The Simulation Dream
June 10, 2013 1:57 PM   Subscribe

There’s an old dream in game design. It drives the design of games like SimCity, Dwarf Fortress, Tropico, The Sims, and Prison Architect. I like to call it the Simulation Dream. - Bioshock Infinite designer Tynan Sylvester on games, complexity, stories and simulation.
posted by Artw (29 comments total) 29 users marked this as a favorite
 
I just happen to be currently reading through The Chronicles of Gemclod, a... thorough example of this concept.
posted by history_denier at 2:23 PM on June 10, 2013


I used to joke that the one thing Maxis couldn't simulate was enjoyment of their products. I actually do love the idea of simulation games creating emergent behavior, and grew up playing and enjoying games like Elite Plus, Hardwar, etc... He's right on about how much of that effort fails to cross over into the player model. It's similar to why the best writers and editors cut a lot of the back story ruthlessly out of their works. The research may need to be done to inform the character's choices or build a realistic world, but reading an encyclopedia or atlas for a fictional world is usually even less exciting than for the real world.
posted by BrotherCaine at 2:55 PM on June 10, 2013 [1 favorite]


This is perhaps why games like The Sims never interested me. It's understandable that in an MMO, the player may not have the tools to track the rabbit population, which affects the wolf population, which affects the tendency of dragons to attack, etc. But that's because the game's interface is not built to visualize that complexity. By the same token, if you had to play Sim City in first-person perspective, it would turn out to be too complex too.

So I agree with Sylvester up to a point: complexity that doesn't affect gameplay in a big way is superfluous, uninteresting, and distracting to the player. Ultima Online is not a game about rabbit populations. But Sim City is a game about dynamical systems. I would feel… not cheated, but not exactly fulfilled to find out that all the buildings have the same occupancy regardless of size, or that all roads have the same capacity regardless of how they look.

Ultimately, your game has an underlying data model. Players interact with this model to learn about it. I don't put a hospital down in Sim City because I imagine what it would look like as I drive up to it. I place it because I know that it will affect the surrounding area within some radius. There could exist a different, completely sandbox game where you arrange some completely non-functional buildings and then drive around for fun. But it wouldn't be Sim City. Sim City is about visualizing complexity.

Complexity can be wonderful if it is a part of the game. Even a giant game is ultimately a repetitive activity in which players engage with "moving parts" to affect and bring about somewhat predictable outcomes. Superfluous complexity is when you're playing spades and someone sits down next to you and starts playing hearts. You could try to keep track of the game of hearts happening next to you, but you're already playing spades, so why bother?

It bothers me when, for example, a Japanese-style RPG has a bolted-on crafting system, or cooking system, or alchemy system, or whatever-the-hell. I am playing a game about random encounters and turn-based combat. The crafting-cooking-alchemy game is a separate game. If its only connection to the main game is that it's accessible through the same interface, then it's just a tedious distraction. If all I can do is produce an item that I am unlikely to use in the main game, what's the point? If they are closely connected parts of the same game, what's the problem?
posted by Nomyte at 3:02 PM on June 10, 2013 [1 favorite]


Such simulations may be intellectually interesting, but will not generate effective stories because they are emotionally hollow.

Intellectual interest is an emotion.
posted by LogicalDash at 3:52 PM on June 10, 2013 [4 favorites]


I refuse to believe Boatmurdered was nothing more than apophenia. Those sim-dwarves knew what they were doing.
posted by qxntpqbbbqxl at 4:00 PM on June 10, 2013 [3 favorites]


Part of the issue is with on-rails games themselves. For every degree of freedom that you give the game world, you may have to give the player equal or more freedom. How awesome would that rabbit-to-dragons mechanic be if there was a level 1 pull-a-rabbit-out-of-a-hat spell that everybody uses once and then forgets about... Until one day some weak mage figures it out and wins the king's bounty for ending the great dragon rampage.

Is anyone out there applying expert systems to game spaces?

I'm playing Fallout: New Vegas right now, and the narrative is deliberately structured that you have to side with one faction and accept the moral consequence of taking down another good karma faction. A slave obeys. The game is balanced in such a way that a difficult choice is required, but I kind of want to say, screw all y'all, I'm incredibly over-powered and you know it and I don't believe the ends justify the means so advance the plot come heck or highwater.

But even despite plot rails, the game still suffers glitches like finding named NPCs who have wandered a quarter of the way across the map from their hometowns.

