Tuesday, June 17, 2008

"Living World" Massively Single-Player Games

I've been thinking lately about what a massively single-player computer roleplaying game -- a Living World -- might look like.

The idea here would be to create a very world-y game like Oblivion or Two Worlds, but which is so large that it would take a player years -- in real time -- to explore it all, even if he plays the game a few hours every day. The play experience of a Living World game would be that it becomes a sort of alternate reality that the player could return to over and over again for years.

And the point of this is to offer an alternative to the "played it, beat it, sold it" experience of games today. Instead of being like cotton candy, insubstantial and quickly forgotten, an extremely world-y single player game would be a highly memorable place that the gamer could return to at any time, whether for new adventures or for a sense of coming home to something familiar.

I first started really thinking about this Living World concept after playing Oblivion. (Certainly Spore has been on my mind since I first heard about it as well, but while it has some of the qualities of a Living World game that I discuss below, it's not quite the RPG I'm thinking of.)

The world of Tamriel in Oblivion was large (relative to other games), seamless (while aboveground and outside of cities), well-populated with locations and quest-giving NPCs, and full of creature encounters. There was even regional weather. All of these things combined to produce a game environment that felt like a place. Being able to keep playing after the main quest line had been completed helped as well.

But there were still limits. Although the game world of Oblivion felt large in the early stages of the first time playing through it, it could be completely explored over a few weeks of play. And once the main quest and all the side quests had been completed, the only novel content remaining was the eventual respawning of previously cleared dungeons with random enemies (as opposed to "named" NPCs with unique goals and dialogue). While there were small user mods, those tended to be GUI tweaks, not full-scale new content. And although Bethesda supplied a couple of expansions (including the significantly large "Shivering Isle" expansion), they clearly indicated that they didn't want to spend any more time providing such content. These things aren't flaws, exactly -- they're simply limits to a game that was designed with finite content.

So what if that design choice could be challenged? What if it were possible to build a game whose various forms of content would take even a dedicated player years to fully explore? What features, if blended properly, would add up to a single-player Living World RPG that could be not just played in but lived in for years?

Let's consider some of the design possibilities of a single-player game intended to last for years.

Enormous area

Not counting dungeon areas, Oblivion offered a large amount of land to explore. Once the player character was leveled up, however, it was possible to run from one side of the world to the other in less than an hour. Two Worlds was considerably larger, but even it could be traversed in an evening and fully mapped in a month of evening and weekend play.

The kind of Living World game I imagine would -- relative to the player character -- be about the size of Europe (or perhaps a bit larger). Alternately, if we were talking about a science fiction setting rather than the usual medieval-era fantasy setting, I'd look for something on the order of hundreds of thousands of planets.

The point of this kind of scale would be to insure that the game contains unknown places for a long time. Having a total area that's very large allows the possibility of multiple political entities controlling meaningfully large territories, which opens up opportunities for epic storylines. Rather than the entire gameworld sharing a single culture (and language), in a huge gameworld there could be places where the player character simply would not be able to go (at least for a while) due to limitations on travel speed, cultural biases ("you can't come in; our country is at war with your country") or simply not knowing the local language. This allows for a very extended process of world exploration, giving the player a reason to keep coming back to the gameworld over a long time.

Passage of time

As in a MMORPG, time in the Living World CRPG would continue to pass even when the player is not actively playing.

This has pros and cons. On the plus side, it supports the feeling that the game world is a real place whose existence is independent of the player. It also provides a mechanism for refreshing content -- as the player explores the game world, changing it by his actions then moving on, explored content can be repopulated or even replaced.

Most interestingly, the game world could change over the long term. As years pass in the game world, NPC behaviors in the aggregate could alter some parts of the game environment:
  • in prosperous areas, some villages would grow into towns and new villages could be created
  • forest areas would be cleared for farming
  • in poor areas, or as a result of famine/disease/war, some settlements could be abandoned
  • abandoned places could be occupied by barbaric NPCs or reclaimed by forest
  • NPCs could die of old age and be replaced in their roles (perhaps by their descendants)
  • over the long term, large-scale social changes -- especially invasions -- could change the political map of the world
  • over the very long term, natural environmental changes -- temperature, rainfall -- could reshape creature and NPC populations
(Some of these concepts obviously apply to a fantasy-based Living World, but similar effects could be generated for a science fiction gameworld.)

