Puzzles of Aesthetics

A few weeks ago, I complained about the casual game Home Sweet Home that it wasn’t a very entertaining game, being asked to decorate a house to client specifications. (I ragged even more on the “construction” part of the game, which manages to be easy and annoying at the same time, and to bear little or no resemblance to the real-life activity that it is supposed to be simulating.) Other people evidently liked the game more than I did.

Since then, though, I’ve been thinking about this question: how do you design a puzzle or goal-oriented interaction in which the player’s job is to make aesthetic judgments?

I’ve seen a number of different gestures towards this kind of puzzle in recent games.

The obvious challenge here is that, the more you challenge the player to make aesthetic choices, the less possible it is to apply any kind of objective scoring, grading, or standard. So you can either remove the concept of goal-oriented play entirely and allow the player to romp in an aesthetic sandbox (which I think is what Home Sweet Home is trying to do, except that I didn’t find the elements very enjoyable); or you can impose some external standard, curtailing the player’s effective freedom.

Chocolatier 2 allows the player to assemble ingredients together to come up with new recipes, a feature that I thought considerably improved it over the original game; but ultimately the player only succeeds by concocting one of the game’s numerous pre-set recipes. So this barely counts, in the sense that the player is not designing anything novel. At most, we’re using our aesthetic sensibilities to guess what the game designers also thought sounded like a delicious chocolate recipe. (There are other ways to approach the strategy of this level, though; if you don’t want to think about cooking, you can take a Mastermind-esque strategy, laying out ingredients in groups and disqualifying useless ones until you’ve narrowed in on the remaining possibilities.)

JoJo’s Fashion Show takes a different tack. The player’s task is to assemble outfits from an assortment of tops, skirts, pants, shoes, etc., in order to match certain general “looks”, like “hippie”, “rock and roll”, or “summer”. Many different qualities of a piece — the color, material, texture, cut, and decoration — all help determine which styles it best fits; so game play is mostly about learning to recognize the potential associations of a given article of clothing, and assemble outfits rapidly. As far as I can tell from the hour of demo play, the game is not concerned with whether two pieces go with each other well — for instance, if you’ve got two patterned garments, you can use them both in a fall outfit even if the patterns clash horribly. And, of course, there are all kinds of aspects of couture design that don’t come into play; we’re never allowed to meddle with hemlines directly, for instance, but only to pick and choose between pre-created articles of clothing.

Still, I found this reasonably entertaining, as a puzzle in optimizing the outfits I could make out of the pieces available to me. Making outfits that I also found aesthetically appealing was a side bonus, something I couldn’t always achieve with the available materials. But there was considerably more freedom of invention for the player than in the Chocolatier recipe-making scenario.

Some people have noted that JoJo’s Fashion Show can be optimized as a numbers game. There are certain powerups that allow you to see, numerically, how many points a given outfit contributes to a given look; so in theory, if you are thorough enough and work fast enough and memorize point values, you can get an optimal outcome all the time without exercising any kind of subjective aesthetic sense at all. But the thing is, that wouldn’t be a very entertaining way to play, and the structure of the game discourages it, through time limits and restricted access to the point value information.

(This is good. There are other games whose structure encourages boring play — IF doesn’t do this as much as some game styles, but there are still IF puzzles that can be more reliably brute-forced than solved “properly”, and so people do play them that way. See: mazes, combination locks with small combinations, conversations with few but important options, and randomized but UNDO-able combat.)

So I think JoJo’s Fashion Show is on to a good idea here. Roughly, the idea is: what we think is a subjective assessment is really not as subjective as all that; rather it’s an intuitive simultaneous application of a large number of fuzzy rules, in which any given object can be significant in a variety of different ways. The computer can afford to explicitly model these rules, but they are too complicated for people to solve quickly.

This suggests other odd possibilities. What about a photography composition game, where you manipulate lighting and camera settings and select and frame a shot, which is then judged by a large number of internal criteria? And how do we decide what the internal criteria should be? I think for best results you’d probably want something less simple than JoJo’s Fashion Show; I think you’d probably want to start the computer off with some general rules, like the idea of the rule of threes, and then do some AI-training stuff, where the computer collected a lot of feedback from human viewers in order to adjust the relative weights given to these rules. But we’re now wandering far from the realm of my expertise into a shadowy grove of things I know about only from occasional reading in the computer section of Barnes and Noble.

Do any of these possibilities plug back into IF in any interesting way? That I’m also not sure about. It’s hard to envision a text game that meaningly assessed the player’s visual aesthetic ability. And literary aesthetics? Wow, are we not even close to being able to judge that algorithmically. It’s amusing to imagine a game in which the player makes progress by submitting poetry to his NPC Roxane, only to be rejected again and again until he learns to write– but how implausible.

