Simulationism

This is from comments on the Terra Nova blog in 2004.

To bring up some hoary MUD-Dev terms, in part we’re discussing the simulationist approach. Now, obviously, to some degree, these are all simulations. But simulationist refers to attempting to model basic behaviors in order to permit higher-order behaviors to emerge. What Cory describes for clouds and pianos is exactly a simulationist approach. The opposite approach is essentially stagecraft; you make something that looks like a piano, and behaves like a piano, but is not actually a simulation of how a piano works.

Searching for “simulationist” in the MUD-Dev archives should pull up a few references–and I believe I discussed some of this with Ted last week as well in my neverending monologue on economic design in muds->UO->SWG. Perhaps he can summarize–clearly, my prolixity was such that I am incapable of reducing the topic to something short, since I spoke at him for a good three-four hours about it.

I think it’s worth pointing out that by far the main current of virtual design is NOT simulationist. The number of simulationist games is fairly small both in the commercial and the hobbyist world, and there’s only one full-blown simulationist world in the 100k club–UO. (SWG is half-heartedly simulationist, and only in some areas–the economy, most notably). One of the chief inspirations for simulationist design in the hobbyist community was the game DartMUD.

Just today I got a MUD-Dev post that asserted the following (this is heavily excerpted):

How can the current computer RPGs be improved? Why do I not find them particularly interesting? The following are the points that I think designers are underestimating in their players.

– Players are creative

Most RPGs of today rely on the idea of experiencing pre-created content: players should experience the game content much like a very interactive book… The designers assume that players like to be entertained in their laziness, rather than entertained in their creativity… While often appealing by graphics and professional content, such games easily break if new content is not constantly added, which really isn’t cost effective. They work by “throwing” tons of content on the players…  Most players are creative enough to provide most game content if they are given an open ended interesting system. Non-creative players will live off of the creative ones. In order to make the game interesting enough it must have detailed game mechanics…

– Players are individual

Designers often assume a lot about the player’s taste and needs. Players do not want to be fitted in to predefined roles and ways of playing, they want to pursue their own specific lifestyle; they want to play *their* role…

– Players want detailed realism

Most players of RPGs want an extraordinary experience that’s different from the real world… A game would break by having too simplified and too non-realistic game mechanics…. Detailed game mechanics adds a depth to the game that can not be achieved in other ways… for instance expanding the number of item attributes, creature skills and game concepts, and  letting them form complex realistic relations.

/Björn Morén

Basically, a simulationist manifesto. But at the same time, as someone with strong simulationist leanings myself, I have to play Devil’s Advocate. All evidence suggests that rather than the premises Mr. Morén starts from, the truth is that

• players want to be entertained, not to be creative
• players want to adhere to archetypes, not create their own roles
• players are perfectly happy with stagecraft, and prefer clear rules rather than unpredictability

The reasons why the simulationist approach is so interesting to some designers and players can be boiled down to these reasons:

• it helps with content creation by providing blanket solutions to problems
• it provides an ever-changing landscape
• it avoids issues with consistency (it’s damnably easy to get inconsistent stagecraft)
• it can be approached and understood as a system, cognitively speaking, rather than as an isolated special case

But problems inhere–and yes, some of these are self-contradictory, yet they always come up.

• the content generated, because it is systemic, lacks unique touches. One noise-generated cloud is much like another; silver linings need to be hand-implemented. 🙂
• the ever-changing landscape can actually irritate people, many of whom prefer a predictable environment. If Fred slew the Dragon in the Hoary Mntns, everyone wants to accomplish the same thing. The idea that the Dragon is gone forever is deeply disturbing because it feels like a lost chance for fun.
• you get consistency, which also means predictability, and predictable content may as well not be there.
• approaching it as a system means that vast swaths of the net game content can be “consumed” at once, because once you recognize the underlying rules of the sim, you essentially have cognitively consumed the entire system. Individual instances of the sim appearing are no longer interesting.

You can see echoes here, of course, of the complaints against FedEx quests, against fractally generated landscapes, against all forms of generated content, etc.

Long term, I remain convinced that the future lies in simulationist approaches. However, stagecraft is not going away for the foreseeable future, for our sims will have practical limits, and in many cases, the sim is tougher to execute on then (and provides no noticeable benefits over) the stagecraft solution for the same problem. Despite the interesting design possibilities and gameplay possibilities that good simulationist design can open up, it currently cannot compete with stagecraft, and so we see games like WoW ignoring basically all simulationist approaches in favor of simply very well-executed stagecraft. The issue of course, and what will likely drive the industry towards simulation over time, is cost. As the bar rises on the stagecraft, the cost to develop large MMOs with competitive content bases will become insupportable.

Simulationist approaches are not at all necessarily tied to “realism.” They are tied to “consistency” far more strongly.