There are a couple of possible downsides to this continuous passage of time. One is that long-term changes in creature and NPC populations could bother gamers who prefer a gameworld to be a stable place. For example, allowing named NPCs to die could make the game less fun for narrative- and roleplay-focused gamers who form emotional connections with some strongly-realized NPCs. Providing in-game tools for discovering and understanding change, such as genealogy charts and "newspapers," could help players whose enjoyment of a gameworld comes from feeling connected with the characters in it.

Another potential negative is that it will probably be necessary to impose a creative stasis on a Living World CRPG. It might be possible to allow for a very, very small number of NPC-generated inventions (political, social, and economic as well as technological), but allowing NPCs to be creative would probably wind up changing the game world too much. At some point there will have been so much progress that the game would have to end. A medieval-era fantasy Living World, for example, should probably not be allowed to invent itself into a Renaissance -- instead, the cultural tools available to NPCs at the start of the game should be what subsequent NPCs can use regardless of how much time passes.

(Having said this, the possibility of implementing subsequent eras, with their modifications to code, art, sound and other assets, as major expansion packs to the original game might be worth considering.)

While this restriction on NPC creativity is likely to make the game less satisfying to Simulationists, it's necessary in order to avoid having to develop assets at launch time that players might not see for years, and to prevent the game from ending by reaching the end of time!

Environmental, ecological and social simulation

A Living World game probably needs to have extensive simulation capabilities, both to provide dynamic content for exploring while the player is in the game and to allow the gameworld to appear to persist (and change) when the player isn't in the game.

Environmental simulation means modeling macro-level physical phenomena such as day/night cycles, topography, temperature, and geographically appropriate weather.

Ecological simulation will reflect the effects and movement over time of groups of living things as they live, reproduce and die in large numbers. On a planet, this will include both plants and animals -- for example, given environmental conditions in a particular area a forest might over many years become a prairie or a desert, and vice versa. The elimination of predators in an area might lead to overpopulations of their normal prey species. Humanoid habitations would move to follow more attractive conditions and better food supplies -- a player who is out of the game for a while might return to find that a village they once visited had been abandoned. (There might need to be some places that would be exempt from such changes in order to insure that a player has a secure place to store valuable items.)

Social simulations would model high-level cultural changes. These would include changes in the opinions of NPC factions toward each other, and possibly even the automated extinction, merging, and creation of factions as small as towns and as large as nations. Social simulation would also model large-scale economic, military and cultural behaviors. This level of simulation would be extremely useful in creating a dynamic gameworld that can change enough to be interesting over the long (real-world) term.

Constantly refreshing content

This is probably the single hardest problem that would have to be solved to make a Living World game. While it's nice to think that the game would ship with so much content that no player could explore it all in two years, some players would race to see everything at a very shallow level. They aren't necessarily the kind of player a Living World game is meant for, but there's no reason not to provide some content they can enjoy as long as it's made clear what kind of game this is.

So it would probably make sense to provide some mechanism for refreshing the existing content in a Living World game. The three most plausible approaches to accomplishing that goal are:
  • dynamically-generated content
  • user-generated content
  • developer-generated content
Random dynamically-generated content that looks good and plays well might be possible for some definitions of "content." Explorers in particular might enjoy trying to solve different kinds of randomly-generated puzzles. More conventional content, however, such as quests given by and including memorable NPCs, would be very difficult to create in this way -- for example, where would the audio dialogue come from for an NPC who's part of a dynamically generated quest?

Allowing users to easily generate their own content, or allowing users to generate content that can be easily shared, is another possible approach to providing continuously refreshed content for a Living World CRPG. Nothing makes this approach impossible, as the success of mods for Doom, Half-Life, Neverwinter Nights and Morrowind have demonstrated. A more recent game, Spore, is actually built on the premise that user-generated content will be used to populate the individual player's gameworld. It's possible that a Living World game could be designed to allow some combination of the world-building tools available for games like Neverwinter Nights and Morrowind and the content-rating system of Spore to refresh some game content with high-quality replacements.

