Thursday, July 24, 2008

Step-Ladders and Demon-Traps

Edgar Allen Poe once wrote a piece about his “Philosophy of Composition,” in which he describes the principles that shaped his poem The Raven. He explains that his poetry is not (as many think) the result of the attempt to render some private visionary experience but a matter of carefully constructing a piece that will produce the desired effect on the audience: “Most writers- poets in especial- prefer having it understood that they compose by a species of fine frenzy- an ecstatic intuition- and would positively shudder at letting the public take a peep behind the scenes... at the wheels and pinions- the tackle for scene-shifting- the step-ladders, and demon-traps- the cock's feathers, the red paint and the black patches, which, in ninety-nine cases out of a hundred, constitute the properties of the literary histrio.” The business of crafting a work of art is all about the prosaic business of jiggering the poem's mechanics-- inserting certain turns of phrase, discarding others-- in order to create the desired impression. The effect of Poe's description is kind of disenchanting: “I next bethought me of the nature of my refrain. Since its application was to be repeatedly varied it was clear that the refrain itself must be brief, for there would have been an insurmountable difficulty in frequent variations of application in any sentence of length... Having made up my mind to a refrain... such a close, to have force, must be sonorous and susceptible of protracted emphasis, admitted no doubt, and these considerations inevitably led me to the long o as the most sonorous vowel in connection with r as the most producible consonant... [in] such a search it would have been absolutely impossible to overlook the word "Nevermore." In fact it was the very first which presented itself.” I couldn't help but feel disillusioned to learn that the meditation on death and loss-- even the character of the raven itself-- are all there ultimately just to provide an excuse to repeat the sonorous “o.”

I had the opposite reaction to the developer commentary included in The Orange Box, which gives a revealing look at the wheels and pinions behind Valve's classic games. If you enjoy Half-Life 2 or Portal, it is worth it to return to these games with the commentary turned on-- as you play through the game you can activate audio clips in which the game's creators describe the creative process of making the games.

One of the real strengths of these games, as Shawn Elliott has remarked, is the subtlety with which they communicate the solutions to their challenges. The commentary abounds in descriptions of how small environmental cues-- the color of a panel, the placement of a cube-- had been designed to draw the player's attention to the significant elements of a puzzle. In the Portal commentary, designers reveal how they structured the introductory levels in order to gradually teach the player the different aspects of the game's physics that would be necessary to solve the complex later levels. The importance of playtesting is a running theme-- the designers constantly return to how running average players through the levels allowed them to fine-tune the design of the levels and achieve the right level of nuance. All this sheds a good deal of light on how these games achieve their characteristic effects, especially those “aha” moments where the scales fall from your eyes and the perfected shape of a solution comes into view. Because of Valve's dedication to being patient with the player-- their approach to the player is downright courtly-- their games provide a distinctive sense of earned satisfaction when you tease out the underlying logic of a situation. Zelda may have invented these moments, but hearing the commentary you get the feeling that Valve has refined the art of their manufacture to near-perfection.

I feel like these commentaries amount to a sort of master-class in game design (Valve's philosophy of composition, maybe), and they supplement the game's existent pleasures with feeling of recognition that I associate with good games criticism. So much of making your way through a game is a matter of this subterranean sense of things you get-- you know the space under that panel is important, but you don't know why or how you know it-- and the retrospective making-explicit of all this implicit know-how adds something to the experience as a whole. There's just something satisfying about getting a glimpse of the way these excellent games were organized and structured in order create the memorable experiences that they provide.


Nels Anderson said...

I too found Valve's commentaries to be both enlightening and fascinating. As you mentioned, one of their strengths as a studio is how much emphasis they put on iterative design.

Maybe it's just that Valve is more transparent about their processes, but it seems like they're leading the way when it comes to data-driven feedback and iteration. There may be some costs to implementing automatic feedback systems like this, but as Valve demonstrates, the value is *immense*. By tracking what classes are being played in TF2, and how, they were able to focus their class pack efforts towards the least served classes. Last I heard, Medic playing jumped immensely when the pack was released, but it's since leveled off to pleasantly above pre-pack numbers.

By also tracking similar usage data for single-player games, they're able to determine which areas are the most difficult, how many players give up, etc. Instead of relying on a limited subset of playtesters and irregular (at best) feedback on the internet, tracking play statistics like this transform every player into a kind of playtester. While prototyping and test before the game is released is important, there are patterns that can only be revealed when thousands "playtest" the released product.

It's great to see some of the proven techniques in HCI/UX (my current domain, but I'm trying to move into game development) being utilized successfully by the industry. Creating great interactions in any discipline requires setting objectives, measuring whether or not they've been met and using those lessons to inform future designs. Much as Poe constantly tweaked and adjusted his language, Valve is able to use player feedback gathered both pre- and post-hoc to determine where their design succeeded, where it failed and how it can be improved on the next iteration.

It's a little disenchanting to hear that these things don't spring fully-formed from the minds of designers, but acknowledging that measurement and feedback are essential tools for creating good games might just inspire others to follow and even improve upon Valve's methods. And that means better games for all of us.

Iroquois Pliskin said...


I totally agree with you here, and I didn't know that Valve did that tweaking with the TF classes.

I guess the playtesting info can be disenchanting if you approach games from the perspective of art: you think the poet as someone who just puts his vision out there and is heedless of what the public thinks of it. It's his vision, public be damned if they can't figure out what to make of it.

And when you think of something being "focusgrouped" you get this icky sort of lowest-common-denominator feel, like the content is being made as simple and inoffensive as possible.

But I think the real difference is that games are a functional art: they have to work right, and be intelligible to the player, before they can get around to doing anything else (like making provocative and unpopular statements about the state of the world.) I'm really interested in the functional elements of game design-- how they go about doing what they do-- so i found the commentary really fascinating.