An example of stagecraft: in NiftyWorld there’s a locked door that can have its lock either picked or it can be bashed down. It is, however, the ONLY door in the entire game that has those qualities, because it’s stagecraft. It’s a special case, created because some quest demanded it. It is not generalized to other doors, other locks, other objects made of wood, because it’s the equivalent of a special-case prop.

In a simulationist world, you’d try to abstract the qualities of doors and make sure that all doors could be lockpicked or bashed.

Now, this is a spectrum. Most worlds do do some abstraction. But the degree to which things are currently special-cased is very very high. The most obvious case is everything remotely narrative–it’s all special-cased. Most everything you see that gives text is special-cased. In many combat muds, the combat AI for any higher level creature is special-cased, custom-crafted for thatcreature.

The flipside of this is that emergent behavior doesn’t appear from special-cased content. It’s specified to a T. You might be able to reuse aspects of it, but it’s not designed to be generally used throughout the game.

As far as the push towards realism–in pretty much all the visual arts, the turn away from realism doesn’t tend to come until the forms of realism are pretty thoroughly mastered. We’ve got a ways to go.

Found this which may help clarify.

http://gamestudies.cdis.org/~rocketship/ionstormterms.htm

Simulations Vs Emulation

A simulation is actually trying to represent the way a system works with consistent rules that are universally applicable. An emulated event is not as likely to be predictable to the experienced player (unless he has played through the area), since it is not repeatable under laws or operating parameters within the context of the game. Once the player learns the rules, events in simulated systems can be predicted.

Example 01: In the (heavy-emulation) game NOLF, there is a combat encounter that allows the player to fight and kill sharks with a speargun. Later in the game, the player is trapped on a (retracting) bridge above a shark tank. If the player enters the water and attempts to kill the shark with his speargun, the shark is unaffected. (To get past the shark, the player must have earlier knocked a guard into the water–the shark will then attack the guard, allowing the player to move past.)

Example 02: In the game System Shock 2, one player reported (on the web) a combat encounter that is the result of the game’s simulated environment. The player-character was out of ammo and fleeing a hostile mutant. The PC ran into a room and was attacked by a turret. The PC hid in a corner, protected from the turret fire, trying to decide what to do. The mutant moved closer, searching for the PC. The PC then used his telekinetic psi ability to pull an explosive barrel toward him from the other side of the room. As the barrel passed through the stream of turret fire, it exploded, destroying the turret and killing the mutant.

Yeah, I am still playing Devil’s Advocate to a degree. I’m a firm believer in simulationist approaches over the long haul. I’ve just been tilting at the windmill long enough to want to make sure that people understand the (sizable) obstacles there are to overcome.

Example #1 is the default state of the art. I think it’s important that you realize that.

Trading is, I think, a poor example of the triumph of simulationism. In fact, most VR economies are emulated. Prices are fixed by developers and do not fluctuate, supply and demand is explicitly cased out of the system, and the AI doesn’t attempt to deal with the reality of the economy. This is again, the default state of the art. Generally, what happens is that players use these static system as mere money faucets, and there’s no real semblance of an economy. The result is typically mudflation.

Alternate currencies is a common side effect of this. The mud community noted the use of alternate currencies as far as 1985 in Habitat, and it was reinforced quite a lot during the early 90s.

The approaches these days tend to go for either this sort of centrally managed faucet-drain economy (which DOES reach an equilibrium point, btw–it just may not be one you expect) or to shift towards a more fully-player-driven approach. But it’s worth noting that the player-driven approaches don’t tend to have heavy simulation behind them, and instead rely on simply facilitating low-friction exchanges between users. There’s generally little to no attempt to simulate a larger economy operating behind the scenes. As a result, the economies tend to be (by real-world standards) relatively small. Examples of this approach would include SWG, 2L [Second Life], and There.

What exactly do you mean by real-world standards? Obviously this back of the envelope, but if you annualize SL GOM sales and expand our population to that of the US, you get a GDP of ~$20 trillion, which is 2 times the US GDP for 2002. Given that economies grow faster than linear, one could argue that this is a conservative estimate. The same analysis of UO eBay sales gets $4 trillion, so in the ballpark of the US.

OK, sure, expanding a virtual world to the size of the US is somewhat forward looking, to say the least, but it is clear that for the size of their populations, virtual worlds can generate powerful economies.

In terms of whether SL is simulating a “real” economy, allowing sites like GOM to operate and to provide efficient monetary exchange and to establish value seems to be even better than a simulation, right? – Cory Ondrejka

What I meant was that the economies tend to be less diverse, include far fewer participants and goods, and far fewer variables. Real world economies are simply more complex and have all sorts of backwaters and tentacles to them. What we’ve made are good toy models, with enough of the variables and behaviors that we see behaviors we recognize as classically economic occurring entirely within our framework, rather than escaping our framework.

