General Collective Intelligence as a System for Facilitating Parallel and Distributed Computing

Abstract

A General Collective Intelligence or GCI is a hypothetical platform able to self-organize individuals or intelligent agents working on behalf of those individuals, into potentially massive networks of cooperation on a self-sustaining basis, where those networks might exponentially increase the complexity and duration of reasoning the group is able to execute. When applied to the computing processes that might be executed by each individual, the algorithm represented by GCI creates the potential to take any computing process that has been abstracted into a logical model (a model in which each component of logic might be implemented in a separate functional component), and to automatically search for and apply patterns of solution by which it might be possible to increase outcomes. Such patterns might involve distributing processing across multiple instances of those functional components, and scaling processing by increasing the instances of those functional components in series and in parallel, as well as ensuring this effort has more value than cost to performance so it is possible to sustain the effort at doing so, until computing outcomes are scaled as necessary to achieve a given target, or until returns from scaling diminish to the point that further investment in scaling is unproductive. This paper explores how such processes of cooperation might include all forms of parallel and distributed activity, how that cooperative activity might be abstracted into a logical model, and how GCI might facilitate parallel and distributed computing through that model.