+ Goosespeak

human goose interface

The design of the goose translation database is a radical departure from traditional animal research contexts and institutions. OOZ explores a class of problems that are multi-parameter, cross-disciplinary, and that resist formal definitions. All of these problems, I argue, can be fruitfully addressed by diverse non experts, by open structures of participation. The advantage offered by open structures of participation on these intractable problems is simply that more diverse interpretative resources increase the probability of finding coherent or persuasive interpretations.

Basing this design on an open model of interpretation exploits contemporary linguistic theory and extensive evidence that language is a 'bottom-up' phenomena. Innovation in language comes from immature users, marginal participants and external languages, not from expert language users. Meaning emerges through many people having similar interpretations. The thesis of the project/public experiment is to decisively demonstrate this bottom-up characteristic. That is, this database can show that many lay attempts at interpretation provide robust meaning--that lay intelligence matters--rather than singular expert interpretations. The OOZ database will be the first experiment to exploit online connective resources (i.e. lots of people) to demonstrate this. Needless to say, demonstrating this effectively will make a claim about the importance of preserving and enabling open interpretation in other contexts, i.e. national security, politics, environmental data collection and the expanding realm of citizen science

robotic goose / goosespeak interface architecture

OOZ also juxtaposes an open world model of informatics with the "closed world" simulations that dominate the use of computational research, databases, and particularly net art. The limited epistemological potential of closed world / sim world aesthetics is that they require the 'rules' to be a priori. Therefore you can only know and test what you expect to know and test, and the dynamics you can observe stay within the parameters of the world you designed. The open world model encoded into the design of the OOZ database sustains multiple post facto interpretations - it is an open-ended archive of material that is not generated by a set of assumptions and rules. The project is open to the empirical richness of the actual interactions generated by the robotic geese, to the character and moods of the geese. The interface is not passive, it does not obscure the human participants and it does not produce data that claims to passively represent. The robot is a very active interface that inserts itself among the geese, interfering with and generating responses. The robotic goose embodies the idea that to observe a system is to interfere with it.

The OOZ database (a functional prototype exists) demonstrates how the computational (virtual) realm can be designed to have deliberate effects and relationships with the non-computational world. Other online environments draw attention away from the dynamic material context of the offline world, yet the goose database draws attention to the very environmental interactions and phenomena that are in need of attention. Translation is performed by many people, that is, anyone who drives the robot goose can upload an interpretation, or can annotate another's interpretation. The video interpretation of any one clip is a micro (php based) wiki. The geese also are making sense of the robotic goose and adapting their behavior to the actions and utterances of humans (channeling geese) accordingly.

Screenshot of the database's user inteface

The database is assembled with fields that, by default, display in temporal order. The important function that differentiates this project from other databases is that rich, dynamically built interpretive units can accrete in the annotations. This is organized through the "cog blog" system, a unique interface designed in collaboration with PhD student Mariano Belinky, that allows keyword searches to accumulate meaning. Although users can search different key words (e.g. "greeting" or "hello") to retrieve distinct samples of video with annotations containing these words (as they would in any other system), the power of the cogblog system is its ability to redundantly group these distinct terms, and in doing so, to highlight semantic relationships. This augments the interpretive work of participants and provides a system that can organize, concatenate and parse invited diversity. Another unique function is the cogblog's ability to assemble and organize video clips along similar semantic parameters. I have written extensively in the Market Research project and the BANGBANG project on the role of video (i.e. open ended representation format) in engaging diverse people. This project also follows this tradition of using the populist genre of video as data.

Evidence from the initial releases of this project shows that 1) people enjoy driving the robotic goose and 'talking' to actual geese; 2) people immediately speculate about the meaning of the goose responses, and can be guided by others' interpretations, to test, verify, or contest these meanings; 3) people become more interested in their goose neighbors and significantly shift their attitude toward valuing their presence.