Innovation in Indie Games

…the panel!

So, I’m moderating a panel at the Independent Games Summit at the Game Developer’s Conference this year. The topic is “Innovation in Indie Games.” Here is the brief, per the

“Join the luminary creators of the Experimental Gameplay Project at CMU, IGF-winning Braid, flOw, and Cloud, and the brilliant Everyday Shooter as they dissect innovation in indie games. How do we generate Earth-shattering ideas that will change the face of gaming? Can small teams innovate? Is ‘innovation’ really what we want?”

Needless to say, I’m over the moon. I got exactly who I wanted for the panel:

John Blow – Co-founder of the Indie Game Jam and Experimental Gameplay Workshop, creator of the IGF-winning “Braid“.

Jon MakHis works include Gate 88, ToJam Thing (a Toronto Game Jam ’06 contribution), and the soon-to-be released (and highly IGF-winning) Everyday Shooter.

Kyle Gabler - Co-creator of the Experimental Gameplay Project at Carnegie Mellon and the accompanying website, games, and, arguably, the whole rapid prototyping craze that’s been sweeping around lately.

Jenova Chen – Co-creator of flOw and Cloud, founder and creative director of That Game Company, and designer on the DS version of Spore.

I’ve designed the format for my panel and written a long list of questions, such as:

Do constraints breed creativity? If you had unlimited resources, what game would you make?

How much can you innovate inside the context of a game people want to play? Not fun per se but just something that people want to play.

What are the metrics of success for innovation?

The Independent Games Summit has sold out, meaning that it has 500 attendees registered. What, I wonder, would everyone else like to ask these amazing, creative, luminary designers? If you’re in the audience you’re free (but not guaranteed) to ask your questions in person. If not, I’d love to hear from you.

Have anything you want to ask John, Jon, Jenova, and/or Kyle?

Deconstructing “Feel” (3 of 3)

I’m just going to go ahead and post this. I’ve got some other stuff I want to write about :).

Tight and Responsive
“The controls feel tight and very responsive; there’s almost none of the “lag” that you get in some other simulation games where you need to wait for a player to finish his animation before passing or shooting.” (World Tour Soccer ’06)

“The ships have an appropriately floaty feel to them, without completely sacrificing responsiveness.” (Quantum Redshift)

Tight and responsive seem to be on the same spectrum as floaty and twitchy, possibly in the center. Both floaty and twitchy feels are generally considered negative, whereas a tight, responsive feel seems to be highly desirable. There is just the right amount of lag between input and action and the level objects are spaced in such a way that the player feels they have ample time to respond to obstacles or changes in the terrain when moving at speed. In many games, such as Grand Theft Auto, this feeling of responsiveness comes from a non-linear mapping between forward motion and turning, as described above.

Loose, Fluid, Relaxed versus Sloppy, Sluggish, Unresponsive

“…the controls feel kind of sloppy and loose at low speed, but the plane gets tight and responsive as you speed up.” (X-Plane)

“Unlike Tony’s game, which revels in its tight fluid-like controls, ESPN seems to almost fight against you with its sluggishness and unnatural feel.” (ESPN X Games Skateboarding)

“You’ll find that the controls are just a little too sluggish, making it feel as though you’re driving a run-down school bus instead of a quick and nimble jet fighter or helicopter.” (Aero Elite: Combat Academy)

This feeling comes from a delay between input and reaction, often caused by dampening or softening of motion (as described earlier – a way of attaining more reaction sensitivity.) A rough way to measure this is timing the delay between input and complete reaction. In some instances, this seems to be a good thing, such as the controls of the Warthog vehicle in Halo. In these cases, the descriptors tend towards loose, fluid, and relaxed, and are generally understood to be positive. Again, this seems to have a lot to do with context: if the obstacles and challenges presented are spaced such that the player has plenty of time to react, the feel will be good. When the challenges come too quickly or when the dampening is overdone, players tend to describe the controls as sloppy, sluggish, and unresponsive.


