Thursday, June 16, 2011

Dark Past (part 3): Letting Go of the Immersive Sim, of flu viruses, ghosting, and why we're all Kate Winslets at heart.

This is a series of posts that analyzes the immersive sim. It's a play on the excellent RPS feature, Dark Futures.

In past posts, I argued that (part 1) immersive sims were so cool we got overprotective of them and suffocated them, but (part 2) we can still extend the same design theory to contemporary single player design.

For part 3, I'd like to explore the limits of "immersive sim theory" and even criticize it in light of recent research. This devil's advocate stuff will help us in part 4.

Both system dynamics (sort of the science of systems) and Looking Glass Studios came out of MIT in the 70's or 80's or some time around there -- and for the convenience of my argument, let's assume it wasn't a coincidence...

In her amazing primer to this system lens, "Thinking in Systems," Donella Meadows explained some (heretical? controversial?) consequences of system dynamics:
  • The flu virus does not attack you; you set up the conditions for it to flourish within you.
  • No one "killed" your immersive sims. Instead, it was caught in a negative feedback loop as both designers and players lost interest in it.
Okay, that last one was mine. Here's a couple more:
  • If a player doesn't complete a level, it's because the level did not help nor motivate the player to correct their input... NOT because they're bad at games.
  • If I am a player is trapped in Lab X-10 with low health, a backpack of broken guns and no ammo, it's because... you get the point.
The first one is (now) common sense, most famously championed by Valve, but that last one is a near-failure state that creates an aesthetic of tension and fear. You have to let the player fall so they can triumphantly rise again.


Such polarized, near-failure states are desirable -- so much so that some games have taken to faking them, i.e. suddenly plunging you into near-death mode as part of a scripted sequence.

At that point, it's actually in the game's interest to preserve your barely conscious life.

Your health flashes red and suddenly your enemies shoot like Imperial stormtroopers, and you just barely get out of the fire with a few health points left and your heart is pounding. Suddenly you have more pain tolerance, your accuracy is a bit better, you move a bit faster, just barely... just barely...

The game's rigged.

In this way, single player levels aren't just systems; they're profoundly unstable, unbalanced, rigged systems. It's what made that notorious first Cuba level of Call of Duty: Black Ops possible. Designers figure we're already rigging it for the player, so why not rig it some more?

Thief is rigged. Deus Ex is rigged. Perhaps the failure of Blops was that it failed to hide it.

The dominating trend in the typical single player FPS game, as theorized by Dan Pinchbeck.
In his PhD dissertation, “Story as a function of gameplay in First Person Shooters: an analysis of FPS diegetic content 1998-2007,” (my summary, written in human English, is here), Dan Pinchbeck argues that one of many defining traits of single player FPS games and levels is decreasing complexity over time.  

Eventually, the system gets "solved." You hack the whole security system and turn-off all the cameras. You unlock all the doors. You fetch the Companion Cube. You kill the striders. You snipe everyone in the warehouse. You clear the dungeon. And so on.

This "intentional instability" contrasts with the system dynamics used in a lot of multiplayer FPS level design, where the layout must be "balanced" and allow for any team or player to fight-back against the opposition and re-stabilize the system. (Though there are MP maps like cp_steel that can definitely destabilize balance.)

Imagine the single player equivalent: a Portal level that's constantly unsolving itself in order to maintain the complexity of the problem space, or an Elder Scrolls game where all monsters gain experience levels alongside you, or a gated arena full of endlessly respawning enemies. (What is this, arcade mode?)

"It's not fair," you would scream -- all while the game regenerates your health, gives you speed power-ups, constrains the NPC movement patterns and forces them to bark constantly so you know where they are. And do NPCs get to quick-load or quick-save? You're not being fair either.


So, if we often think of single player level design as systems to be solved, as puzzle design rigged in the player's favor... what lives outside of this thinking? Unsurprisingly, some immersive sims and their descendents:

Expert Thief players pride themselves on "ghosting," a style of play that requires players to not extinguish any torches, nor neutralize / alert any guards, etc. -- the implication being that authentic "professionals" don't need to solve the puzzle or reduce the complexity of the system. (And only amateurs use gas arrows!)

This is, however, a totally artificial, socially-constructed constraint that exists outside of the designed video game. There are no specific rules that enforce it. Less charitably, it might be described as self-imposed instafail mode.

It always strikes me as paradoxical: Ghosting is a set of rules designed to guarantee the complexity and integrity of the game system, but they're actually already violating the integrity simply by existing outside of it.

(Like a madman with an assault rifle, threatening to shoot unless we respect the ban on assault rifles.)

(Not to imply that those TTLGers who ghost are sociopaths.)


Well, here's a more mechanical example: several single player FPS games (Stalker, Far Cry 2) shift notions of level design toward procedurally populating worlds with encounters. However, sometimes the output feels like broken experiences, though occasionally it's profound, I'm told.

When playing almost any immersive sim or open world or sandbox game, there's a mental distinction between the structured / authored "missions" and the dynamics / collisions of systems. (e.g. your killing a civilian then commandeering the ambulance isn't "content"). One type of play is sold as legitimate DLC, the other is the "filler" in-between.

