Tuesday, November 19, 2013

Emergent gameplay vs whatever the other kind is

I spent the weekend at a delightful little game-dev conference at NYU. Much cool stuff happened there. However, I want to focus on Saturday morning.

Saturday's first talk was by Warren Spector, who has recently switched from developing games (Deus Ex, etc) to teaching the subject at UTexas. His thesis was simple: emergent gameplay, emergent gameplay, emergent gameplay are you listening people.

Here's a writeup of Spector's talk, thanks to Leigh Alexander and Gamasutra.

(Footnote: the quality of emergent gameplay should be referred to as "emergency". As in, "Yeah, that game had a lot of emergency." Hat tip to Vernor Vinge for pointing this out.)

Spector tried not to say "Everything else sucks." He stated right off that he was oversimplifying, and that he's just presenting the kind of games that interest him. But it was hard to avoid the subtext that any scripted, linear, or single-solution interaction was inferior -- bad game design. Inherently. That if players tried the emergent (simulative, rules-based) gameplay they'd be happier and never go back.

This led to a lot of backchannel muttering among the audience; you can scroll back through the #practice2013 Twitter hashtag if you want the dirt. I disagreed, Emily Short disagreed. But one can only fit so much pith into tweets, of course.

Then the next two speakers (Soren Johnson and Keith Burgun) started their presentations with exactly the same analytic framework: linear versus generative, scripted versus emergent. And they seemed to assume that this question was settled -- that all of us were already committed to the emergent end of that dichotomy.

So. I am not part of that stampede. I disagree with the conclusion, and I disagree with the premise.

First: this emergent-vs-linear thing is oversimplified, just as Spector said. Any analytical model is simplified, but this is the bad kind; it trims away something crucial. One of the later speakers posted a slide with a linear-to-emergent scale: games could be placed at any point along the scale. But this is still too simple!

A game can do more than one thing. Most do. To describe a game, you need a whole stack of scales. For example, Bioshock (the first) gives you:
  • free movement in a complex spatial environment;
  • a fairly rich array of tactical combos for combat (built on power and weapon upgrades chosen over time);
  • exploration through a branching tree of rooms (with much backtracking as you achieve goals and are rewarded with new tools and options);
  • an irregular sequence of environmental puzzles, each of which has a single solution to be discovered;
  • an infrequent sequence of binary choices for dealing with Little Sisters;
  • a fixed sequence of story chapters leading to a narrowly forked ending scene.

So what in there makes Bioshock a "linear" game? Certainly the last aspect. Certainly not the first two. The middle ones are worth an argument in their own right. So, can you ignore the combat or the Little Sisters or the overall storyline, and still claim to describe Bioshock as a whole?

Heck, look at Myst. For twenty years, gamers have been dismissing Myst as a linear slideshow -- while other gamers remember it as a completely open, unconstrained, explorable environment. I refuse to declare that either view is wrong. Surely this demonstrates that there's more than one layer here? Every "emergent" game has scripted aspects to it, and every "linear" game has aspects of surprise, and they can both be happening at the same time in different ways.

Moving on.

One of Spector's repeated points was (I paraphrase) "If you create a clever puzzle with a solution, you're showing how clever you are. Let the players show how clever they are." And much other language about "putting players in control."

I find this painfully misleading. For a start, complicated systems are expressions of the designer's intent! If they weren't, we wouldn't have to spend so much effort tweaking, adjusting, and getting them right! (The word "right" is itself an admission.) To quote some of the replies:

ok look when "emergent" interactions occur they're part of the possibility space the designer set up, not magic out of nowhere (-- Michael Brough)

it is possible for the rule set itself to express a world view; emergent gameplay != absent designer (-- Emily Short)

Or, in my own words: bringing more player agency into the experience does not mean pushing authorial agency out. It makes authorial agency different, more complicated, yes. But the simplistic see-saw trade-off is a phantom terror.

Go back to Spector's immediate statement. Is it bad to work very hard, to be extremely creative, in designing a puzzle? Shouldn't we laud that effort, when the designer chooses to put it in? Surely the point of emergent gameplay isn't to let the designer be lazy. (If so, it's not working, nohow.)

No, what Spector wants -- rightly -- is to permit the player to be creative. We both treasure games that require the player to think creatively. We don't seem to agree on what that means, though.