“…while the gameplay is basically similar to the TH games, ESPN’s controls are tragically stiff and unresponsive.” (ESPN X Games Skateboarding)

Stiffness is the opposite of fluidity or a relaxed feel and often arises when a triggered action locks the player into a predetermined path or action for some duration, when there is little or no reaction sensitivity. As noted earlier, in the game Ghosts and Goblins Arthur always follows the same trajectory in his jumps and comes to a complete halt upon landing. This, especially when compared to the reaction sensitive Super Mario Brothers, is an extremely stiff feel. Instead of disconnecting the controls for a certain duration, consider changing to a reduced value (a state change – more below) or adding global dampening to the system (and raising the movement parameters commensurately to compensate.)

4. State Management & Transitions – Altering mapping and/or tuning in real time to afford the player more expressivity and manipulate the game’s feel.

When a state change takes place, it alters either tuning or mapping. The classic example is jumping in Super Mario Brothers. When Mario is in the air, the speed of his left and right movement (local tuning) is reduced. Functionally, we’re mapping more reactions to a single set of inputs. We still have the same input device, the controller, which still has its two buttons and directional pad, but now we have two different feels in one. Mario feels different in the air than on the ground. The change in feel brings each feel into greater relief and context. In Mario this is a harmonious juxtaposition, each complimenting the other. This shows that there can be a great beauty in switching states and experiencing different feels in rapid succession.

Another benefit of switching states is greater expressivity. Using the same set of inputs, we’ve achieved greater reaction sensitivity. Some games take this even further. For example the Tony Hawk series provides five different main states (air, ground, grind, manual, run) from which each button and each combination of buttons on the controller triggers a completely different move. Interestingly, the transitions are seamless: the player simply views it as an ability, available for use at any time.

Once again returning to our simple Asteroids game, let’s add a ‘turbo’ state. We need to modify our system design to accommodate three new parameters: turbo thruster, turbo left, and turbo right, and add another button to our mapping, the turbo button. When held down, the turbo button will change the values of rotate left, rotate right, and thruster to our modified turbo versions of those same parameters. When released, they revert. Because we’re doing the rotational dampening globally, we’ll still have the benefits of the rotation gradually speeding up to its maximum and slowing gradually back down when the button is released. Once we have this set up, we tune the new turbo numbers and test, honing in on the feel we’re going for, which is an increased sense of speed with a reduced rotational control. Perhaps we reduce the left and right rotation as we increase the forward speed. And we may need to adjust our level design to accommodate this new higher maximum speed, spacing the asteroids further apart.

Again, because these are virtual sensations, impression is the only thing which conveys feel. Switching from one feel to another aids impression by providing context for each separate feel. A feel which would otherwise be considered floaty becomes tight and snappy when juxtaposed with one which is much looser, which has less carving.

“Feel” is an aspect of games that players and designers discuss in abstract, intuitive, subjective terms. Mechanic design consists of four major disciplines that inform the feel of interactive aesthetics: system design, mapping, tuning, and state management. Feel is one of the most interesting emergent properties of human-computer interaction and the methodical categorizations of its components will assist gamers and designers alike. I would be excited to pursue a deeper, more structured study of players’ descriptions of the easily recognized but poorly articulated phenomenon of feel in digital games.

Deconstructing “Feel” (2 of 3)

3. Tuning - extensive, minute adjustments of the specific parameters governing the movement of the player-controlled avatar.

Once our system is designed and our motions mapped, we are left with a set of parameters. Ideally, these parameters can be viewed simultaneously, because what is important is to view them as a cohesive whole, to understand the relationships between them. Balancing them against one another – making small alterations, testing the results – is the primary way to arrive at a certain feel.