One is sometimes worth extra money... the other is cynically described as "a chunk of the game that was cut-off to sell separately to us," or even worse, we dismiss it as horse armor.

We value the aesthetics that systems produce (a mission) but not the systems themselves (the traveling in-between missions), aside from the occasional YouTube video of weird stuff -- thus, the somewhat recent success of stuff like fast travel or the ghost town of LA Noire.


Left 4 Dead's workaround was to brand the set of pacing algorithms as an "AI Director," a god out of the machine whom you can curse and blame for setting three Tanks on you. (THREE?!)

Such a conceptual shift can be powerful: in those instances, we don't necessarily say the game is broken as in Far Cry 2 -- instead, the Director is just being an asshole. (One wonders whether FC2 would've been better received if it were branded with a "director.")

But this is a mind hack at best, a blatantly non-ludodiegetic device that violates the credo of immersive sims... Unless we're fighting a Shodan, or an Andrew Ryan, or a GladOS. Narrative conflict is not an "emergent collision of systems" but is purposeful, strained and artificial: it requires black and white, good and evil, or a puzzle-solver and a puzzle-designer.

So it sounds like we don't actually want to transcend puzzle design.

We don't really want an infinitely replayable single player experience, because we already have it in many games but we usually put authored "content" on the pedestal instead. We even respond much better when we imagine there's a "director" or a god orchestrating everything, instead of a cold unfeeling "dynamic system."

What's more palatable to the average person: staring at a screen of ASCII characters for months before you can decode some sequence of events, or reading the impeccably authored / designed tales of Bronzemurder and Oil Furnace?

Here take this short quiz: which has been more viral and popular? Emergent collisions of systems and blah blah blah / the dinette set... or rich, textual, stylish authored content / a brand new car?


At the end of her book, Meadows warns about glorifying the system lens. You might understand how something works, but it'll just bring up more questions -- and even if you know how to design a system, implementation is a different story. Knowing the mechanics of addiction and quitting a smoking habit are two very different things.

Meadows also offers this insight: "Self-organizing, nonlinear, feedback systems [i.e. the world] are inherently unpredictable. They are not controllable. They are understandable only in the most general way." (Imagine that printed on the front of the Deus Ex 3 box: "now with self-organizing nonlinear gameplay that's barely comprehensible!")

So can systems guarantee a certain understanding or meaning? I'd argue no:
  • Does a gas arrow in Thief mean "useful power-up" or "amateur's crutch"? What is a "cheap" tactic?
  • What's up with quicksaving constantly? Is it condoned by the game design? How does that fit into pacing? Would you quicksave less in front of an audience?
  • In Counter-Strike, is it poor sportsmanship to use smoke grenades, aka "lag grenades", or would you argue that if it's in the game it's okay? 
  • How does the concept of sportsmanship fit into systems thinking and procedurality?
What if a game has goals, but no designed mechanics? (In a game where being unfair is "totally okay now," then what is too unfair or not okay?)

* * *

(TL;DR) System dynamics is the foundation of immersive sims, but we're constantly pissing on it.
  • We praise the "intelligent AI" in Halo, but the AI always lets us win. If it doesn't, it's cheating. We design our systems to lie to us, then get angry when we catch them lying to us.
  • We say we want emergence, but when given the choice we often privilege and commodify authored "content" over procedural generation. We want puzzles, stories and mirrors -- not new worlds or universes. (e.g. I have no desire to play something so alien like The Void, even though I should.)
  • We say game systems are so powerful and determine the range of interaction, but then Thief players (protectors of robust game systems everywhere!) flout them every day when they "perform" as a ghost.
Thus, I argue, we do not actually value the "systems" in immersive sims because we limit their complexity to a scope we can easily comprehend, master and manipulate. It's the same "partnership" or "conversation" as between master and slave.

... and that's okay. (Well, slavery isn't, but...)

These systems aren't even that powerful anyway! We impose new meanings on games all the time. People are still finding new ways to play Quake, Starcraft, basketball, etc.

... and that's okay.

But what's NOT okay is lying to ourselves.

Maybe we need to re-think why we loved immersive sims in the first place. It's not because we enjoy complex and emergent interactions between systems, considering how heavily we cripple and limit these systems and devalue the interactions that weren't directly authored by someone.

Meadows proposes an alternative to this whole trap: let go.


Let go and know when the system lens helps you and when it doesn't. Let go and know that using systems is a system in itself, or something.

Know that there are other ways of thinking about games, outside of procedurality.

The immersive sim is better off, its organs donated to other needy genres, as evident from all the games I've cited that aren't strictly "immersive sims qua immersive sims" but have inherited many of the traits.

Yes, the game industry is making the zombie-like Deus Ex 3 because that's what we asked for... But if you've read the Dark Futures interviews, the founding fathers and mothers of the immersive sim have already let go.

Why can't we quit smoking? Why can't we quit you? Why can't we let go?

Because we're all Kate Winslets at heart.


Next and last time:
Dark Past (part 4): never let go, a valence theory of stealth level design.