by Nik Gaffney - position paper for 'designing robotic artifacts' workshop, 16th October 2010 at NordiCHI 2010

Based on three assumptions; 1) that we are living in a “robot cargo cult”, 2) that animism and materialism are incompatible, and 3) that we need ways of better understanding contextually sensitive human-machine complexes if we are to successfully design semi-autonomous artefacts.

We are building machines to live up to our expectations of how they should behave, based on what we have been told they can do. Pop culture, cinema, science fiction are thick with description, yet our artefacts are often created in “imitative rituals that are conducted without understanding the underlying cause of a phenomenon.” We are living in a Robot Cargo Cult¹⁾. The central claim of “Animism, Rinri, Modernization”²⁾ is that an Animist perspective on robotics, is at odds with the Materialist assumption that a robotic subjectivity is realistic, achievable and substrate independent. The animist position suggests ethics as an essential part of robotic development, while a materialist perspective suggests a flexible, rewritable, moral software as something independent.

Conversely, in “Bots on The Ground”³⁾, it seems that an imposed subjectivity, identity or intent is inevitable. Hyper-functionalist, mine clearing robots, in a war zone, become trusted companions of the soldiers that use them. Robots designed with no sense of character or personality are imbued with such characteristics by virtue of their involvement in human activities. These perspectives provide insight into what may be a deeper relation, a shared contextuality that is often absent in the design, development and deployment of robotic systems. What changes with a materialist to animist shift? Are there points of compatibility?⁴⁾ how does this affect the design, engineering and use of robots or semi-autonomous artefacts?

There may be existing tools to help with this transition; for example, Panarchic⁵⁾ control systems can provide the designer with the possibility to model “evolving hierarchical systems with multiple interrelated elements”, while Rosen Modeling Relations⁶⁾ (or similar contextual modeling formalisms) may provide a theoretical grounding for highly context sensitive complex systems.

This suggests a series of shifts; from…

1. accuracy to quirks
2. repeatability to consistency
3. error to ambiguity
4. processing to play
5. control to emergence
6. mechanism to dynamism
7. fragility to redundancy
8. mono- to multi-

(…)

— see also; category robotics, project lirec