NPCs powered by expert, goal based systems would be unpredictable as all-get-out, but at least you would know that they had an objective in mind when they wandered into the desert to be attacked by radscorpians.
posted by Skwirl at 4:16 PM on June 10, 2013 [2 favorites]


Will Wright has spoken compellingly about the gap between a good simulation and a good game. Where simulation seems to work best is when it adds some complexity and unpredictability to the world, while still staying within the confines of scripted set-pieces. Ie: fancy Havok physics for explosions, even if the game is an on-the-rail shooter. Or Radiant AI stuff in Skyrim, etc where part of the NPC activities are simulated, but the overall narrative structure is fixed.
posted by Nelson at 4:17 PM on June 10, 2013


I suspect it works better when you don't have a bunch of very expensive media assets that, if you're paying for them, you really want to direct the user to pass through all of them.
posted by Artw at 4:28 PM on June 10, 2013 [1 favorite]


I strongly disagree with both the article's premises and its conclusions.

Point 1: Games don't have to generate stories. Scrabble, Bejeweled, Settlers of Catan, poker... there are lots of games that don't depend on narrative for their appeal or success. Even within the realm of computer-based simulations, I don't think games like SimCity, Flight Simulator, Kerbal Space Program or Minecraft are primarily about narrative. (Unless "narrative" means "the succession of things that happen to the player, regardless of whether they form a coherent story or not" -- in which case, I think that's too expansive a definition.)

Point 2: The article seems to be making the tired argument that worldbuilding in a game needs to be used on the players or it's somehow wasted, worthless, pointless and wrong: "In general, lean on the simple side. You don’t have to simulate that much." Well, worldbuilding ain't bad, and complexity isn't harmful to player enjoyment. In fact, player enjoyment often goes up when complexity generates experiences (sometimes stories, but not always) that wouldn't have happened otherwise.

I come primarily from a tabletop RPG background, where most kinds of complexity are a lot easier to generate (because human brains are still a lot better at creating and relating complexity than computers are), but even within the realm of computer games, seemingly 'pointless' nuance can lead to interactions that wouldn't have been possible otherwise. Sandbox games, where players can create their own worlds, narratives, etc., rely on the ability to generate complexity out of seemingly simple subsystems. If they relied on the game creator saying, for example, "I can never imagine players wanting free camera movement, so I'm skipping that", we wouldn't get all the cool webcomics that games seem to generate. And I'm pretty sure that Tarn and Zach Adams would include every bit of complexity they could in Dwarf Fortress, if computers were up to the task; and I'm pretty sure that players would enjoy that complexity, too. Well, some would; but my point isn't that everyone enjoys complexity, just that complexity isn't something to always pare down to the smallest amount possible.

Now, there certainly are limits on what complexity a creator should try to put before the players: don't go beyond the specs of the computers you're designing for, for example, or the tabletop session's personal bandwidth for communication, or what you have time to create. It's not a good idea to spend countless expensive developer hours creating an ecosystem if the players are just going to kill it all immediately; but even that might not be a failure of planning, just a failure in presentation. Except for limits of practicality, there's no reason in principle why complex worldbuilding shouldn't be fun for both creator and player.

Point B: If your argument uses as a support the idea that Middle Earth is boring (which the article seems to imply: "And even if you did, it would be boring, because even Middle Earth isn’t very story rich"), that may be a good support to rethink. Because while Tolkien's storytelling may not be to everyone's tastes (it certainly isn't my preference), his worldbuilding is widely recognized as superb. Generations of people have been fascinated by it, many enough to write PhD dissertations about it, and there are many highly successful Middle Earth books that are reference materials -- highly successful, while not having coherent, sustained narratives at all. "Story-rich" is far from the only or best metric to use in judging the success of a game, or its setting.
posted by jiawen at 4:47 PM on June 10, 2013 [7 favorites]


I'm on your side jiawen But I would add that games like Settlers of Catan generate narratives as you play them, the story of the game, the upsets and triumphs. You say that's meaninglessly broad, but it's something that, the more the game's designer and writer scripts out the events of the game, tends to be lost. So rather than attacking your point I think I'm supporting it.

