Invisible Labor in Open Source Software Ecosystems

TL;DR

This study addresses invisible labor in OSS by defining labor visibility/invisibility, developing a cognitive anchoring survey, and applying explanatory mixed methods with $n=142$ participants. It finds that about $1/2$ of OSS labor is invisible and that attribution is inconsistently provided, with visibility framing biasing perceived visibility and credit importance. The work reveals tensions among attribution motivations (expressive, instrumental, non-attribution) and shows that governance and resource constraints shape what is credited. It argues for participatory, multi-stakeholder design of attribution and compensation to improve fairness and transparency in OSS ecosystems, while acknowledging methodological limitations and the need for targeted future research.

Abstract

Invisible labor is work that is either not fully visible or not appropriately compensated. In open source software (OSS) ecosystems, essential tasks that do not involve code (like content moderation) often become invisible to the detriment of individuals and organizations. However, invisible labor is sufficiently difficult to measure that we do not know how much of OSS activities are invisible. Our study addresses this challenge, demonstrating that roughly half of OSS work is invisible. We do this by developing a cognitive anchoring survey technique that measures OSS developer self-assessments of labor visibility and attribution. Survey respondents (n=142) reported that their work is more likely to be invisible (2 in 3 tasks) than visible, and that half (50.1%) is uncompensated. Priming participants with the idea of visibility caused participants to think their work was more visible, and that visibility was less important, than those primed with invisibility. We also found evidence that tensions between attribution motivations probably increase how common invisible labor is. This suggests that advertising OSS activities as "open" may lead contributors to overestimate how visible their labor actually is. Our findings suggest benefits to working with varied stakeholders to make select, collectively valued activities visible, and increasing compensation in valued forms (like attribution, opportunities, or pay) when possible. This could improve fairness in software development while providing greater transparency into work designs that help organizations and communities achieve their goals.

Figures (7)

• Figure 1: (a) Invisible labor is labor that is not fully visible and/or undercompensated. Labor is not fully visible when labor data doesn't exist, isn't shared, or isn't accessible. Labor is undercompensated when it doesn't receive enough credit, pay, or new opportunities, among others forms of compensation. (b) Our study demonstrates that roughly two thirds of labor in OSS is not visible, either by not being visible or only shared with one other person. We also found that about half of OSS labor activities do not receive credit, as demonstrated here by the 9 in 10 individuals who do not receive credit from some fraction of projects they contribute to.
• Figure 2: Distributions of (a) artifacts and (b) use cases that survey participants contribute to. Questions were multiple select, so percents shown represent the fraction of participants who expressed that they work on the artifacts and use cases shown.
• Figure 3: Distributions of responses to questions about how often people receive credit (a) from specific projects and (b) for specific tasks; and distributions for how satisfied individuals are with (c) how often they receive credit and (d) the mediums through which they receive credit. Dots and error bars represent mean values with one standard deviation on the linearized scales. Colors show how individuals responded to (a), tracking those same individuals across the other questions.
• Figure 4: Responses to "How often did {2 or more people, 1 other person, nobody else} know that you performed those tasks?" as described by response (a) means and (b-d) distributions. Participants who saw the questions in ascending order (0,1,$\geq2$) tended to report that their work is more likely to be invisible (seen by nobody else) and less likely to be partially invisible (seen by 1 other person) than those who saw questions in the descending order ($\geq2$,1,0). In (a) arrows indicate question viewing order, and errorbars show 95% confidence interval of difference between means.
• Figure 5: Correlations between numerical survey variables, ordered by clustering. Colored boxes are statistically significant ($p<0.01$) while gray boxes are not. Being seen by nobody else (I) was correlated with low scores on several other variables. Conversely, being seen by $\geq2$ people (D) was often correlated with high scores on other variables. Being seen by 1 other person yielded mixed results.