The Uncanny Valley of Videogames


I’ve recently praised the use of states and derided their absence, but it’s not a one-way street.

To put it simply, states are great for abstractions. They fit unrealistic settings very well, and give us clear signs of what’s actually going on inside the game. They can also diminish the need for complex transitions while making the best use out of limited resources. A clear benefit of this approach is gameplay that can rely on instant, i.e., twitch, responses. For example, a character in Street Fighter II can go from doing a leg sweep to a backflip in a split second without looking too awkward.


Despite being a 3D game, Wind Waker -- like most Zelda titles -- made great use of state-based mechanics.

States are not directly tied to arcade titles, though, nor are they unfit for 3D games. The correlation here is between states and the lack of realism, not states and projection types. The further a game gets away from what we know of real life, the more explanation and experimentation it requires. Sure, Super Mario Bros. might be pretty intuitive, but you can’t really tell how fast Mario can run and jump until you try it out yourself (or see it happen). Conversely, when initially approaching Call of Duty 4, there are many preconceptions for how the characters should move and animate because of their depictions.

Of course CoD4 isn’t a life simulation, but it does aim for what “feels” right and consistently follows its own rules. As does SMB, actually — both titles have a real sense of verisimilitude. One’s just more abstract than the other, and as a result can get away with being much less realistic.

Which leads me to my main point: as the fidelity of games approaches real life, state-based mechanics increasingly detract from that illusion.

But first, let’s take a look at two drastically different titles: Snow Bros. 2 and Grand Theft Auto IV.

In Snow Bros. 2, the enemies have a few basic states. There’s walk, jump, covered in snow (1/4, 2/4, 3/4, or fully), and a couple more. The physics behind movements are very basic, while the transitions are instant. As soon as an enemy gets hit with a shot, its visual representation changes to an animation that shows it on its back struggling to get free.


Say what you will about Assassin's Creed, but the way it allowed Altair to interact with his environment spoiled the audience of third-person action games.

In GTA IV, all game objects respond to a wide variety of variables. The cars don’t have a simple moving/idle/dead state, and their visual representation is a reflection of their physical properties. The cars accelerate on a curve and dip with each turn, while a drunken character’s skeleton animates him as he hangs on to the door handle.

Trying to apply the state mechanics of Snow Bros. 2 to GTA IV would result in ludicrous situations that would detract from its sense of realism. In fact, making the world feel more organic was one of the major improvement of GTA IV over GTA III.

And this gradual raising-of-the-bar is to be expected in videogames in general. Dialogues were once just text, with the occasional frame or two of a “talking head.” These days they’re fully voiced and lipsynced, and character models even emote and use body language. However, many games that rely on a realistic presentation still insist on state-based mechanics. I think one of the more notable examples of this is the upcoming Final Fantasy XIII.

Here’s a video of its demo:

Various fans have praised this game’s detail and fidelity, providing glowing commentary on its rendering of hair and other such tidbits. Well, it is quite a long stretch from the deformed pixel art of the older titles, but it’s also a clear example of dissonance between visuals and mechanics. The player character’s movement has an instant acceleration, and, when she gets stuck on a wall, she performs that old running-man animation. The topography of the obstacle at the 0:52 mark is also quite complex, but its collidable surface is represented by a giant, invisible block. Furthermore, traversing this obstacle is done with a single button press that initiates an instant and perfect jump. This movement is entirely scripted, and it looks quite awkward and unreal when contrasted with the scope of the environment and the proportions of the character.

Now I’m not sure if this is technically an example of the uncanny. After all, that phenomenon describes a feeling of unease brought on by an almost-but-not-quite-real object, and I don’t think anyone would describe the above example as being entirely realistic. Still, the uncanny concept deals with the contradictions between what’s expected and what’s witnessed, and I think high-fidelity games that rely on state-driven mechanics embody that point quite well.

, , , , , , , , , , , , , , , , , ,

  1. #1 by John T on June 9, 2009 - 1:06 pm

    Very interesting read. And its about time someone stopped fawning over FFXIII and its kin, and nailed them on some pretty obvious flaws.

  2. #2 by Chris on July 19, 2009 - 6:21 am

    It’s not the uncanny valley effect, it’s a failure to create an environment that gives the player ‘willing suspension of disbelief’.

    Neogaf had a thread a few days about it.

  3. #3 by The Management on July 19, 2009 - 1:35 pm

    Well, suspension of disbelief refers to accepting fantastic or unrealistic elements of a setting/story as whole. What I tried to focus on in the above post were realistic-looking entities whose behaviour is driven by simple state-based mechanics, which in turn can bring on the sensation caused by the uncanny valley.

  4. #4 by William Moore on June 30, 2010 - 3:53 am

    Grand Theft Auto is so damn addicting game. I play 8 hours a day of GTA.:-*

  5. #5 by Hayden Robertson on July 21, 2010 - 1:29 am

    the latest version of Grand Theft Auto have more detail on its graphics, nicely done.”`*

  6. #6 by Paul on June 27, 2011 - 9:28 pm

    Wrong. The uncanny valley deals with the perception of robots or computer animations which try to emulate real human beings. It’s got little to do with some character instantly starting to run at full speed.

    This is especially important when trying to animate facial expressions or gestures. Again: Only if the subject claims to be a real human but falls short in terms of realism. Link of “Wind Waker” does not look real by any means. One can clearly identify him as a caricature. Emulations of humans that fail to look real appear to us like diseased or ill individuals and our subconsciousness tells us to avoid them.

    • #7 by The Management on June 27, 2011 - 9:39 pm

      Wind Waker wasn’t used as an example of the uncanny valley, FF XIII was, and that game had some serious problems with its somewhat life-like characters.

(will not be published)