It's true that the narratives produced by playing board games, or very random procedurally-generated computer games, doesn't fit the traditional three-act ideal going back to Aristotle. I think this is actually one of the reasons I and others like them; they don't try to tell you the story, they aren't shoehorning the events into a trite formula, so they have a power to surprise that explicitly-written stories lack.
posted by JHarris at 5:10 PM on June 10, 2013 [1 favorite]


JHarris, I think we agree. It seems to me that the notion of "narrative" needs to be split into several sub-types. There are three-act stories, as you mention; anecdotes; sense experiences divorced from time flow; overarching ways of framing a situation without actually detailing the events or chronology; and others. I think part of the problem with the article is that Mr. Sylvester is conflating several types of narrative as "story", and applying standards for one to all.
posted by jiawen at 5:17 PM on June 10, 2013


jiawen, I don't think Sylvester is saying Tolkien's worldbuilding is boring; I think he's saying living in Middle Earth would be boring. Tolkien has already done all that filtering down to story-rich moments; that's why we have the Silmarillion instead of book after book of The Lives of Gondorian Peasants (side note: what do you think the ratio of human dirt-farmers to elven kings is in Middle Earth? What with immortality and all).

And I'd agree with BrotherCaine that worldbuilding is a fantastic tool for creating rich stories, but that you don't need to put it all in the finished product (although I love having access after the fact; I'd read the heck out of a fleshed out history of Westeros). And, really, worldbuilding isn't equivalent to simulation; it's already heavily biased towards what Sylvester calls human values, which is probably a better, more specific term than narrative for what he's talking about - i.e., conflict, tension, resolution, all of which certainly exist in Scrabble and poker (although I think they also have more strictly mechanical pleasures as well, which he's not addressing here - and which can provide a great deal of pleasure on their own).
posted by Runcible! at 5:51 PM on June 10, 2013 [1 favorite]


I guess I don't understand the specific distinction you're drawing between worldbuilding and living in a world, Runcible! All Sylvester says is "[y]ou can’t just simulate a super-complex world because players won’t understand it", but "simulate" here could mean "tell stories set within", "present in a non-narrative format", "create tension within the framework of", or something else. Can you explain?

I certainly don't think all the worldbuilding someone does should be dumped on the players/readers/target audience. Lots of worldbuilders seem to get a guilt complex from the notion that if the players/readers/audience don't get exposed to every iota of worldbuilding, then that worldbuilding has been "wasteful" and "wrong". I'm convinced that this is why some GMs in tabletop RPGs feel the need to exposit at their players for far too long. They're not necessarily frustrated novelists; rather, they're guilt-ridden worldbuilders.

However, the lesson should not be "...therefore don't do any worldbuilding that you can't dump on your target audience". The lesson should instead be "don't do worldbuilding that you don't actually have time for, and don't feel the need to exposit pointlessly, but otherwise go at it, enjoy it by yourself if that's appropriate or share with others if that's appropriate, and have fun!" And worldbuilding can indeed be very fun. I feel like Sylvester's article is drawing the former, erroneous, conclusion.

For me, "reading an encyclopedia or atlas for a fictional world is usually even less exciting than for the real world" is totally untrue. I love reading things like the Dune Encyclopedia, A Guide to Middle Earth, Expedition, plot-less RPG setting materials and other such fictional non-fiction. Worldbuilding sans plot isn't for everyone, it's true; but for many people it's highly enjoyable, and I get annoyed by arguments like Sylvester's that seem to say no one should or could enjoy pure worldbuilding.
posted by jiawen at 6:36 PM on June 10, 2013 [2 favorites]


The same happened on BioShock. While BioShock retained some valuable vestiges of its simulation-heavy beginnings,

Can someone explain the backstory I'm missing here -- what simulation-heavy beginnings? And what vestiges made it into the game? I sure didn't notice anything of the sort (which I suppose supports the author's point, if there were any there to be noticed...)
posted by ook at 6:38 PM on June 10, 2013


Oh, I guess he means Infinite. Never mind, carry on, I'm just being dense today
posted by ook at 6:44 PM on June 10, 2013


It sounds like Bioshock to me:

Early in the development of BioShock, that game had an ecology too. There were three parts to it. Splicers would hunt Gatherers, who were in turn guarded by Protectors. The player was supposed to interact with and manipulate this ecology to survive.

There's certainly signs of a less linear game in there, with the environment looking like it was more of a hub and spokes arrangement at some point.
posted by Artw at 7:10 PM on June 10, 2013


What I mean by 'living in a world' vs 'worldbuilding' is that if you narrated everything every person did in Middle Earth, it would be a lot of really, really boring stuff, with a few interesting bits mixed in. There would be page upon page about cow-milking, or threshing, or watching sheep, and only the occasional world-shattering quest. Whereas in worldbuilding, you can skip over most of the cow-milking. Perhaps you have a section on how cows are milked, or on the development of lactose tolerance, or on cowpox, but it's a much lower percent than reality.