But what about a game of symbolic magic, in which you accomplished things by juxtaposing significant items? What if a spell wasn’t done with a magic word or even (as so beautifully in the case of Suveh Nux) by a magic sentence, or by waving a magic wand or using an amulet, or by sympathetic magic as in Savoir-Faire, but simply by setting together a collection of ordinary game-world objects and performing some sort of activation?

Given the right prompting and feedback, I think the player might be able to develop an aesthetic sense of meaning and potency, rather than a precise set of rules.


13 thoughts on “Puzzles of Aesthetics”

  1. The Nintendo 64 game “Pokemon Snap” was a casual game that used the idea of taking photographs as the main mechanic. At the end of each ride sequence, all of your snapshots were judged by the game and given a grade. Unfortunately, the game preferred close up shots where the creature was centered in the picture, making more artistic compositions not worth as much (hadn’t they thought of the rule of thirds?!).

  2. Probably the most recent game in the photography genre that would interest is is the IGF nominee Secret of Bird Island.

    In the GrailQuest gamebook series, the Poetic Fiend would occassionally require you to write a piece of poetry to match his specifications before you could go on. Because the Poetic Fiend was an awful poet, aesthetic judgement didn’t matter plotwise. I also presume a lot of players skipped the metagame stuff that came up in GrailQuest.

  3. The screenshots on Bird Island make it look like mostly you’re trying to frame the birds accurately; maybe I’ll try it later, though. Clearly there’s more of this kind of thing than I suspected.

  4. Having now tried the Secret of Bird Island: this is kind of what I imagined and kind of not. A lot of the challenge is about speed/accuracy in framing a shot of a moving object (which I imagine is pretty accurate to wildlife photography, but is more about setup and reflexes than it is about aesthetic judgements about what would look nice where). Still, an interesting case.

  5. This is a fascinating and important problem. Is it possible to formalize aesthetics? Can we code art? One approach comes to mind: the completely abstract game. By removing all representative aspects from the game, we remove need to bind ourselves to traditional aesthetic values. In a game without dresses, recipes, guns, people, or anything else real, there is no opportunity to say that this dress clashes, this recipe sucks, etc.

    But it’s still possible for this abstract game to have a system or structure that evaluates the player’s input and determines its potency or effectiveness. Suppose you have a game in which the goal is to move your agent from Point A to Point B, but the speed at which you move is determined by the way in which you control your agent: you notice that if you try to move in a straight line, you move slowly, but if you corkscrew, you move more quickly. After time, you discover that there is a complex pattern that when adhered to, enables you to move lightning fast.

    The most effective method of interacting with this or another abstract game may be a complex, byzantine system that players must spend days experimenting, practising, and learning. This complex system is, in essence, a system of aesthetics: it is mysterious and unfathomable, yet rewards effort with results. It requires patience and understanding, and an experienced player can easily be distinguished from a novice. As a sculptor reveals the meaning hidden in a block of marble, the player of an abstract, labyrinthine game slowly reveals the meaning hidden in the code.

    It occurs to me that what I’ve been talking about suggests that the act of playing the game itself is the artistic, creative action. A player works and works, and eventually achieves mastery over the game, feels at one with the game, aligned with internal structure of the game.

    I see this as a top down approach, where the aesthetic system comes from the game and is learned by the player. I think a bottom up approach is possible as well, in which the game learns the aesthetic system from the players. I’ll give some thought to that one.

  6. There is a game called “Snapshot Adventures” by Reflexive in which you take pictures of birds. The pictures are rated based on whether the bird is centered, close up, in flight, singing, obstructed, lighted well, facing toward or away from the camera, and a few other things, I think. Part of the game involves asking the player to choose between two photographs he has taken which is better, which can be fairly difficult.

    I found the game to be quite boring, although this may reflect on my aesthetic sense as much as the game’s design.

    As for the idea of the player developing some sense of ‘meaning and potency’ of items in your hypothetical approach to magic, I submit that we already do this sort of thing in any game where we have incomplete information. What you’re suggesting seems, to me, like a replayable puzzle, which we gain more information about as we go, until we can construct an appropriate solution, like Spider and Web or especially Rematch.

    The only problem I see is with the idea that the player would not gain knowledge of a precise set of rules; as long as the player is able to acquire enough information, he will probably deduce the rules, or near enough as makes no difference. You can build in a little uncertainty, but since I usually find randomness more frustrating than interesting, I’m not sure that it would help. I don’t know that there would be any way to get around this, since we must produce the rules when writing the game, but it may be possible.

    It’s an interesting idea, though, which I will have to give more thought.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: