next »

[unsilentwill]

These titles are getting more pretentious as I go on, sorry! This is not a thesis, just a point for discussion with you guys.

My personal definition of emergence is starting a world with a few simple objects repeated infinitly by means of simple rules allowing the objects to continually interact. It seems to be the fundamental principle of, er, the universe. It's a very anti-platonic style of thought. Plato said that everything that exists is defined by it's specific form in intellectual hyperspace. I use the word define very exactly here.

For example, if Socrates would ask "What is a chair?" he would respond that it is a thing that chairs, or exhibits the strengths of immortal chairhood.

When actually a chair is a concept that emerged from a cleverly balanced pile of wood, often used as for sitting on, for a step, or for burning.



So game design to me seems to start with genre, eg. I want to make a puzzle game or I want to make an MMO or platform shooter. What is a platform shooter? What would Socrates say it is? Defining is setting some ideal form and it ends up with a list of required macguffins that have to be expertly taped together, bosses, chests, doubled jump etc. This is more or less how most people design games.

There are quite a few examples of emergent design, such as Grand Theft Auto, Dwarf Fortress, Sleep is Death, Space Chem, Saint's Row and Knytt (sort of) but I can't hide my unabashed joy with the design process of Minecraft. I could honestly write a paper on the things the game does well, I'm going to focus on the crafting mechanic, item use, and world generation.

Being able to craft items means inventing your own tech tree, and is a basic human response that games tend to ignore. Being able to use any item to mine and attack and being able to carry (almost) every item is an example of simple rules applying to everything. Referring to the suicide of the old school adventure game, at a certain point the player's solution is not applicable because the designer's puzzle choice is somehow better. I want to combine the lipstick and the chicken: nope, well why not?



If you allow this creativity, the game begins to design itself and emerge as something truly new. This is actually pretty obvious considering how DnD works. It's a way for the players to take back the narrative, and designer narrative falls into lore and physics.

Thinking about emergence in a world seems to dramatically change the way graphics could be done, music could be done, but especially how writing could be done. Some of the best stories come from characters acting emergently. Authors often speak of the story writing itself by having characters (objects) being well defined (by simple rules) fully able to interact with each other. Because that's exactly what happens, humans are storytelling machines who put the phenomenon of the real or virtual world in a narrative context, even of their “own” actions creating a story of ones self.

So what are the limitations of emergent design? I don't know nearly enough about programming to say except that computer logic and language are key and emergence is philosophically opposed to the strict rigidity of those systems. An iron rod found in a musical instrument would be taken from the class “string” and instead defined by the simple (empirical) rules as “hard, sharp, flexible, etc.” Meaning any object “soft,  smooth, full of blood (blood defined as liquid, red and warm)” is a clever and gruesome way of killing someone with piano wire found in an actual piano, or maybe using it as a key to a locked door. It sounds like an infinite amount of objects need to be made, but the solution could be a procedural combination of traits that defines the appearance and use of objects. Don't limit the story by telling they player they can't do something, allow them to do it and show how negative or worthless it is. Failure is important because failure is how people learn.



Should the systems be as complex as Dwarf Fortress? Theoretically we all understand how awesome that would be, and the absurd amount of content in Minecraft modding suggests it, but no. Keeping situations closed with a fairly limited number of objects could still improve how games are played. In fact, the less realistic rules and objects the more interesting games could be as far as new worlds go. Think about it for a second, small changes in physics have very large scale effects. Or biology, consider trilateral symmetry rather than bilateral, different gravitational effects, different light patterns, etc. Allowing for new organism evolution from incredibly simple systems (shades of Incredipede) allows you to think about how life works, but also would mean aliens would finally be alien (adapted to their specific environment) instead of goshdarn blue people.

So what do you think? What happens when you let the world create itself and emergent from simple rules, and what happens when you allow the player to act within the rules? Being god, or a game designer is not about forcing your will on the people, but creating a system where their narrative affects them in the way you want it to. This can be done by changing how characters act with them, or how time and space work. Remember though, you can chop down a tree with a herring, and it might not be a good idea and have serious consequences, but allowing those objects to interact and the player to make that choice makes a totally unique, personal and memorable experience.