Most of my text games hearken back to old-school IF: puzzle situations, unique solutions, hand-crafted outcomes. Two people who finish Spider and Web will, ultimately, have found the same solution for every puzzle. And I thought of that solution before either of them. But does this mean that they have not been playing creatively?

I say they have been. The work of solving these puzzles -- the play experience -- is of experimentation, discovery, and then synthesis of the results in a way which was not immediately obvious. That's creative thought. Dismissing this as "square key in square hole" is ignoring the point.

IF traditionally builds a complex, rule-based world out of hand-crafted, unique responses. One action is shallow. A hundred actions, revealing common underlying rules, is a fluid environment. (That's why "square key in square hole" is an oversimplification. A puzzle with one clue, one option, and one action is a trivial toy.) Seeing unexpected possibilities in a fluid environment is... exactly what Spector says he wants.

random thought: at one point if a dialog tree is big enough, it will FEEL emergent. Is that what matters? (-- Reynaldo Vargas)

Yes. (I could quibble about "tree" being a prejudicial term here. Make it big and stateful, and it'll stop being a tree. Down that path lies Versu.)

Switch to a more familiar example: the crossword puzzle. It has a single solution (unless the designer has been really creative). But you have to be clever to find it. Crossword solving is not a monotonous dictionary attack. Puzzle fiends then move on to baroque variations (the cryptic crossword, the variety cryptic... the MIT Mystery Hunt) which require even more creative thought to solve. Yes, the designer has to be cleverer yet. Crosswords are harder to construct than to solve. I don't see that as a reason to criticize; I'm grateful to the designers.

Let me wrap up by stepping back. I am not an enemy of emergent gameplay. It's awesome. I try to build my games, even the tiny ones, around an explorable mechanic with complex, generative results. I love Spector's pithy metric: how much do you know in advance about a player's play experience? That's an important question; it permeates every design decision you will make. (To say nothing of the crass realities of replayability and Internet walkthroughs.)

But I do not accept that this is a quality metric. It doesn't tell you whether a game is superior, or even (if you dare ask) "more fun". And it's not a simple metric. You don't ask that question once per game. You ask it over and over, interrogating each aspect and element of your design. Never "is this an emergent game" (blech); rather "what are the emergent interactions in this game?"

Go now and do likewise.

Footnote 1: There's a whole subdiscussion to be had about minimalist games -- games which have been boiled down to a single core mechanic. (This was the focus of Keith Burgun's talk.) To the extent this is possible (Go already exists!) it avoids much of my argument. Some games really are doing only one thing! And if so, you'd better get as much oomph as possible from that one mechanic.

But let's not confuse the simple case with the general case. Commercial games tend to be large, rather than minimal. And both in and out of industry, genres love to hybridize, forming interesting compounds and complexes.

Footnote 2: I forget what footnote 2 was.

Footnote 3: Don't imagine that the entire conference was a buzzsaw of absolutist "emergent" rhetoric. Later on Saturday, Emily Short sat next to a couple of the Walking Dead designers and they talked about the balance between generativity and prescription. No, it wasn't a counter-spinning buzzsaw of authorial control. Nobody wants that. Tradeoffs are always where the interesting design is.

Comments imported from Gameshelf

Andrew Plotkin (Nov 19, 2013 at 9:51 PM):

Figured out what footnote 2 was supposed to be.

At some point we'll need a grand theory of interactivity as player perception and understanding. We've gotten good at talking about interactivity as player action -- too good. It needs to be a two-way street, afferent and efferent.

Figuring something out is a form of agency, even though it is entirely in the player's head.

Amer Khan (Nov 20, 2013 at 9:26 AM):

Sorry, off topic but how to subscribe through email?

WhineAboutGames (Nov 20, 2013 at 12:02 PM):

I was thinking of something touching on these lines while reading this. There was the suggestion that every linear game has emergent aspects, and that made me immediately state "What about pure visual novels?" whose interactive space is so heavily constrained that there is no perceptible difference between one person's playthrough and another... Except that the main interesting difference between players is in their reactions to the information they uncover. What they see, what they understand, what they anticipate, what meaning they derive from it all.

It's only interesting to watch someone else play a VN if you get to see and hear their reactions along the way

Reynaldo Vargas (Nov 20, 2013 at 8:41 PM):

Your response to my quote is precisely what I meant. I'm using the lingua franca, so to speak, of the people who were to have read it. I meant to amend the tweet with the concept of rhizomatic structures, but I had too many twitter arguments at the time.

