The Feedback Machine

This browser does not have a Java Plug-in.
Get the latest Java Plug-in here.

Please allow the application a moment to fully load.

How do I use the application? What should I expect?

… it is quite possible for a person to talk to a machine, a machine to a person, and a machine to a machine (Wiener, The Human Use of Human Beings, 76)

To begin with, the console will prompt you to enter your name. Press and hold down the space bar, and say your name aloud. When you release, a new question will appear. Press the spacebar again while you answer this question, and so on.

At first, you can expect silence as you respond to the prompts. Assuming your computer is configured to take audio input from a microphone by default, however (the case for most laptop computers), you will hear snippets of your own answers come back to you. At first, the snippets will be short and completely unintelligible. As the program continues, however, the audio clips will become longer, and will intermingle with machine noises (printers, scanners, modems, sewing machines).

The questions will tend more toward arithmetic, then binary addition. Near the end of the questioning, the audio will transition into chaos—voice clips overlaying voice clips, machine sound making it impossible to tell one from the other. The sound of your own voice relaying endless, echoing digits will dominate. Because the program was playing both machine and human noises as you recorded your answers, the feedback from those recordings will build up and create noise.

When you reach the end of the program, the human sounds will peter out and be replaced by sporadic machine noises, and brief phrases will write themselves across the console as the program completes. Each experience will be different, because almost every step forward is dependent on one or many calls to a random function deterring the order and content of the questions, the vocal audio chosen, and the machine snippets played.

Why these questions, specifically?

These arguments take the form, "I grant you that you can make machines do all the things you have mentioned but you will never be able to make one to do X." Numerous features X are suggested in this connexion. I offer a selection:
Be kind, resourceful, beautiful, friendly, have initiative, have a sense of humour, tell right from wrong, make mistakes, fall in love, enjoy strawberries and cream, make some one fall in love with it, learn from experience, use words properly, be the subject of its own thought, have as much diversity of behaviour as a man, do something really new.

No mechanism could feel (and not merely artificially signal, an easy contrivance) pleasure at its successes, grief when its valves fuse, be warmed by flattery, be made miserable by its mistakes, be charmed by sex, be angry or depressed when it cannot get what it wants."
(Turing, “Computing Machinery and Intelligence”)

The questions asked by the console are divided into three categories: human, arithmetic and machine. The human questions are formed in response to the challenges of machine “thinking,” as explored by Turing in “Computing Machinery and Intelligence.” They represent the questions the program uses to evoke samples of your humanity, as exemplified in your ability to describe sensations of human life, such as emotion, humor and self-reflection, and on your originality.

The arithmetic questions serve as a stylistic bridge between the human questions and the binary computation commands. The binary computations are kept simple so that the user doesn’t just skip them in frustration—so that the program will ultimately have a corpus of “zero/one” voice samples to draw on as the audio response component progresses.

How and why are the samples mixed together?

Another use for speech scramblers could be to impose thought control on a mass scale. Consider the human body and nervous system as unscrambling devices… remember that when the human nervous system unscrambles a scrambled message this will seem to the subject like his very own ideas which just occurred to him, which indeed it did. (Burroughs, “Electronic Revolution,” 24).

The samples are mixed up together to create an effect reminiscent of Burroughs’ “cut-up technique.” Burroughs theorized that the process of deciphering a collection of chopped-up, spliced, and doctored sounds and images would cause the unscrambler to unconsciously ascribe qualities of each sample to one another. Thus, splicing sex noises and the sounds of a disapproving mass into the speech of a senator would effect outrage in his constituents upon playback (14).

More relevantly, Burroughs imagines the use of spliced images to reactivate a virus in a human body recently recovered from an illness via the use of a “cold tape” (30-31). This notion unearths the potential of the cut-up technique to activate a fundamental change within an individual, based on replayed audio and visual images of him or herself.

Why machine noises?

