The pull-based development paradigm is widely adopted by modern open-source software (OSS) projects, enabling a pull request (PR) to pass through multiple validation stages, from PR assignment and continuous integration testing to the actual code review, before eventually being merged into the project or rejected. Due to the distributed collaboration characteristics of open-source projects, PRs often get delayed across the PR stages, including for the first response, slowing down software development.

In this paper, we conduct an exploratory study on the time-to-first-response. By analyzing 111,094 closed pull requests from ten popular OSS projects on GitHub, we find that bots are frequently used to generate the first response in a PR. The timing of those bot-generated first responses is significantly different from the human one. We further conduct an empirical study to understand the characteristics of the bot- and human-generated first responses, including their relationship with the lifetime of a PR. The results of our study indicate that the presence of a bot is an important factor explaining the time-to-first-response in the pull-based development paradigm and, thus, has to be separately analyzed from human responses. Moreover, we also find that projects with a low PR success rate, heavy existing developer team workload, and newly created projects have a higher correlation with a longer waiting time for the first human response to a pull request. These findings are important for newcomers to understand the delays they experience for their pull requests.