I really like the idea of 'human values', or perhaps drama, as the variable here; simulation has a low 'human value' ratio, worldbuilding has a higher ratio, and a well-edited novel has a higher ratio yet. Or maybe you could break 'worldbuilding' up into encyclopedias and dictionaries and history books; dictionaries (or constructed languages) are less drama-dense than encyclopedias, which are less drama-dense than histories, which are less drama-dense than historical novels, etc. All of which is entirely normal; a novel is essentially life, but only the interesting bits.

All of that's only in general, though; different people have different 'human values', so while I can happily read wikipedia articles about sail arrangements, someone else might be bored silly. I get bored with sports, generally, but a lot of people extract much more value from them. There are some things (conflict, romance) that are generally quite good bets for value production, and some bits (sheep) that aren't. So if you're going to simulate things, it's important to simulate mostly the former, and just enough of the latter to keep things feeling grounded.

I think where simulation-based games often go wrong is that at a certain level of complexity, you start getting emergent behavior which may or may not be at all interesting, and the computer has no idea. So when you start from really basic rules (wolves eat rabbits, and also deer, dragons eat deer and also people, etc.), there's a chance that it'll all end up looking like something entirely boring (everything's dead). Theoretically, it would be possible to tweak variables until something interesting always came out, but the more complex the system, the more time that takes, probably in a quadratic scale, until you need eons to balance the system. Whereas a simpler system with more human-like factors is much easier to get right.

And, of course, since tastes vary, sometimes you wake up and theres a whole thriving industry of pure simulation games for things you think are a bit boring, so there's that, too.
posted by Runcible! at 7:12 PM on June 10, 2013 [2 favorites]


What I mean by 'living in a world' vs 'worldbuilding' is that if you narrated everything every person did in Middle Earth, it would be a lot of really, really boring stuff, with a few interesting bits mixed in. There would be page upon page about cow-milking, or threshing, or watching sheep, and only the occasional world-shattering quest.

Yup, still sounds a lot like Dwarf Fortress.

Sure, sure, you need to iterate over a lot of life before you get to the interesting stuff, but that game seems to do all right with 200 dwarves and a year or two of game time.
posted by LogicalDash at 7:27 PM on June 10, 2013 [2 favorites]


Thanks, Artw. I managed to skim right past that paragraph; I am being dense today.
posted by ook at 7:37 PM on June 10, 2013


...what vestiges made it into [Bioshock]?

It was possible to sucker Splicers (or their flying robots) into accidentally grazing a Little Sister or a Big Daddy. The battle that followed (Big Daddy pitilessly gores Splicers) was arguably sort-of-emergent behavior, in the sense that the goals and capabilities of the individual actors had been specified but the specific blow-by-blow of their interaction had not. I expect the behavior of hacked robots and hypnotized Big Daddies was similar.

Big Daddies would knock on a mouse-hole to call forth a Little Sister, escort the Little Sister across a good chunk of the map, then deposit her in another mouse-hole. IIRC, the Little Sisters would lead the way, and they would alter their route so as to stumble on the freshest Splicer bodies. So the Big Daddies' routes were ultimately affected by player actions. That was interesting, but generally of very little game-play consequence.

Bioshock gave the player access to volatile, exploding objects, which could be used in several ways. They could be thrown away or deliberately exploded to eliminate the hazard, they could be thrown directly at enemies, or they could be strategically placed and detonated from afar. A big pile of them could kill the most powerful adversaries in the game instantly. All of that appeared to be emergent from the simulated physics.

None of that is especially impressive, though. Plenty of games have this relatively shallow degree of emergence in them. The original Deus Ex was another game that was reportedly conceived this way, but shipped with only a few glimpses of that original vision. The Thief games are all about pitting the lone player against a flock of AIs. GTAIII included some parts that were strictly goal-oriented; any way the player found to use the environment to meet the goal counted as success. Fallout 3 and Fallout New Vegas certainly have plenty of that sort of thing.
posted by Western Infidels at 8:13 PM on June 10, 2013 [2 favorites]


I'm pretty sure that for a while there was talk of how much further Deus Ex: Invisible War was going to take those aspects and expand on them, which if you've played the resulting horrible game is bitterly ironic.
posted by Artw at 8:47 PM on June 10, 2013


Runcible!, I'm unsure, but it sounds like you're making the argument that encyclopedias and dictionaries or sheep are inherently less interesting than (in your example) novels conflict or romance, which I don't accept. Perhaps generally more interesting to more people, though. So I don't accept that "it's important to simulate mostly [conflict and romance], and just enough of [sheep] to keep things feeling grounded." As you say, many people find farming simulators more interesting than more conflict-full games.