At this point, is useful intellectually to delineate between local and global parameters. Local parameters apply only to the specific avatar object which is being controlled. Mario’s jump, for example, is a local parameter, as is the speed of his left and right movement (which will be tuned as one parameter as it should always be the same left as right.) Generally speaking, any motion that is directly triggered by the player is a local parameter. It is in these parameters primarily we can address player feedback about feel.

A global parameter is one that affects all objects equally, such as gravity. Oftentimes, pairs of parameters work as foils to one another and must be balanced in pairs. Gravity versus jump power, for example. Less gravity means greater jump height and vice versa. Remember, though, that gravity is a constant that every object in the game uses and therefore changing it has significant ramifications for the interaction of all game objects. To return to the system design for our Asteroids example, do we want all objects in our world to be dampened? For this game, probably not. We want the asteroids to float around as if in space, and the ship to continue forwards endlessly, frictionless, until the player fires the thruster while facing the opposite direction to slow down or change course. So, all we really want is rotational dampening on the ship, a local parameter. We alter our system to accommodate this, and re-tune all our parameters to achieve a more responsive feel.

One final consideration for tuning is level design. The spatial context in which the movement of the avatar occurs is of paramount importance. Much as the sensation of speed in an airplane at 10,000 feet is less than a car on the freeway, the feel of a mechanic needs context to have meaning. The plane is moving much more quickly than the car, obviously, but there is no impression of speed because there’s nothing flying by the window to use as visual reference. Virtual sensation is entirely comprised of these kinds of impressions, so the design of the level, the layout of objects around the avatar, their size and so on, is the framework for processing the feel of a game. A driving mechanic may feel clunky and unresponsive if obstacles are spaced too closely relative to forward speed and turning (causing constant collisions.) If the obstacles are spaced further apart or the forward acceleration and steering parameters are adjusted, the feel becomes smooth. In this way, level design is simply another parameter to balance local and global values against.

Player classifications of feel and their meaning:
Below, I have classified the common player descriptions of feel mentioned earlier by attempting to correlate them to the parameters, global and local, which give rise to and affect them. There seem to be some redundancies and overlap, such as between “twitchy” and “touchy”, and in many cases the descriptors seem to pair off as opposite extremes (tight versus loose, stiff versus smooth and so on.) A possible next step in formalizing these descriptions of feel in game could be survey and interview-based study looking for more native categorizations and attempting to further correlate them to parameter relationships and known behavioral phenomenon.

“…the action in the game just feels too floaty overall. There’s very little sense of speed or acceleration, either while on the powder or in the air.” (Amped 2)

When a player says a mechanic feels too floaty, they are often referring to the relationship between forward movement and rotation. Specifically, how quickly a player’s turning input causes a change in direction. If the avatar object can pivot a great deal before a change in heading occurs, as though it is sliding across ice, it will feel floaty. If instead it “carves”, seeming to grip the terrain, dig in, and cause a quick, arcing direction change, it will be perceived as being tighter-feeling. If the object being controlled is in contact with the ground, we can emulate the natural phenomenon that causes cars and other vehicles to make sharp, arcing turns, friction. It is also possible to add a dampening force proportional to the amount and/or duration of turning. For example, if a turning force is applied while a car avatar is moving forwards at 10 units per second, an arbitrary force could be applied in a direction opposite to the car’s forward movement depending on the sharpness of the turn, assuming that the input device being used has enough sensitivity to accommodate a nice range of turning. This will cause the car to appear to carve as if on dry asphalt rather than sliding sideways, as if on ice.

Twitchy or Touchy
“Unfortunately, these games are also the most unevenly re-created, with spotty and unfaithful sounds and–worst of all–twitchy controls, regardless of what input method you select…as previously mentioned, the controls for all of these games are very twitchy and overly sensitive.” (Atari: 80 Classic Games in One)

Twitchy seems to be the opposite of floaty, when the controls are too responsive and the player feels as though the slightest movement will send them veering off their desired course. In this case, we would seem to have too much sensitivity, or not enough range of sensitivity. Try making the range between the smallest and largest possible force applied (especially applicable to turning forces) larger, and mapping it non-linearly, such as with a Bezier curve. GTA Mapping

For small movements of the input device, the reaction is very slight. The stronger the input (the farther a thumbstick is pulled away from its neutral position, the faster a mouse is dragged, the longer a button is held) the stronger the turning force.


