Subservient Chicken

Subservient Chicken (Mask)

If we conceive of text command line art as “you type (words and phrases), it types back” then what if it talks back… or displays photo collages… or plays video clips?

The “Subservient Chicken” web interface is an interesting example of the freedoms and limits that come with wiring an IF or chatbot command line style typing interface to a non-textual response.

“Subservient Chicken” was a viral marketing campaign done for Burger King in April 2004 by marketing firm Crispin, Porter and Bogusky. The goal was to promote the BK Chicken Tender sandwich. It presents a mock webcam, in which an actor dressed in a large homemade chicken costume and garter belts appears to wait in his living room, hoping to be ordered around. The parody of net-fetish webcam cultures makes the piece feel edgy and fresh - the concept is off the wall, although it appears to riff on the somewhat subservient Burger King motto “have it your way” and evn might echo an old urban legend about Perdue ads mistranslating “It takes a tough man to make a tender chicken” into “It takes a hard man to make a chicken aroused.”

At any rate, rather than a live actor on a webcam, a simple Eliza-style parser picks words out of the typed text and chooses which of a hundred or so video clips to play in response - although, as with most chatbots, the trick is to make a small number of clips and recognition patters appear to cover a very large number of interactions. These clips are generally ambiguous charades of various kinds. My general observation is that, by escaping written language, the ad effectively creates the illusion that much more sense is being made than is actually the case. By comparison, sophisticated chatbots and IF characters take on much more - and their failures are more evident for it. Like many chatbots and IF characters, failures of sense in this case can be ascribed to the unusual nature of the interlocutor - that is, he is clearly nuts (where others of often ornery, drunk, absorbed, etc. etc.).

Snopes article
Clickz article
Clickz interview

6 Responses to “Subservient Chicken”

  1. 1 Mark Marino

    To me this counts as effective symbolic exchange. After hearing students complain about how boring SmarterChild is all evening, I was relieved to hear them laughing at Subservient Chicken. The dark decor, the flicker of the webcam, the choppy images, and the outrageous and somewhat disturbing costume entertain while other chatbots merely wait or their animated eyes follow the cursor around the screen. Paula SG’s video has a similar effect. The discontinuous looping video draws us in and as we help fill out the diegetic world we are engrossed. There is something very compelling about digital video/photographs staring at us. Also, the responses are amusing and unpredictable, like some of the most rewarding IF. The question is: What can text-only chabot makers learn from the bot.

  2. 2 Christy Dena

    There is a few issues that both of you have brought up:

    Jeremy said:

    My general observation is that, by escaping written language, the ad effectively creates the illusion that much more sense is being made than is actually the case.

    This is a good point. But I’m wondering if it is an illusion of sense being made or a diversion away from the expectation of it making sense? You’re probably familiar with Masahiro Mori’s theory of the ‘uncanny valley’ that explains how when an avatar or robot is actually like a human there comes a point when it will fail to be real and so the user is suddenly repelled by the ‘uncanniness’ of the, now, creature. This is often brought up in the debate about ‘believable characters’ and the reason why many practitioners/researchers, such as Ruth Aylett, choose to use cartoon-like characters. A person in a chicken-suit is pretty abstract and puts the user into a whole other paradigm of expectation.

    If we go back to the ‘illusion of sense’ then we’ve got to ask what is different to facilitate this impression? An image? The lack of text as you suggest? I presume you mean that text output has the propensity of showing gross deviations in grammar and meaning. Whereas the chicken process of a limited amount of input options with only one default (that of the chicken moving to the camera and waving it’s finger in disapproval - well this is the only one I found) guarantees that every input will be addressed with a response that makes sense. But that is because the sphere of possibility has been limited, the rules of interaction clear and the benefits of adhering to the rules great. This, I believe, can be achieved with text. But many bots are created with the intention to be capable to capture every input as a human would — which is what your point about ‘chatbots and IF characters take on much more - and their failures are more evident for it’ recognises. Not all botmasters though.

    I think on the one hand there is a need, because the natural-language processing technology is not ?there? yet, to define the bounds of the interaction. What will work and what won?t. Indeed, this is the basis of any interactive program ? constraints need to be set. So, I see part of the success of the sense of the chicken has been because of the limited input/output options and the movement away from written text. But the later, the movement away from text, is not because of the ?failures? you so sweepingly designate to chatbots and IF characters (poor NLP) but because of the power of the WORD. You know: the effect a typed word has on a person ? the direct cable to imagination, almost unmediated communication?the star of this blog. Typed text over the Internet also has close ties with what real people do. Text then is like a humanoid and so suffers the same ?uncanny? fate.

    But as Mark observes, it is refreshing to have something jump around and give us an aesthetic other than the cursor-following avatars. They are/have rapidly lost their impact within seconds of use unfortunately. Although I think it?s wonderfully exciting to have your text read out by an embodied bot with a flick of a free program ? they seem to have fallen under the same curse as PowerPoint.

    Mark finishes with a good question:

    The question is: What can text-only chabot makers learn from the bot.

    I think I?ve addressed this question somewhat, and this post has helped me to clarify the point. Bots that are enjoyable to me are ones that I can GET TO DO things. The bot then becomes a generator of sorts. Indeed, what interactive work doesn?t function under the system of generation? The chicken shows text-based botmasters that users want bots to jump hoops, or at least have the option to tell them to. The jumping hoops can be outside of the storyworld and often are: like telling me how I can get a transcript of the conversation, change skins, be told a joke or a story. Alan is a great example of this proactive interaction. Funnily, I become ?closer? to a bot when these things occur and feel the pressure of willing-the-bot-into-life floats away. If bots are to be ?real? then they would know how to do cool things with interfaces, with computers, they would teach us how to hack the server whilst their botmaster is offline.

  3. 3 Jeremy Douglass

    A followup:

    It looks like Microsoft has rolled the basic text-parser-with-video-matching formula into its search engine interface “Ms. Dewey” (via The Star Tribune). The use of fade techniques and blue-screen creates an impressive prototype and a far more polished experience than the rough-cut viral-marketing look in Subservient Chicken.

    Still, the design concept is a bit mystifying, even as a novelty. Ms. Dewey snarks, smirks, vamps, and waxes impatient in a constant stream of high quality video clips - in other words, she does her level best to distract from and derail anything resembling a search process, which occurs in the form of 2-4 partly visible results in a little side window, and a list that can be awkwardly navigated through mouse hovering. If there is a killer application for the command-line-and-video-clip genre, this isn’t it.

  4. 4 Mark Marino

    Excellent site, Jeremy. I agree with your assessment of the human computer interface usability aspect.

    Like many embodied conversational agents, she reads as having a non-Caucasian race (the actress is of Indian and Dutch descent, according to the wonderful wikipedia.) and dressed as she is in her sleek black outfit, she is played by Janina Gavankar from the L-Word. This exemplifies my argument in my dissertation about the tendency to build sexualized chatbot “others.”

  5. 5 Christy Dena

    I don’t understand how this can encourage using this as a search engine when all I get is abused for my inane choices and for taking too long! It looks slick, sure, which is great, but I don’t want to be abused while I’m working!

  1. 1 My Cybertwin, a social web chatbot at WRT: Writer Response Theory

Leave a Reply

thesis writing service