Bottom-up live coding: Analysis of continuous interactions towards predicting programming behaviours

Paper in proceeding, 2022

This paper explores a minimalist approach to live coding using a single input parameter to manipulate the graph structure of a finite state machine through a stream of bits. This constitutes an example of bottom-up live coding, which operates on a low level language to generate a high level structure output. Here we examine systematically how to apply mappings of continuous gestural interactions to develop a bottom-up system for predicting programming behaviours. We conducted a statistical analysis based on a controlled data generation procedure. The findings concur with the subjective experience of the behavior of the system when the user modulates the sampling frequency of a variable clock using a knob as an input device. This suggests that a sequential predictive model may be applied towards the development of a tactically predictive system according to Tanimoto’s hierarchy of liveness. The code is provided in a git repository.