Part 3 of 3 coming on Saturday (I teach from 8am to 10pm Fridays).

– Swink

Deconstructing “Feel” (1 of 3)

As promised, here’s the text of my submission to Supple Interfaces. I’m experimenting with a good way to divide by pages; I’ve sunk a surprising amount of time into trying to get various WordPress plugins to do this properly, but it seems to come down to something in the Theme. If you know a good way to do this, please let me know :). For now, I’m just making seperate posts.

Part 2 = Wednesday. Enjoy!

Digital games are a new medium and, as such, include many unexplored areas. This paper examines one such area, the aesthetic properties of interactivity, more commonly known as the “feel” of controlling a game. While the descriptions used to articulate this feel are often vague and esoteric, they tend to be consistent across players and game designers. I will attempt to classify these poorly articulated descriptions in systematic terms, correlating them to four specific, practicable disciplines of interactive aesthetics.

There exists in the collective minds of video game players a deep and nuanced classification of the “feel” of the games they play and enjoy. This feel is afforded by real-time interactivity. A player is able to trigger an input – move a mouse, press a button – and the game reacts immediately. With the barriers between intent, action, and reaction thusly removed it is possible to experience a kind of “virtual sensation”, exercising kinesthetic control over a purely digital entity in the same way you would steer a bike or drive a car. Like cars, each controllable object in a game has its own feel, based on things like weight, turning radius, and suspension. The similarities continue: virtual sensation can give the same pleasurable feelings of mastery and control, challenge and reward. In many ways virtual sensation is better equipped to create pleasurable experiences than real sensation. In a virtual world, any property of any object can be changed in a heartbeat in favor of one that feels better to control. Gravity can be reduced, friction eliminated, difficulty can be adjusted, reality changed. Physical danger is removed. Feel in a digital game, then, is an evolution of the satisfying, life-enriching sensation of skillful manipulation.

Feel is not artwork, immersion, or theme. The interactive aesthetics that give rise to feel must be separated from traditional visual and aural aesthetics. The quality of the painting and drawing, musical composition, graphic design, sound design, character design, and animation; these things are traditional aesthetics, not interactive ones. There’s no question that the character design and texture painting in a modern Final Fantasy game is well executed from an artistic standpoint, but this has no effect on how it feels to control a character in the game. Also, players will invoke feel to describe the theme of the game “a cool Western feel” or the qualities of immersion (flow) they’re feeling while playing a game “You feel like you’re really there, like you’re in the game.” “The HUD It´s frickin ugly, it totally spoils the feeling of being there.” These are red herrings when discussing feel in digital games.

When players do describe the feel of exercising control over a game avatar, they do so in kinesthetic terms. The game feels “floaty”, “loose”, or “twitchy.” When players say things like “the controls feel tight and very responsive”, “the controls feel sloppy and loose”, or “that crisp feel of control”, they are responding to the interactive aesthetics that give rise to feel. Floaty, loose, twitchy, sloppy, tight, sluggish, responsive, relaxed, stiff, fluid, unnatural, smooth, clunky, touchy; these are the common, recurring descriptions players use when attempting to articulate the feel of controlling a game. As a game designer, these descriptions are frustratingly vague. If a player tells you that your game “feels too floaty”, how do you reconcile that with the abstracted variables of your system? What numbers do you change, and by how much? How do you know when you’ve got it right? How do we as designers come to terms with how players feel our games? How is feel in games created?

Disciplines of Interactive Aesthetics
Below I have outlined four practical disciplines that, in my experience, must be applied to create a good feeling game. In addition, I offer anecdotes, examples, and advice for practitioners.

