Coordinating open-source software (OSS) development requires both the division and integration of developers’ attention. How to combine these contradictory logics effectively, however, remains unclear. Adopting a configurational lens, we distinguish between developers’ primary attention, or their top-priority focus, and secondary attention, or their spare capacity. Along these two lines, division and integration take on different forms. Recombining these forms yields two new coordination logics: interlacing, which combines primary division with secondary integration, and interlocking, which combines primary integration with secondary division. Applying the same or different logics across OSS’s two key work types—discursive and technical—produces a broader set of coordination configurations. Analyzing 4,291 release cycles from 1,004 GitHub machine-learning projects, we find that isomorphic interlacing—applying interlacing in both work types—delivers the strongest performance of development speed, novelty, and quality. Configurations pairing interlacing with division or integration also perform well under particular contingencies of task interdependence, project ownership, and release-cycle length. By contrast, configurations involving interlocking consistently underperform. Our study advances OSS coordination theory by identifying a high-performing configuration that systematically combines division and integration and clarifying when alternative configurations can substitute for it.
Public service communication aims to raise awareness and change behaviors on critical public issues, such as health, safety, and the environment. Public service communication typically takes one of two forms: a public service announcement (PSA), such as a TV or radio advertisement that broadcasts information (a monologue), or a public service interaction (PSI), such as a meeting with officials that both provides and collects information (a dialogue). Celebrity involvement has traditionally been limited to PSAs due to cost and scalability challenges, but advancements in AI now enable the use of digital humans as scalable, interactive “celebrity-as-a-service” solutions. These AI-controlled digital celebrities mimic real people, enabling one-on-one celebrity interactions at scale. In an online experiment, we compared a digital celebrity to a non-celebrity digital human delivering a skin cancer awareness message in both a PSA and a one-on-one PSI. Results showed that both perceived celebrity (celebrity vs. non-celebrity) and interactivity (PSI vs. PSA) increased trustworthiness and enjoyment. Trustworthiness, in turn, increased the intention to comply with the public service message, while both trustworthiness and enjoyment increased the likelihood of sharing the message. We conclude that AI-controlled digital celebrities open the door to incorporating celebrities into one-on-one PSIs to drive more effective, scalable, and cost-effective public service communication.
Crowdfunding platforms can influence project success through their promotion policy, which determines which projects receive enhanced visibility. Promotion policies can directly influence inclusivity on the platform and potentially address well-documented disparities across creators. This study examines how alternative machine learning (ML) problem framings for promotion policy affect fairness and business performance. Using data from 118,780 Kickstarter projects, we compare three ML framings: a process framing that learns to reproduce historical employee promotion decisions, a success framing that predicts whether projects will meet their funding goals, and a backers framing that predicts whether projects will attract a large number of backers. We evaluate the resulting counterfactual promotion sets using multiple fairness and performance measures. In our context, the success framing produces promotion sets with more favorable fairness outcomes and business performance outcomes than the other framings, possibly because it applies more consistent promotion decisions across similar projects. These findings demonstrate that alternative ML problem framings can produce meaningfully different promotion policies, even when trained on the same underlying data. While the success framing prioritizes practical projects, the backers framing favors projects with higher funding goals. Importantly, carefully designing the ML problem framing for promotion policies can align fairness and performance objectives, countering concerns that inclusivity necessarily comes at the expense of performance. This research contributes to the information systems literature by demonstrating that ML problem framing is a consequential choice in platform governance. It offers practical guidance for digital platforms seeking to improve fairness and business performance through ML-assisted promotion policies.
Digital resilience has become increasingly crucial for organizations and professionals in times of major exogenous shocks and escalating crises. Complementing capability-oriented accounts of digital resilience, we examined how resources are created through evolving relations, offering a process view of digital resilience. Through a longitudinal, qualitative study of Greek primary school teachers, we explored how digital resilience emerges through evolving relations despite limited organizational support. Drawing on a resourcing lens, our study establishes that digital resilience emerges and is sustained when professionals, motivated by value-laden beliefs that make them assume responsibility for turning potential resources into resources-in-use to respond to a shock, engage in a process we call relational resourcing. Our model of relational resourcing shows how resourcing convictions are enacted through processes including appropriating personal digital technologies, enlarging relational functionalities, and routinizing new enactments of resourcing convictions. Our findings expand resourcing theory by showing how resourcing activities entail new ways of relating from which resource value derives. Furthermore, the findings add to the relational view of technology by theorizing how and why entities or professionals learn to relate.
Actual sales data are valuable but often inaccessible. Thus, researchers and practitioners rely on sales–sales rank relationship models, such as the power-law distribution, to infer sales information. However, these models often poorly fit the sales distribution of online markets due to the long-tail phenomenon, creating a need for a more precise functional model. This paper proposes a novel model discovery framework, genetic programming with two-phase pruning (GPTP), which combines genetic programming with a newly developed two-phase pruning method to automatically learn a functional model relating sales to sales rank. Through analysis of sales data from both offline and online markets for 203,994 product items in 148 categories, we discover a new natural-power-law model that outperforms the traditional power-law model and provides a more accurate mathematical means of quantifying the long-tail distribution in online markets. The findings demonstrate that the proposed GPTP framework can intelligently learn the generalized knowledge underlying large business datasets and precisely determine the universal functional relationship model across multiple product categories in a white-box manner. The new model will enable researchers and practitioners to more accurately predict actual sales and explore various marketing issues.