To paraphrase something I wrote elsewhere:

Players are a plant. When a plant escapes a trellis, it’s a credit to the plant, not to the trellis designer.

Mind you, I don’t think that the sorts of economies that we see in SWG, 2L, and There are simulated OR emulated. I think there’s enough simulation of economic structure to allow REAL economies to manifest within the structure, in the case . I oppose this to cases where there is an emulated structure that cannot really support an economy because of its lack of detail, and which therefore results in a parallel economy being created.

Imagine the economy as a fluid. We’re building pipes and tubing. An emulated system builds some wood pipes and fake faucets, and it looks just like a framework for a real economy. But it leaks. It spills water all over the floor–which is to say, you get alternate currencies and so on, players finding alternate paths for the economic waters to flow through. A simulated system actually builds pipes and faucets. We’re making very SIMPLE plumbing at the moment, but at least it’s plumbing, and not stagedressing.

I notice that Gamezone are reporting a new land rush in UO. Houses that haven’t been used for 90 days start to decay. 5 days later, they disappear and their contents are available for anyone who wants them. The land they were standing on also goes to the first person (or, presumably, first commercial organisation that thinks they cn make a quick buck from resales) to stake a claim.

http://www.gamezone.com/news/01_27_04_05_29PM.htm

Would this count as simulation or emulation? – Richard Bartle

Again, not sure it is either. The distinction I am making is primarily one about how content is created, not one about player behavior. Emergent player behavior given a system is, well, emergent player behavior. Interesting stuff happens either way–whether they are dealing with an ecological simulation spawninig system, or with zonewide resets from the days of yore.

Raph — I don’t have time to post a new thread on this simulation/emulation thing today, but aren’t we just talking about degrees here? The ultimate emulation must be where the player presses the right button & sees the victory animation. But the ultimate simulation must be reality. Everything short of reality must take shortcuts and thereby limit potential player actions, which seems to be the characteristic of emulation.

There does seem to be another dimension on the sim/em difference that you highlight, and that is the universal behavioral equivalence of equivalent objects. So the ultimate sim objects would be Legos and the ultimate em objects would be Bertie Botts Beans.

I’m still grappling with how that plays out at the level of fine detail. By suggesting we’re just talking about degrees, I don’t mean that it isn’t a valid distinction. E.g., “hot/cold” is a degree distinction.

But I guess I’d be more interested in the reasons why you’re in the sim camp — I get the sense (correct me if I’m wrong), that you think the sim path holds more revolutionary potential for VWs as a medium? – Greg Lastowka

When are we NOT talking about degrees? 🙂 I was always accused of binary thinking when I was in graduate school, but what I am really attempting is to define useful axes on which to consider common practices.

I think that by saying,

The ultimate emulation must be where the player presses the right button & sees the victory animation

You’re missing the point a little bit; that’s why I prefer “stagecraft” to “simulation.” It’s not about the player, it’s about the authorial choices in how to present the setting. There’s a difference between an early Jackie Chan flick and most Hollywood action movies; the fights are far more simulated in the Chan movies than the Hollywood ones, where camera tricks are used to hide the fact that mostly, the fight didn’t really happen. Some directors wait to shoot at night; others use a blue filter.

Of course, the task we’re taking on in a simulationist endeavor isn’t so much waiting to shoot at night as it is attempting to model planetary rotation. 😉 Often the answer is “don’t be dumb, it’s far easier to just put in day/night cycles than to model planetary rotation. It’ll look the same.” That’s good stagecraft. And it’s all wonderful until you need an eclipse. Then you add in more stagecraft, etc. The more demands you put on your night sky, the more difficult the stagecraft becomes, until you say “maybe I should have just built a virtual orrery,” which is a simulationist approach. It’s not attempting to actually replicate reality–it’s still working at a level of abstraction. But it’s trying to make a good toy model of reality that follows the general principles.

Yes, I am in the sim camp, and have been struggling uphill with it for years. This despite the fact that given my background as a writer, as an artist, and so on, my formal training is in stagecraft. Why? It boils down to this:

Players always exhibit emergent behavior. Stagecraft never does. Systems sometimes do.

And that matters to me because I feel the medium hitting a wall, a wall on its ability to stage-manage entire worlds to the level that the audience now demands. It’s not an insurmountable wall, it’s one of cost and return on investment. Back in 1996, Jessica Mulligan used to be bemoan the fact that production values were about to shoot sky-high with UO, because it meant that the innovative but lower budget shops were going to find themselves unable to compete. What I am echoing is the same dilemma, except that I am saying that the big budget shops are against a wall too.

Note that I am not saying that the wall is insurmountable. I am saying that we may hit business realities that alter the way the market gives access to VW creators.