[s0]

I think Europa Universalis, Hearts of Iron fit your defintion as well. I know that some game design geeks who post on Gamasutra or w/e consider it  "bad form," but I love games that feel like the designer made a simulation first and then grafted "game" mechanics onto it. Dwarf Fortress is a good candidate for being my most played game ever. I like interacting with fairly open ended systems I guess.

[Falmil]

Every single interaction between objects must be defined. By assigning qualities to objects, and programming around those, it can be easier to make these kinds of systems by making the kind of objects interacting variable, but the actions themselves still have to be defined and the rules for such interactions must be made as well. A object that shoots projectiles is useless without physics and without defining the damage that can be caused by the projectile colliding with another object.
The other place emergence comes in at is AI, in which complex behaviors can be seen. Such emergence still requires identification of objects as having certain qualities and defining specific methods of interaction. It is a shame that methods of interaction cannot simply emerge as they do in the real world. Biology relies on chemistry which relies on physics which relies on unknown systems and rules aside from mathematics.

[eclectocrat]

Emergence is fantastic, but so hard to do well if you want the emergent behaviour to approximately model the real world. If you're fine with very simplistic systems, or bizarre and unexpected outcomes then it's pretty straightforward, otherwise you're better off merging a simple emergence with designer input. An approach that I see being somewhat scalable is the interaction of propertied objects (emergent behaviour) with the option of customizing results (designer input).

Code:

-- Strike object B with object A
function Strike (objectA, objectB)
    if  objectA.strike(objectB) or 
        objectB.strike(objectA) then
        -- effect has been applied
        return
    end
    if IsHarder (objectA.material, objectB.material) 
        then -- object B takes more damage than object A
        else -- object A takes more damage than object B
    end
end

BTW, this is not meant as an example of good software design, you at least want to add multiple dispatch to avoid inconsistent results.

Anyways. From a purely technical perspective, I think it's good to start with a very simple system which you can always under all conditions fall back to if there is not a more specific implementation to use. This core will provide some level of emergent results, since if the game can't make sense of an interaction it just goes back to basic rules which give a consistent (if not altogether realistic) result which can in turn be built on for additional interactions. So in the above case you can take a hammer and strike a piece of plywood and the plywood will take damage until it breaks, despite there being no specific designed instructions that plywood should break when hit with a hammer.

It's up to the designer to implement more specific consequences to interactions. This is the only way to maintain a rough isomorphism with the way the real world functions, because we just don't have enough horsepower to realistically model the myriad interactions of matter that are possible. So if a designer makes a chisel object, it can include a function to check if it's being hit with a hammer and apply a chisel effect to whatever it's held against. And because you have properties as a base to fall back on, then the chisel can even check if it's being hit by an appropriately heavy object to apply it's effect anyways, perhaps with some penalty for not using the most appropriate tool.

So what do you think? What happens when you let the world create itself and emergent from simple rules, and what happens when you allow the player to act within the rules? Being god, or a game designer is not about forcing your will on the people, but creating a system where their narrative affects them in the way you want it to. This can be done by changing how characters act with them, or how time and space work. Remember though, you can chop down a tree with a herring, and it might not be a good idea and have serious consequences, but allowing those objects to interact and the player to make that choice makes a totally unique, personal and memorable experience.

I'm obviously coming from the perspective of someone who wants to create an emergent world that roughly resembles the one we live in. If you're going for very simple rules, or totally abstract worlds, then thats a whole different story, and my only answer is that don't expect anything whatsoever from an emergent system. They can be totally unpredictable. Which is awesome.

PS. Have you read Wolframs "A New Kind Of Science"? He comes off as an egotistical ass, but it's still an amazing look at complex patterns emerging from very simple algorithms.

[unsilentwill]

Thank you for the responses guys, this is probably one of the most interesting topics in game design for me, and I really would like to hear more opinions coming from different disciplines. Emergence is indeed a topic that comes up again and again when looking at the world, so it can look very technical. I tried to put it simply so we could discuss it, but I ended getting very philosophical very fast. If there's anything you want clarified let me know because I really want to talk about this. Tell your friends!

What I think is "bad form" is to ignore the really hard problems with current game design of replayability, cost of visual/musical assets, and designer vs. player freedom, which if I may be so bold emergenct game design seems to solve.