I'm very glad you wrote this, I had the very same sentiments and I was rather excited to jump into this conversation.


Jason Dyer (Nov 20, 2013 at 9:49 PM):

I'm particularly thinking of mystery games which involve menu choices at the end where you finger the culprit, give motives, etc. Certainly the interactivity there is way past the simple menu.

Andrew Plotkin (Nov 21, 2013 at 12:54 AM):

There is no email service for this blog's updates, per se. You may be able to set something up using the RSS feed: http://feeds.feedburner.com/GameshelfBlog

Andrew Plotkin (Nov 21, 2013 at 12:57 AM):

You're welcome. I had not heard the term "rhizomatic structure"... well, not for ten years or more, anyhow.

Andrew Plotkin (Nov 21, 2013 at 1:08 AM):

Also, the notion of "reflective choice": choosing what you think of your character, or of another character, or of some story event. This may not be represented in the game state at all, but it's still a choice. (Emily and others have been writing about this sort of choice for years.)

Jerome West (Nov 21, 2013 at 9:53 AM):

My first ever experience with emergent gameplay (at least, as far as I recall) was in the old-school adventure game The Hobbit.

As a child, it used to amuse me no end to round up Thorin and whoever else I could drag along, bring along a bunch of food and wine, and throw an impromptu party. Drunkenness and gluttony ensued, often descending into violence. Even at that tender age I had some awareness that this was something pretty special happening; that I was creating something in this game world which the designers themselves had possibly never even seen.

Whilst I agree entirely with your point that emergence is not the only game in town, it's that kind of thing that I really miss in modern IF.

Vulpis (Dec 12, 2013 at 9:56 PM):

Hmmm. I'm reading this, and thinking among other things, of emergent gameplay as being a bunch of lines hooked together.

That said..I think there's a bit of an overemphasis on 'emergent' experience being overexaggerated here--if only because of the problem that your system can be emergent and flexible as you like, and still end up with players who are a line of flaming lemmings. Witness many MMOs who provide all kinds of options for character builds, only for the playerbase to settle into a set of One True Builds for anything.

Frankly, I think the ideal that emergent play is shooting for isn't going to be even close until we are able to do what could be best called 'reactant' play, where the computer side of things is sufficiently flexible to react to anything the player might try, whether the original designer thought of the scenario or not and provide a response better than 'You can't do that'. This would be similar to how a human-moderated RPG runs, or perhaps the game from Ender's Game.

Andrew Plotkin (Dec 12, 2013 at 10:24 PM):

What you call "reactant" is sort of the eternal imaginary solution -- like artificial intelligence. (Which, arguably, it is.) It's not a reason to stop trying today. I think game designers have more practical ideals than you allege.

(The notion of "anything the player might try" is a huge lever in its own right, but that would be another whole post!)

The question of providing choices without defining The Best Choice is interesting. And if a system has lots of unobvious consequences, it becomes impossible for the designer to balance them. But this just means we have to step back and think about what "balance" means, and what kinds of games don't require it in the first place.

Greg Collins (Jan 10, 2014 at 4:33 PM):

A very erudite and informative discussion, but I still think that "game" players boil down to two types (how linear of me). Those who enjoy a challenge and those who don't. Think of a Rubik's Cube and a dollhouse. The first is gaming challenge on steroids, the second is wide open emergent play -- one simply follows one's own imagination, no rules, no challenges, no end even. Both types of play can be fun. People should simply know what type they prefer and games should be suitably labeled. I certainly don't buy the notion, if I'm understanding it correctly, that linear gameplay is less interesting than wide-open play. Sokoban is a great puzzle "game" with really only one game mechanic and yet I never tire of its gameplay. Of course, games can mix in both types of play. The most obvious example is the classic adventure game. A great adventure is an intelligent, creative mix of puzzle and exploration. I love unabashed puzzles, but a puzzle creatively inserted into an adventure game provides the double pleasure of story element with creative challenge. If, that is, you enjoy that sort of thing.

Andrew Plotkin (Jan 10, 2014 at 4:43 PM):

I can't agree that "wide open" play equates to "players don't enjoy a challenge". The people most ardent about that sort of thing talk about inventing challenges for themselves.

In any case, it's not a boolean property. (That's not linear of you, it's zero-dimensional!) Labelling A or B is never going to convey what the game is like.

No comments:

Post a Comment