But developer-generated content may be the best way for a Living World game developer to go. As with MMORPGs, perhaps providing a Living World game with a steady stream of new content and enhancements long after the core game is purchased and installed would be a successful business model for a developer. Players would be assured of getting high-quality new content (including professional voice acting), while the developer would gain the long-term revenue stream that would help to pay back the high cost of developing a Living World game.

Content adapts to player's playstyle

To enable a Living World game to be attractive to more than just a few hardcore Explorer-types, consideration should be given to finding ways to populate the gameworld with forms of content preferred by different kinds of players.

How this is achieved will depend on the content generation method (as described above). If content is dynamically generated, the generation system could either automatically create new high-level content (which gets broken down into specific tasks to perform or other opportunities for player action) based on the kinds of things the player has chosen to do. A player who, when given the choice, typically solves game problems by killing and destruction could find that the gameworld over time becomes more warlike, presenting more opportunities for that player to be rewarded for doing what he enjoys. Similarly, a player who usually solves problems through listening and negotiating with NPCs might tend to receive content that emphasizes and rewards diplomacy and understanding of emotional/story ideas. Players who approach problem-solving through economic power or system redesign could likewise find their gameworld adapting to generate new content that tends to be most rewarding to those styles.

If new content is mostly user-generated and downloaded, then it would be useful to tag that content to define what kinds of gameplay rewards are available. A user-generated mission to locate and destroy a large, dangerous pirate vessel, if written primarly to reward direct combat action, should be marked to indicate to potential downloaders what kind of gameplay to expect. This would allow users of a clearinghouse for content to select the new content they find most enjoyable. (I would hope, however, that any system allowing users to generate their own content would allow for multiple solutions. To reuse the pirate ship example, maybe players could negotiate with NPCs to gang up on the pirates, or could pay mercenaries to take out the threat, or could study intel to identify a flaw in the vessel's design that could allow a small strike team with the appropriate technology to destroy the ship.)

And if new content is created by the game's developers? This will depend on the granularity of that content. If the developers can create lots of small bits of content, then the clearinghouse approach of tagging content by type might still be effective. If however content tends to be produced more like that found in MMORPGs, where it tends to be big, detailed, polished, and rare, then players will simply have to take what they get. In this case, it will be up to the developer to try to insure that multiple playstyles are respected. Even in this case, however, it could be pretty amazing if the gameworld did adapt in some ways to the preferred playstyle of the player.

PC based

The size and scope of a Living World game, as well as the likely need to continually refresh content, make the PC platform a natural choice. While some consoles are moving toward online components, the PC is already there. The architecture of the PC also allows for much larger spans of world-space, as opposed to having to break up the world into pieces to fit into the memory of a console. While games like Oblivion demonstrate that this can be done with a minimum of interruption to the player, scaling a Living World game to the PC platform by design eliminates the need to have hundreds or even thousands of small areas at all.

With regard to the question of piracy that supposedly is rampant on the PC platform, designing a Living World game to benefit (like online games) from regular downloads of new content shifts the revenue stream to a long-term model. While not eliminating all losses due to piracy, the downloadable content design reduces the need to earn back all money purely from sales of the base game.

Maintain consistent and appropriate challenge level over years of play

One problem that open-world RPGs like Oblivion and Two Worlds have is that players, by taking the many side quests offered, can effectively level up their characters to the point that subsequent challenges are too easy. This is a problem that would be particularly acute in a Living World game that could be played for years after a character has effectively hit the game's level cap.

The usual solution in these games (as well as in Dungeons & Dragons, the prototype for class/level-based RPGs) is simply to retire the old character and start a new character. While that might be workable in a Living World game, it might be preferable to consider alternatives. How could a long-term game be designed so that players want to keep playing for years with a single familiar character?

One approach (used by the original Star Wars Galaxies) would be to use a skill-based system rather than a class/level model, and permit characters to "forget" known skills in order to learn new skills. Depending on how this is implemented, players could tailor their characters for different styles of play depending on their preference. For example, Sony is designing The Agency to allow players to effectively "change classes" by simply swapping their outfits. Something like this could help to keep a Living World game fresh for a longer period of time.