In considering the functions of the mind or the brain we find certain operations which we can explain in purely mechanical terms. This we say does not correspond to the real mind: it is a sort of skin which we must strip off if we are to find the real mind. But then in what remains we find a further skin to be stripped off, and so on. Proceeding in this way do we ever come to the "real" mind, or do we eventually come to the skin which has nothing in it? In the latter case the whole mind is mechanical. (Turing)

Turing delineates many of the arguments against the machine’s potential for thought, and arrives at the conclusion that many of these arguments are based more upon the general limitations of existing technologies—rather than the fundamental limitations of digital processing. He writes:

A man has seen thousands of machines in his lifetime. From what he sees of them he draws a number of general conclusions. They are ugly, each is designed for a very limited purpose… many of these limitations are associated with the very small storage capacity of most machines.

By setting aside for a moment the limitations in storage capacity and computational complexity of modern machines, Turing achieves a perspective on the potential of the machine to successfully process information, learn and respond as a human does. He proposes that there is no unique human element which could preclude the presence of a mind in a machine.

Using machine noises in the context of Burroughs’ cut-up technique allows us to explore the potential break-down of this formerly rigid boundary between computer and human being by juxtaposing aural images of human and computer and inviting you, the user, to unscramble the message. It is important that the resultant audio collage include all three elements: machine sounds, vocalized binary arithmetic, and deeply “human” expressions such as poetry and sensory descriptions. The unscrambled message, then, transcends the image of a human imitating a machine, and addresses the issue of whether the expressions we consider to be uniquely human are fundamentally out of reach for the machine (and vice versa).

What are the phrases which scroll across the console at the end of the program?

These phrases are a further borrowing from Burroughs, who theorized that the inclusion of Ron Hubbard’s “Reactive Mind Phrases,” which supposedly induce illness or mental disturbance in their subject. By reformatting the phrases to suit the machine context (changing “me” to “it,” and replacing, in some cases “body” or “bodies” with “machine” and “machines”), I aim to strengthen the impact of the experience in Burroughs’ tradition.

Why is the project presented within a console-like visual interface?

When I give an order to a machine, the situation is not essentially different from that which arises when I give an order to a person. (Wiener, 16)

The machine which acts on the external world by means of messages is also familiar. (Wiener 23)

The interface becomes a site of simultaneous freedom and obedience for the user—the program initially invites you to contribute input to an open and deeply personal question, but the language gradually shifts from subjective, inviting queries to straightforward commands. The proportion of machine prompts ending in question marks diminishes noticeably as the experience progresses.

Thus, as the program explores the boundary of human and machine via audio collage, the format of the prompts troubles the traditional position of human as commander and machine as slave. By translating the text into computer speech for each prompt, the dichotomy between machine voice and human voice is reinforced to contrast with the progressive breakdown of the power relation between the two entities.

Either machines or humans are in control. However, since the latter possibility is just as obvious as it is trivial, everything depends on how the former is played out. (Kittler, “On the Implementation of Knowledge: Towards a Theory of Hardware”)

Why does the audio become so difficult to understand at the end?

Firstly, unintelligibility is a precondition for the success of the unscrambling process.

However, more importantly, the fact that you ultimately end up simultaneously recording audio input and experiencing replay of human and machine noises on the speakers of your computer means that the computer is recording both new sounds and the echoes of already-recorded ones (note: this only works if you avoid using headphones or otherwise bypassing speakers during the experience). Ultimately, the audio builds up over time, and the final audio playback is highly ridden with noise as a result of the feedback between microphone and speakers during the course of the program.

This actually conflicts with Wiener’s conception of feedback as the factor that makes action possible, via communication between the environment and the individual attempting action. He defines feedback as the “control of the machine on the basis of its actual performance rather than its expected performance” (24). Wiener views feedback as a means for controlling the entropy of a system, and concludes that both human and machine exemplify the reduction of chaos, writing that “there is no reason why [machines] may not resemble human beings in representing pockets of decreasing entropy in a framework in which the large entropy tends to increase” (32).

The main difference lies in Wiener’s focus on negative feedback to the exclusion of positive feedback. He is certainly correct that humans and machines individually rely on negative feedback to interact with their environment. It is interesting that here, at their intersection, entropy, chaos and noise appear to accelerate uncontrollably as a byproduct of positive aural feedback. The Feedback Machine poses a challenge to Wiener’s “theory of the message among men, machines and in society as a sequence of events which… strives to hold back nature’s tendency towards disorder” (27).

Source code: FeedbackMachine.java