Attention to Innovation: Linking Developer Gaze and Coding Approaches

Creativity is widely seen as a key quality of software developers, yet empirical software engineering rarely examines its manifestation in coding. We take a process-level view by linking psychometric creativity assessments to fine-grained eye-tracking and IDE interaction data from a controlled game implementation task. In a laboratory study with 40 Python-proficient participants implementing scoring logic for a Pong game, we derive programming strategy profiles from joint gaze and code editing behavior and relate them to scores on three well-validated creativity tests.We identify three recurring strategies that differ in IDE interactions, gaze behavior, and the level of visual and structural elaboration. The efficient and functionally focused group shows consistently higher divergent scores and is characterized by a compact, targeted implementation style. These participants focus on a small number of well-placed changes to ensure correct game mechanics, rather than settling for a partial solution or investing heavily in decorative enhancements. Our results offer early evidence of how creativity appears in programming behavior and point to the potential for creativity-aware tools and pedagogical approaches.This Figshare item provides the supplementary materials for our ICPC paper and includes the processed dataset and analysis code used in the study.Contents1.task1-final.csvA consolidated dataset for Task 1 containing:IDE interaction metrics extracted from CodeGRITSCreativity test scores (TTCT, RAT, DAT)The full set of extracted programming and coding strategy features2.analysis.ipynbA Python notebook containing the analysis code used to generate the results reported in the paper.Related publicationThis dataset accompanies: [Attention to Innovation: Linking Developer Gaze and Coding Approaches] (ICPC-ERA 2026).