And as LogicalDash pointed out, things like Dwarf Fortress really do try to simulate all the nitty gritty little itty bitties. Flight Simulator tries to simulate every single switch and gauge, and players seem to want ever-higher levels of detail in the simulation, even though less detailed versions might pack more action and thrills into every moment (at least for some people). I bet Dwarf Fortress will continue to simulate ever more details as computer specs get better. Lots of successful games have that push behind them. A lot of people don't just want the conflict-full, romance-full detail; they want all the detail they can get.

Mr. Sylvester's article is pointing out something very important: that no one has infinite time or infinite resources, and thus no one can simulate reality perfectly within a game. That is an important lesson to learn. (I wish I'd known that when, as a teenager, I started trying to build an RPG world and followed ICE's advice that I should start by detailing all the plant species!) I just disagree with most of the rest of Sylvester's argument. ^_^
posted by jiawen at 9:36 PM on June 10, 2013


Is anyone out there applying expert systems to game spaces?

Left 4 Dead uses an AI Director to decide when to spawn a horde attack or special Zombies. I'm not sure about how much of The Last Express was AI vs scripting, but it had actors with their own goals moving in real time.
posted by 23 at 10:41 PM on June 10, 2013 [1 favorite]


I'm playing Fallout: New Vegas right now, and the narrative is deliberately structured that you have to side with one faction and accept the moral consequence of taking down another good karma faction. A slave obeys. The game is balanced in such a way that a difficult choice is required, but I kind of want to say, screw all y'all, I'm incredibly over-powered and you know it and I don't believe the ends justify the means so advance the plot come heck or highwater.

I found that there were only a couple of decisions in New Vegas where I genuinely felt conflicted (Mr House, The Powder Gangers). The rest of the time you can thread the moral needle if you're careful. Although it's easy to slip up and find yourself having to dispose of some former friends. Or inadvertently triggering their massacre. IIRC, keeping the Brotherhood and the NCR on your side is particularly tricky.

Keeping frequent saves is key. But then the game is so flaky that you're wise to do that anyway, just in case.
posted by inpHilltr8r at 12:50 AM on June 11, 2013


Oh, I guess he means Infinite.

No I think he means the original. IIRC, the Little Sisters, were supposed to be part of some complicated economy of Adam, but all that remains is that they'll drain a splicer corpse if one is nearby.

I couldn't see any trace of simulation in Infinite.
posted by inpHilltr8r at 12:54 AM on June 11, 2013


Teach me to post before finishing the thread...
posted by inpHilltr8r at 12:56 AM on June 11, 2013


I'm just gonna leave this here...

http://en.wikipedia.org/wiki/Microcosmic_God (the full short story isn't hard to find via google)

It's what comes to mind every time I play with "Fishbowl Games". (Now reading the original article)
posted by DigDoug at 5:59 AM on June 11, 2013


This is kind of my "thing," since (among other things) I play Sims 3 and write about it professionally.

Here is where the guy lost me, when he said: "The Sims 3 has a very simple computer model of social interactions which does not really depict deep human emotions like jealousy and anger."

That is completely untrue. So untrue that it makes it obvious he has never played the game in any meaningful fashion. Sims 3 DOES depict jealousy and anger in very clear and meaningful ways.

Sims in the situation he depicts will get a very strong negative moodlet (buff) due to jealousy, rejection, or an awkward social situation (depending on the context). Their relationship level with each other will drop. Probably into the red. The worse their relationship drops, the more likely that they will fight each other. Interactions like "Insult" and "Slap" are unlocked. And so forth.

Saying that "most of [the story the player created] is apophenia" is dismissive, arrogant, and utterly missing the point.

Creating these stories, nudging them in the direction you want, and (often) laughing at the results is the entire point. It's not a strange phenomena of deluded human behavior. It's collaborative storytelling, where your collaborator is the game, and the audience is the other players you talk to.

Sylvester is acting like the player's narrative emerges as an interesting yet accidental side effect of the game developers' attempts to create as accurate a simulation as possible. This is completely backwards.

I've seen a lot of people get the wrong end of the stick when they talk about Sims games. But this is by far the wrongest.
posted by ErikaB at 7:55 AM on June 11, 2013 [1 favorite]




« Older Acme Supply Co. Presents...   |   Anecdotally... yup Newer »


This thread has been archived and is closed to new comments