From Attention to Achievement: A Theory-Driven Machine Learning Analysis of Popularity and Success in IoT Innovation Communities
Online innovation communities have emerged as vital infrastructures for the development and diffusion of Internet of Things (IoT) technologies (Atzori et al., 2010). Platforms such as Hackster.io allow developers to share projects, attract peer attention, and compete for formal recognition. Despite their growing economic importance — the global IoT ecosystem is projected to exceed $1.1 trillion — prior research has conflated two conceptually distinct outcomes: popularity (community-driven user engagement) (Zadeh & Sharda, 2014) and success (institutionally-validated competitive achievement) (DeLone & McLean, 2003). This conflation obscures fundamentally different generative mechanisms and creates misleading implications for platform governance, developer strategy, and open innovation theory. This study develops and empirically tests a two-stage theoretical framework grounded in Attention Economy theory, Social Proof, Signaling Theory, Human Capital Theory, and Tournament Theory. Using a dataset of over 26,000 IoT projects from Hackster.io (Ghaleb et al., 2022) and three machine learning algorithms — linear regression, decision trees, and neural networks — we model both project likes (Stage 1: attention-driven popularity) and contest-winning status (Stage 2: evaluation-driven success). A bridging hypothesis examines whether popularity translates into success. Results reveal a structural decoupling of attention and achievement. Popularity is primarily driven by developer follower count and content richness (images, description length), whereas success is primarily driven by developer reputation (prior awards), competitive experience (contest participation), and developer tenure. Critically, estimated project completion time — a proxy for creative effort — shows negligible predictive value for either outcome, constituting an 'effort paradox' that challenges rational signaling assumptions. Neural network models substantially outperform regression, confirming non-linear dynamics in popularity formation. Theoretical and practical implications for platform designers, IoT developers, and digital innovation scholars are discussed.