Another possibility is to design the game to have less emphasis on character abilities that can be leveled up. This could be accomplished in many ways:
  • slow down skill progression
  • substitute player skill (reflexes, puzzle-solving, etc.) for character skills
  • minimize the "game" aspect in favor of exploring a vast and complex world
One or a combination of these approaches (or some other design solution) might be worth considering for preventing characters from becoming so capable so quickly that players believe the game is "over."

[EDIT: 2010/03/27] Gameplay by "inhabiting" non-player characters (NPCs)

Yet another option, and perhaps the most interesting, would be to eliminate level-based character advancement entirely. Rather than the standard RPG model of creating a particular character and then leveling up that character through progressively more difficult challenges, something like a Living World might not need to adopt that standard high-level approach to gameplay.

Instead, I've been thinking it might be more appropriate for the primary gameplay system to be designed to highlight the distinctiveness of the Living World. And what comes most strongly to mind is not giving the player a character at all, but instead making gameplay about "jumping into" existing NPCs.

In this model (which, yes, now that I think about it does somewhat resemble the notion behind "Quantum Leap"), players would first experience the gameworld from a disembodied perspective. They would be shown the Living World from a great distance; the camera would then zoom in to various regions and then to individual NPCs going about their lives. The player would be shown how to take control of two different NPCs as a tutorial. After returning to the big picture view of the world (to impress on the player the scope of the Living World), they would then be free to begin exploring the world by inhabiting the NPCs of their choice.

This is where the core gameplay of the Living World would be found. Every NPC would be defined to have a particular role. (Role definition would be part of the toolset for building the content of the Living World.) Players would be able to observe or inspect an NPC to see the role he or she has, and then choose whether or not to inhabit that NPC.

Inhabiting an NPC immediately gives the player access to all the skills -- defined as gameplay activities -- that are associated with the role to which that NPC was assigned. So if you want to fight monsters, inhabit a Ranger; if you want to chase thieves, jump into a City Guard NPC; if you want to practice crafting, inhabit a Blacksmith or Baker; if you're looking for economic gameplay, inhabit a Merchant; and if you feel like trying your hand at the very difficult game of city or kingdom management, you would be able to inhabit a town's Mayor or even the King of an entire nation.

As noted above, in each case the kinds of gameplay available to you would depend on the role of the character you choose to inhabit. These gameplay activities would need to be predefined by the developer as selectable actions which use and/or affect objects inside the gameworld. The "Ranger" role, for example, might be defined to optimize skills (gameplay actions) such as Shortbow, Shortsword, Tracking, Herbal Medicine, and Stealth. Meanwhile, the "City Guard" role could be defined as having special training in Shortsword, Tracking, Negotiation, Perception, and the Guard badge which allows that character to summon other Guards. Other roles would have their own appropriate skill optimizations.

In addition to roles having particular skill optimizations, roles would also be keyed to preconstructed gameplay content. In other words, the role of the NPC you choose determines the kind of prescripted gameplay content offered to you. If you choose to inhabit a Ranger, you would not only be free to explore the gameworld while doing Ranger-y things, the act of inhabiting that NPC would also activate any number of world events (either automatically or at the player's discretion) in which Ranger skills could be particularly useful -- say, a monster invasion, or finding a lost child for some villagers. The same would be true for every role. Even highly mundane roles such as Baker would have gameplay events (e.g., baking challenges) scripted for them that would be fun for someone who voluntarily chose to be a Baker. (Note that this design integrates into the "epic storylines" discussed in the next section.)

The assignment of NPCs to particular roles would need to be integrated into the social dynamics of the gameworld. Because the Living World models the birth, growth, and death of individual NPCs (possibly only at a statistical level except where the player has traveled or currently exists), the mechanism of children maturing into adults would need to be fitted into the role system. For example, a village experiencing dynamic growth would encourage children to fill high-priority roles as openings (due to death, injury, retirement, or possession by the player) occur, then useful roles, then expansion-oriented or support roles as the group's resources and rules permit. A village or social group that can't or doesn't care to continue, on the other hand, would probably find most of its children emigrating to cities or taking on "solo/loner" roles. In other words, the Living World needs to be designed so that role assignments to juveniles are made dynamically based on a determination of whether to maintain/enhance the existing social group or to break it up in favor of forming new groups elsewhere.

One other note on the gameplay design approach of "inhabiting" NPCs: Because the Living World would simulate not just large-scale social movements but personal social structures as well, in many cases the NPC whom the player chooses to inhabit would be part of a social group -- a member of the city guards, or the blacksmith for a village, or a husband or wife who is the parent to some children. This raises the question: should those NPCs realize when someone who is part of their social group is "inhabited" by an entity with the unique power of possessing people's bodies and minds?

I very much like the idea that these NPCs would definitely know when someone in their group is inhabited by the player, and that they would be able to react toward the player according to their beliefs and fears. It seems reasonable that stories of this kind of possession, occurring for thousands of years, would be part of the legends and histories of all the peoples of the Living World. Would you be considered a god? or a demon? or perhaps just a very talented sorceror? If you choose to inhabit an NPC who, in addition to her gameplay role, is a wife and mother, how might her husband and children react to you when they realize that she no longer exists as herself and that you might do anything at all to her, from taking her across the wilderness to getting her killed? These relationships, I think, would also be an excellent opportunity for prescripted gameplay activities to be activated, with the twist that they could have more than the usual amount of emotional resonance -- would you choose to risk the life of an NPC whose children are begging you not to take her away?

Finally, regardless of which approach to gameplay is taken, it will be crucial to very clearly market a game like this as a long-term game, as a world to be lived in. A Living World game needs to take advantage of its unique features by marketing those features. It must not be described to prospective players as a typical shooter or RPG with a developer-defined story that players should expect to "beat" in a few hours.

Truly epic storylines

An opportunity afforded by a Living World game would be the chance to tell significantly larger and more complex stories than those found in other kinds of games. Because the action can take place over several years, the story doesn't have to unfold quickly. And because a Living World game is extremely large, both in terms of geography and NPC cultures, stories can truly be epic in scope.

An interesting possibility here would be to provide players with several outlines of epic stories. Once an epic is selected, large-scale world-changing events are set into motion, and before long the player will be impacted in some way. Epics could be story-based (the princess is in love with a prince of the neighboring enemy country and needs your help), action-oriented (barbarian/alien invasion), or adventure-driven (find a way to appease the angry god who's disrupting the world with volcanos). But in all cases, starting an epic will reshape the physical and/or social structures of the game world in some way that affects (and can be affected by) the player's character.

Customizable rules

As noted above, some players might prefer a world in which NPCs don't die. Other players, however, might enjoy seeing NPCs who die be replaced. Players might wish for more or less aggressive NPC cultures, more or less environmental variation, and so on.

It might be interesting to consider whether a Living World game that's intended to run as a highly simulationist single-player experience could allow for these aspects and others to be optional. Another question would be whether some or all such customizible rules can only be set at the start of a new gameworld, or if they can be changed while a game is running.


I'm just getting warmed up in thinking about these possibilities, but so far I'm feeling that there might be something here, that this could be a novel form of computer-based entertainment that some number of gamers would embrace enthusiastically.

Obviously there's a lot of work to be done to put some real meat on these bones of an idea. In particular, the business case needs to be addressed -- how could any publisher be persuaded to bankroll an idea like this?

I'll try to speak to some of those questions in future blog posts.


  1. Hey, I'm from the TUG forums and I thought I'd check out your blog. An excellent post. I have also found myself fantasizing about something similar (sandbox RPGs that dynamically simulate social change) on the Project Eternity forums recently, though it probably wouldn't fit with the narrative focus of the games that inspired Project Eternity. Seeing as you had this vision years before I did, I thought I might ask a few loosely related questions.

    Firstly, are there any sites or forums that you have found for discussing these kinds of ideas for game design? There are obviously many sites on video games in general, and some more design-oriented sites on tabletop role-playing games, but I have yet to find a good place for people to share their theories about game design. I'd love to enter my own ideas into the mix, but I am not sure whether I'm up really to the task of starting my own personal blog from scratch.

    Secondly, and more relevant to this particular post, how long do you think we have until this approach to "sandbox" RPGs becomes technically feasible? Are there already games approaching this sort of thing beyond the level of Skyrim, or is it a case of the developers not believing that a market exists for such games? I'd really love to see something like this, but mainstream games don't really appear to be moving in this direction as far as I can tell.