I had forgotten to mention Inside a Star Filled Sky and I feel bad about that. He nailed it with that game by using almost entirely emergent procedures to make a shmup way more interesting. The downside with that is that it become entirely abstract, which is where design comes back in as eclectocrat said. Making a new system actually look like something take a lot of direction. Also, I'd much rather time be spent on faster calculations than faster rending if fractal patterns could look as amazing as they do, designed in seconds rather than weeks.

The reason why I really want to talk to all of you about this is the oddity of why emergence is only being noticed recently. Wolfram said (I had watched a few of his talks, but haven't read his books) people didn't think about letting simple systems repeat because they felt they knew what the outcomes would be. This change in attitude is a step in bettering the communication between designer and audience (that all art struggles with) where making information more free to interact, and giving the player more freedom seems to produce better, or at least more complex responses than what designer intended. This change hasn't quite happened in game design to the degree that is should. Platonically defined objects by their intended function is way less interesting that the system of Aristotle where objects are defined by their natures.

Put shortly, I hate it when Macs don't let me change a setting or let me do something I know it can run. The same thing happens in games all the time where the action is already existent in the game (adventure games had plenty of action options defined Falmil) but the option is not there because there is no animation or game state for the result.

Apologies for the wordiness. Please, all thoughts are welcome.

[mirosurabu]

What does it mean for "simple systems to repeat"? I think I understand the overall message of your post, but I cannot quite grasp this sentence yet.

[1982]

Well you quite well described how all game engines should build. I totally agree in every aspect. I also wrote something like this in some other thread, but don't remember where. It was about a "adventure" game happening inside nuclear facility, where everything that happens is based upon nuclear power plant technical simulation and human behavior simulation. How complex it would be - well there are levels how complex you want/could to make it... But the game would eventually create the story by itself.

It is sad that finally when we have the tech to actually build very "complex" games upon this simple philosophy, we rarely see anything. Most of those games which however work like this, excels everything with quite a margin.

[s0]

This change in attitude is a step in bettering the communication between designer and audience (that all art struggles with) where making information more free to interact, and giving the player more freedom seems to produce better, or at least more complex responses than what designer intended. This change hasn't quite happened in game design to the degree that is should.

Well there SEEMED to be a definite trend towards it in the 90s and early 00s, particularly in the PC world. At the time, I thought more emergence, nonlinearity, openness etc. were going to be the future of gaming. It seemed like the logical next step. Unfortunately, that kinda fell flat when the whole "Hollywood envy" thing started.

[gimymblert]

it was medal of honor, everybody was saying script was dead ... then that D-day normandy thingy happen

[s0]

Emergent gameplay is coming back though! Remember that Minecraft, a game with no scripted events, made its creator a multimillionaire. The demand is definitely there.

[baconman]

Does anybody else think it's quite possible to coincide deliberate design with emergent play? Because I'm personally looking to create a balance between the two, not have one come at the expense of the other.

My level chunker project is geared towards combining definitive level design with procedural-gen layouts that can create cause-effect results in neighboring sublevels. I've got a group of character designs with very different personalities and motivations that will inevitably lead to friends and foes, sometimes in unexpected ways; along with a variety of inherent abilities and fighting styles to differenciate them, and a definitively scriptable way of reanimating each of them (thus, once I get a template together, inserting new charas should be clockwork. And I have design work done for an output-editable dynamic music generator as well, although it stems from simple music principles.

Putting all of this into actual code, however? That's a bitch. And so is actually producing the desired effects, since coding and scripting is not always as straightforward as one would like it to be.

[DavidCaruso]

"Emergent gameplay" = an industry buzzword for rulesets that appear to be simple while actually allowing for complex and/or nuanced possibility spaces as the rules interact with one another. So yes, of course you can design a system which has a relatively low barrier of entry ("simple" ruleset) but high complexity and nuance. John Conway did it, after all, and so did the designers of Chess and Go (and GTA). You'll need a lot of testing though, of course.

[mirosurabu]

What is exactly emergent in GTA?

[s0]

"Emergent gameplay" = an industry buzzword for rulesets that appear to be simple while actually allowing for complex and/or nuanced possibility spaces as the rules interact with one another.