1. Mechanic System Design – Creating a framework or system in which it is possible, through mapping and tuning, to produce the desired feel.

This is where the feel of a game begins. Before any tuning or tweaking of parameters can occur, we must first define what those parameters are and the relationships between them. Game Designer Chris Crawford suggests first constructing a “verb list”, defining all the actions that will be available to the player. For example, defining the verbs for a simple game like Asteroids would yield something like this:

• Rotate (Left or Right)
• Fire Thruster
• Fire Shot

Asteroids Clone

Ignoring the shot, we’re left with the motion of the ship, where the feel of Asteroids primarily resides. To construct this system, we will first need an object that can be rotated left and right and moved forwards, so the object needs an obvious forward-pointing direction. A triangle fulfills these requirements nicely. Also, there are certain assumptions underlying the relationships between these parameters. For example, the speed of rotation for left and right rotation needs to be the same. When examined, it seems somewhat arbitrary, but the user will expect it because the two rotational values are presented as mirrored. So, the “simple” act of designing and setting up a system belies a series of subtle design choices: what actions will be available to the player, and what will be the relationship between them? What will the object being controlled look like does it have any special functional requirements (such as having a clearly defined front and back)? Mechanic system design, then, is the big picture; it is informed by the disciplines of mapping and tuning, but is arguably the most important. It is impossible to arrive at a desired feel through tuning if underlying system is not capable of producing it.

2. Mapping – Defining the relationship between user input and game reaction.

To return to Asteroids, we have our triangular object and it will rotate and move. How are we going to trigger these three motions (rotate left, rotate right, and thruster)? This question indicates another set of small, subtle design decisions. Assuming that the input device is a keyboard, which buttons do we choose to map to which motion? Where are they positioned relative to one another? Does the rotation of the ship start when the button is in the pressed state and stop when the button is released? Or does the button press start the ship rotating, waiting another press of the same button stop it? What happens if both rotate buttons are pressed simultaneously? This is the discipline of mapping, defining exactly what response the game will offer given a particular input. When a player says a game feels’ unnatural’, this is the culprit.

Mapping marries the physical motion afforded by the input device to some corresponding reaction in the game. This is neat because, as mentioned earlier, motion in a game is bound by nothing. Anything can be a good positional metaphor, anything can make sense, there are no physical laws binding what can and cannot happen in reaction to a given input. We can control a beetle pushing a golf ball or a star flying through the night sky. The only thing that matters is that there is a strong, intuitive, easy to understand correlation between physical manipulation of the input device and reaction from the game. In addition, it is useful to utilize accepted standards and conventions wherever possible (such as using the keyboard keys W, A, S, and D to control forward, backwards, left, and right motions respectively – a common convention.) If the mapping has gone awry users will inform you immediately and vocally. They will be frustrated and confused, asking very basic questions about how to jump, how to get around. So mapping acts as a gatekeeper: for a player to enjoy the feel of a game, the mapping of their input to game reaction must be so intuitive as to be transparent.

Another consideration when mapping is finding the right amount of expressivity. If we consider, in the most general sense, the expressivity of a mechanic to be the sum total of the physical sensitivity represented by the input device and the virtual sensitivity afforded by the reactions to that input by the game, we can get a rough estimate of the expressivity of a given mechanic. For example, a mouse is a highly sensitive input device, especially as compared to a standard two-state button. Jumping in Super Mario Brothers is highly reaction sensitive (the longer you hold the button, the higher the jump, Mario slides gradually to a halt) whereas the jumping Ghosts and Goblins has far less reaction sensitivity (Arthur always follows the same trajectory in his jumps and comes to a complete halt upon landing.) Very little reaction sensitivity results in what players describe as a stiff or unresponsive feel. The trick is to strike a balance; we want as much expressivity as we can get while keeping the simplest, most intuitive control mapping possible. The lower the barrier to entry, the more quickly they can experience the intended feel of the game and appreciate its beauty.