    Anyway, great post again and I will check out some of the rest of your stuff!

  2. Thanks for the nice words about the Living World concept. This is a design notion that I just cannot get out of my head. If I can't make it, I describe it here in the hopes that someone else can and will.

    To your first question, I'm not aware of any sites that are specifically about highly worldy games. I've frequently been considering creating a new web site explicitly about exploration and Explorers in games -- I believe there's a good market for that. As it is, I already have multiple projects going for which none of them can be given enough time -- sigh.

    In the meantime, I can refer you to Brian "Psychochild" Green's blog site at http://psychochild.org/ . Brian is an experienced game developer and highly competent design thinker (and a darn nice guy), and you may find a lot of good references on his site to your interests. Raph Koster's blog is also frequently provocative.

    To your second question, I don't think we are yet at the point where it's financially feasible for anyone who wants to create a Skyrim-like game to do so. Even if you went with simpler graphics and procedural content generation, the sheer amount of "stuff" that has to be created is still, I think, prohibitive for one person or even a small group.

    That said, I do believe we're on the cusp of getting to the point where that kind of game can be made by a smallish team. It seems like every day now I'm hearing about a new content creation tool or a new low-cost RPG engine. At some point soon, I think we'll hit a critical mass of such tools, and a small group will finally have a decent chance of being able to build a gameworld that is (as we've recently discussed) both broad and deep. Such a game (per the Living World concept) will have lots of places to see and things to do, and it will be populated by critters and NPCs who behave in sufficiently plausible ways that it will be fun to "live in" that world for weeks and months at a time.

    I do think that procedural content generation will have to be one part of that formula (for anybody smaller than Bethesda). One day I'm going to write a blog post that's a survey of that subject, but for now you might be interested in Googling what people like John Whigham and Miguel Cepera are working on. The Procedural Content Generation Wiki is also becoming a pretty inspiring resource: http://pcg.wikidot.com/ .

    I hope you enjoy the rest of the site here -- do let me know if there's anything else you'd like to see more of!

  3. Thanks a lot for all of the resources, I'll check them out.

    I suppose this makes me a bit guilty of acting on impulse, but I created the structure for a basic forum yesterday. I have also been looking for a gaming community that actively promotes crowdsourced projects, open-source games, and other grass-roots approaches to gaming, but my level of success there has been similar. I suspect that you are probably familiar with The Forge, which was an RPG-oriented theory site that spawned the GNS theory. It closed down before I had the chance to participate, and it seems as if there's not anywhere active currently that serves a similar purpose. I'm also inspired by the unreliability of game reviews in mainstream gaming media.

    Anyway, I'm going to try my hand at building a gaming community based around video game (and specifically RPG) theorizing and promoting alternatives to the somewhat unideal AAA industry approach (i.e. games like TUG and Project Eternity). You sound busy, so I understand if you don't feel you can commit to participating, but if you like the sound of it- or if you just feel like offering a bit of advice about how to develop a forum conducive to theory- you can see what I'm working with at http://progressivegaming.boards.net/. At the very least I'd love to link your blog if you wouldn't mind.

    Thanks again.

  4. This article having matured, and coming out when I was a wee lad, and before Windows 7 (Lions, Tigers, and Vista oh my), I wonder if the entry level for Indies to jump aboard this project and tackle it is possible. With better computers, more devs, better tools, and a lower access level, is it feasible? Or is this best left for the big guys? I'm excited to see this great idea spawn into something bigger, and I long for the day it does.

    Thanks for the incredible brain food of an article.

    1. Glad you enjoyed it. This is one of those "long-term dream projects," so I'm always happy to see it rediscovered.

      I think the basic idea was implementable when I first proposed it. The part that really drives up the cost (time and money) is first-person 3D. That perspective feels important to me for maximizing the feeling of scale in a living world. But I have to acknowledge that it's the expensive part.

      A team that dropped first-person 3D as a requirement in favor of some other representation could, I think, make this game now in a year or two, depending on the number and detail of the character roles and the size and dynamic features of the world.

      I believe this kind of game will happen. Someone will make it. The only questions are who, and when.