Yeah it's a buzzword, just like "procedural generation," but unlinke "proc gen" (ugh!), it's actually useful. I use it as a shorthand for situations that arise "naturally" out of a game's rules as opposed to being scripted. It has nothing to do with simplicity or apparent simplicity or w/e. I'm pretty sure e.g. Dwarf Fortress and Europa Universalis, two games whose gameplay is at least 90% emergent, qualify as complex by any meaningful definition.

[DavidCaruso]

"Emergent gameplay" = an industry buzzword for rulesets that appear to be simple while actually allowing for complex and/or nuanced possibility spaces as the rules interact with one another.

Yeah it's a buzzword, just like "procedural generation," but unlinke "proc gen" (ugh!), it's actually useful. I use it as a shorthand for situations that arise "naturally" out of a game's rules as opposed to being scripted. It has nothing to do with simplicity or apparent simplicity or w/e. I'm pretty sure e.g. Dwarf Fortress and Europa Universalis, two games whose gameplay is at least 90% emergent, qualify as complex by any meaningful definition.

Everyone in this thread was talking about simplicity and building complex systems out of fundamentally simple rules so I figured the definition had changed again (last time I heard the term being used it was about "designer intent," and before that when I first heard the term it was about unpredictability). "Procgen" is probably a more useful term than "emergence" because when someone talks about procedural generation I at least know what they're talking about (dynamic randomly generated content as opposed to static predefined content) whereas when someone talks about emergence they could either be talking about designers' intentions, unpredictability, complex systems arising out of apparently simple rules interacting, unscripted shit going down, or probably 10 other things. If I wanted to say that a particular scenario wasn't scripted then I'd just call it an "unscripted scenario" =P (which is still technically a misnomer because nothing in a game is really unscripted etc., but you get the point).

[s0]

This is a bit of a side topic, but I think proc gen is dum if you're just using as a fancy term for randomization, because procedural generation doesn't necessarily imply randomness. A lot of modern 3D games use procedural animations but people don't call them "procedurally generated" games or w/e. I've been playing roguelikes since the 90s and no one used the term "procedural generation" then. Its current usage only became widespread during Will Wright's  pre-release hype campaign for Spore IIRC.

Aaaanyway, I think people are misinterpreting emergent gameplay when they say it's about designer intent. It doesn't literally mean gameplay the designer didn't anticipate. It means gameplay the designer(s) didn't explicitly include but implicitly arises from the game's rules. Whether they had it in mind while designing the game is unimportant.

I see a "script" as a kind of game designer's deus ex machina [insert obvious joke about cyberpunk cop with sunglasses here] that breaks the rules of the game to alter its flow and follows NO mechanically-derived rules itself. Cutscenes and (the majority of) quicktime events would be good examples of "scripted" elements.

[Belimoth]

I don't even know what 'emergent gameplay' is anymore. I thought I had a pretty good handle on it, but then people started using it to describe any game where you can combine things with other things or shoot a barrel occasionally.

[unsilentwill]

Luckily I wrote a little bit on what I mean. Emergence as a theory for design rather than a bullet point for an FPS. It's something that should be thought of in the beginning. In sum, if you have a series of objects define them by their simple attributes, a let the simple rule, usually physics, constantly apply to all of them, letting them all interact unexpectedly.

[s0]

I don't really understand why they have to be simple though? Why can't complex objects also interact in interesting ways? This may depend on your definition of "object" obv.

I the "emergent" design philosophy is also related to the whole linearity vs nonlinearity thing. It's about how much you "railroad" players and tell them what to do vs how much you just design a system for them to interact with, give them a bunch of goals (implicit or explicit) and let them use the system in any way they see fit to try and achieve them.

[unsilentwill]

Well right, emergence just states that complexity it just a simple concept repeated on itself like a the appearance of a snowflake or a coastline. The complexity of say a city is still the interaction of smaller bits following an exact pattern, and it appears very complex but can be repeated through simple rules. I'm a poor teacher, but there's Wikipedia, Radio Lab, or Wolfram himself to explain the marvelous concept.

And yeah it's part being honest with the player what he can or cannot do. If you give someone a gun and say certain things can't be shot, that's awful and yet that happens all the time because the objects in the game are defined as immortal walls rather than the brick they appear to be. If you want them to be barriers make them out of something that can't be blown up, don't lie about it.

Emergent design is making objects with properties rather than using smoke and mirrors and giving the player rubber props.