8월 3주차

For-profit firms increasingly adopt open-source product development by engaging external community members alongside internal employees on social coding platforms such as GitHub. Yet whether and through what mechanisms this engagement affects firm profitability remains an open question. Drawing on the knowledge-based view of the firm, we conceptualize open-source product development as a form of distributed knowledge integration that enhances labor productivity by expanding the specialized expertise available for product development beyond the firm’s internal boundaries. We posit that improvements in labor productivity translate into higher profitability, as labor constitutes a primary input in software product development. However, the labor productivity effect depends on the extent of participation by external contributors, and the resulting profitability gains are shaped by equifinal configurations of firm resource allocation. We examine our theoretical framework using a longitudinal dataset of 977 U.S. high-tech firms from 2001 to 2025 and find that firms adopting open-source product development via GitHub realized, on average, a 4%–5% increase in gross margin. These results are robust across staggered difference-in-differences, generalized synthetic control, instrumental variable, and dynamic panel specifications. A moderated mediation analysis decomposing over 323,000 project-level contributions across more than 44,000 repositories into internal employee and external volunteer sources reveals that labor productivity partially mediates the profitability effect and that this mediation is amplified by external contributor engagement. The indirect effect of open-source development intensity on profitability through labor productivity becomes discernibly positive only beyond a threshold of external volunteer contributions (approximately 35% in our sample). Configurational analysis further reveals that research and development intensity is present across all high-profitability configurations, consistent with the absorptive capacity required to integrate externally sourced knowledge. These findings extend the knowledge-based view to the open-source context and provide managerial guidance for aligning open-source strategies with firms’ resource configurations.

Generative AI content-generation (GCG) platforms enable users to co-create personalized content with remarkable speed. Yet recent research suggests that such immediacy may undermine early engagement: when content appears instantly, users may not realize sufficient value to register on the platform. We address this challenge by introducing fulfillment, i.e., the extent to which co-created content is revealed prior to registration on GCG platforms, as an experiential design lever that shapes value realization in initial interactions. Drawing on value co-creation literature, we suggest that fulfillment operates through two motivational pathways: value-in-use, reflecting users’ recognition that their input meaningfully shaped the output, and curiosity, reflecting anticipatory motivation when the experience remains perceptually open. Using a randomized field experiment on a GCG platform, complemented by a follow-up online experiment, we show that partial fulfillment, which reveals some but not all generated output, outperforms both full and no fulfillment in driving registration. This effect is also conditioned by the framing of the registration message. While loss-framed messages that emphasize the cost of inaction increase registration on average, this effect attenuates under full fulfillment, suggesting a substitution relationship. Formal mediation analyses indicate that although both full and partial fulfillment enhance value-in-use, only partial fulfillment sustains curiosity, and this dual activation explains its effectiveness. Additional analyses delineate the scope of these effects, which persist beyond registration to shape subsequent engagement and return behavior, but arise only when users meaningfully co-produce content and are enhanced by better quality outputs. Together, these findings suggest that registration on GCG platforms depends not on maximizing disclosure or curiosity alone, but on structuring interactions to preserve users’ involvement in shaping generated outputs. In doing so, they highlight how effective design on GCG platforms supports engagement that emerges from complementary human and GenAI contributions, rather than from automation alone.

In the current era of digital business, firms continuously experiment to enhance the online experience of individuals visiting their websites and platforms. The possible changes range from minor tweaks to large product or user experience updates, which are tested before the full rollout to estimate performance and minimize unintended negative outcomes. Because of the clear benefits of digital experimentation, increases in the number of experiments have strained the resource of online participants/customers. In tension with this scarcity, many digital experiments are not carefully powered for their objectives, leading to either over-sampling that wastes resources or under-sampling that weakens inference. These issues warrant methods that can help experimenters balance power and efficient resource use — that is, to sample enough for the proper power without over-sampling. To address this issue, we propose a sequential hypothesis testing method for selecting the best treatment among multiple alternatives for experimenter-specified levels of statistical power and false positive rate. Critically, the method not only samples for no more than the necessary level of statistical power, it also has low sample size variance relative to other methods, meaning that the resulting sample size is a precise estimate of the required sample size, and low bias in the treatment effect estimates, reducing a common problem for adaptive sampling methods. We also demonstrate our method on a dataset from a real multi-armed online experiment which demonstrates the method's efficacy in a realistic scenario. Our method can be implemented in an experimentation pipeline, facilitated with an R package we provide.

Online communities can create complex knowledge products, such as software, by leveraging collaboration and division of labor between their core members and the crowd across tasks. Recent literature has focused on how incentives stimulate crowd contributions. Limited attention has been paid to potential intergroup spillovers, that is, how incentivizing crowd contributions affects nonincentivized core members’ contributions. In this paper, we examine how core members of the Python developer community responded to the Internet Bug Bounty program, an initiative that offers monetary incentives for bug reporting by the crowd. Using a difference-in-differences strategy and the Java community as the control group, we find that the Python core members reduced their contributions to the incentivized task of bug reporting after the program’s implementation. Importantly, they also reduced their contributions to the nonincentivized task of enhancement reporting, a critical knowledge creation task for improving Python. Such reduction has led to a decline in the overall contributions of enhancement reports and related code commits to Python. Further analysis suggests that the negative intergroup spillover of the incentive on core members’ nonincentivized task contributions can plausibly be explained by reduced intertask learning. Our study offers implications for crowdsourcing, the design of incentives for digital public goods, and task delegation. This paper was accepted by Anindya Ghose, information systems. Funding: This research is supported in part by the HKSAR General Research Fund Project 16503620. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2024.04544 .

In a bug bounty program (BBP), organizations incentivize hackers—who could otherwise be “enemies”—to report security vulnerabilities by publicly offering monetary rewards. This paper develops an analytical framework to characterize BBPs and their strategic and economic impacts. Strategically, BBPs induce some hackers to self-select into cooperative vulnerability discovery and reporting, diverting them away from attacking (attack diversion) and delegating part of the protection effort to them (protection delegation). We show that (i) BBPs can be beneficial even when hackers are inefficient at vulnerability identification, yet more participants might make BBPs less attractive to firms. (ii) Although BBPs improve security, they might encourage competitive hackers (i.e., attackers) to exert more efforts. (iii) The coopetition dynamics can render a firm’s provision of rewards and in-house efforts socially inefficient. Common cybersecurity regulations, such as breach penalties and bounty reward subsidies, may exacerbate the inefficiency. We find complementarity between bounty subsidies and breach penalties, suggesting that combining and customizing them according to firm characteristics could enhance BBP efficiency. (iv) Contrary to the conventional wisdom that crowdsourcing rewards increase with participant size, we identify scenarios in which a firm should reduce or make nonmonotonic adjustments to a bounty reward as participant size increases. (v) Legal safe harbor for security testing or policies that reduce duplicate report submissions can lower firm payoffs under BBPs. We draw related implications for research and practice in information security and crowdsourcing. This paper was accepted by Hemant Bhargava, information systems. Funding: This research is supported in part by the HKSAR General Research Fund Project 16503620. Supplemental Material: The online appendix is available at https://doi.org/10.1287/mnsc.2022.02682 .