, and should probably be, so I will keep this short.

Ultima Online was, like I mentioned earlier, a very interesting title dabbling in the dynamic possibilities of the massively multiplayer. It also stands as a warning against implementation without enough testing or fail-safes.

I have one funny anecdote to illustrate this. Origin built a system with the aim of creating a realistic balance between predator and prey. The predators hunted the prey for food. If there would not be enough prey in an area, the predators would move into the next one to hunt there, until they would finally reach the city and attack the humans (the players) there. They tested this with a low population, and it seemed to work fine.

A neat little system making the world feel more alive.

Of course, as soon as the game went live, the massive amount of players meant that both prey and predators were wiped out immediately by people looking for experience, and the system imploded without the playerbase even realizing it ever existed.

Always account for player unpredictability, through rigorous testing and design with fail-safes. Isolating risk must always be an integral part of designing dynamic systems.


So why did Han Solo return to finish the fight? Why did he come back to help destroy the Death Star?

He changed his mind because had become a friend of Luke. He did so because of what was at stake. He came back because he knew he could truly make a difference. He returned because he realized that he had become part of something larger than himself – the Rebellion.

He came back because all of the above reasons had made the conflict between the Empire and the Rebels meaningful to him. If we can capture this sense of meaningfulness, through his and others’ sense of agency instead of relying on static exposition, our players will come back for more even if we dare move the Skinner Box back to the background – to the meta-game it used to be.

Done right, it could be the beginning of the next generation of massively multiplayer RPGs.

If I was to make some sort of conclusion from all I’ve talked about during the session, it would be this:

The nature of the massive game is at odds with traditional, static content, but carries a largely untapped and unique potential for dynamic worlds that empower player communities to craft their own, epic tales.

That was my session. Thanks for listening. If you have any questions, I’d be happy to try to answer them.


Leave a Reply

Your email address will not be published. Required fields are marked *