Continued Wednesday…

Random Musings

@Guitar Hero 3

Mixed feelings. I worked at Neversoft so I know a lot of the people who’ll be working on this. Expect to see a Thunderlords song, and the guy in the Viking hat there, Alan, to be the lead designer. And huge ups to Dave Rowe, the audio-cranking beast machine. It’s gotta be a dream come true. But, seriously, harsh there Activision dudes. Activision has a habit of taking franchises and licenses away from the developers who created them and giving them to other studios they own…with mixed results. It worked out ok with Treyarch and Spider-man, but don’t think for a second there isn’t still some residual animosity at Neversoft over losing Spidey. Neversoft solved the major problems of a Spidey game in 3d waaay back, and were rewarded for their highly successful (2+ million units sold!) efforts by an insulting and unceremonious handing off of the franchise to Treyarch. It makes business sense, obviously; they’re using Neversoft as a multimillion dollar pinch hitter.

Unfortunately, Spiderman ≠ Guitar Hero. Here’s why: Harmonix is a music company. In order to be hired there, I’m told, you must, in addition to being really f’ing smart and really f’ing good at what you do, play an instrument. I’m not talking glockenspiel* here, although it would be badass if you could shred the gloc. You need to be able to hold your own in a jam session with the team at Harmonix and these guys are unbelievable musicians, every one. So, are there people at Neversoft who can make a functional sequel? Yes. Is it a good idea? I’m not so sure it is. And it’s odd, oh so very odd that Neversoft, the darling of Activision, has been tapped as a base runner on this one.

On the plus side, Harmonix is now free to reinvent the music game genre yet again (with Band Hero? Who knows!) The game industry: even when you win, you lose. *sigh*

@ Supple Interfaces

This is coinage by my friend Katherine over at RPI, and part of the title of her workshop “CHI 2007 Workshop on Supple Interfaces”, where CHI = Computer-Human Interaction. I’m conducting a lecture/activity at said workshop and wrote about ten pages “deconstructing feel” in games for my submission, which I’ll post here in chunks starting next Monday. It’s an interesting direction, closely related to my virtual sensation stuff, but it starts with player classifications of feel such as ‘floaty’ and ‘stiff.’ I think I could write an entire book on the subject. *ponder*

@ Flow

Having re-read Jenova’s thesis, I’m more convinced than ever that it was the single most important theoretical contribution to the field last year, and that most everyone missed the point. I think people are confused by his use of Dynamic Difficulty Adjustment (DDA) because of the various connotations and ill-conceived experiments given that name over the years. What he’s espousing is a holistic view of game design that integrates Flow theory into mechanic, system, and rule design. This in contrast to the bolt-on approach most people associate with the acronym DDA. So, as you’re designing your game, as a fundamental part of the process, asking the question ‘how can I allow the player to adjust the challenge to perfectly fit their ability?’ I think the key here is the notion of integration, of making the ability to adjust challenge part of the game’s primary mechanic, giving control of it to the player. Part of it is giving the player freedom and choice, but there are so many more concerns – you need to find the right kind of freedom, the right kind of choice. Giving the player four possible difficulty levels to choose from is a blunt instrument, as is adjusting the actual numeric difficulty of the game based on player performance. Jenova’s insight is to view control over challenge as simply another ability, another verb for the player which needs to be balanced against all the other parts of the system like any other.

The game flOw is a cool experimental step in this more elegant direction, neatly skirting issues of player intelligence (it’s too easy to game a real time number balancing system by lowballing early) and competitive psychology (some players, like me, will always choose the hardest difficulty level.) Anyhow, read the thesis and play the game. They contain a number of simple, startling insights.

*Not to impugn the noble Glockenspiel and its many fine, prodigious players, but it is Guitar Hero for a reason ;).