The inverted machine – Rechnender Raum (Computing Space) is a light-weight sculpture, constructed from sticks, strings and little plumbs. At the same time it is a full functional logic exact neural network (*). Through its strict geometric and otherwise very filigree construction, the observer is able to track the whole processing logic from every viewpoint around the machine. This disclosure of the machines core is enforced by an uncommon distribution of its constructing elements: a nine angled architectural body forms a torus. In contrast to an ordinary alignment of a hidden logic and an outer user facing display its geometric basis is turned inside-out. The core of the machine, with all its computing elements, is shifted outwards on the surface, while the “display” which indicates the results of the tasks is displaced into the center of the system. Even though the tasks and their logic runs directly in front of the viewers eyes and even if one is long sinking into the interaction of the elements which is accompanied by a polyphonic but steady and reassuring buzz, it is not possible to follow the succession of the single conditions of the machine. On one hand by turning the machine inside-out its function is completly transparent, on the other hand a strict self-referentiality and ignorance to the viewer is realized. The machine turns away from the visitor and carries out its computations only for itself. Without depending on interaction or requesting it, it goes through its own states endlessly. The results of the computations are sent inwards -into its own center- they are not intended for the viewer. So an interesting paradox appears: while the machine opens up everything it closes it at the same time, as if it has a secret.
(*) RR units operate similar to a very basic artifical neural network. Like a McCulloch-Pitts-Neuron with static weights (threshold) it implements the basic boolean functions NOT/AND/OR. RR consists of over 200 boolean units that form a parallel operating cellular automaton.