Elon Computing Sciences

SHOULD I SHOUT OR SAY THANKS? A COMPUTATIONAL ANALYSIS OF THE DISCOURSE PATTERNS OF LEADERS OF THE LINUX COMMUNITY

Daniel S. Schneider (Dr. Megan Squire) Department of Computing Sciences

A controversy erupted in the Linux software community in July 2013 when one contributor to the project, Sarah Sharp, accused project leader Linus Torvalds of “advocating for physical intimidation and violence… [and]… advocating for verbal abuse” on the Linux Kernel Mailing List (LKML), and said he was “one of the worst offenders when it comes to verbally abusing people and publicly tearing their emotions apart.” To investigate the validity of these allegations, we use computational techniques to describe the patterns of speech and specific words used by Torvalds in his emails including expletives, grade level score, parts of speech, and most common words. Additionally, we use a Naïve Bayes Classifier to compare Torvalds’ text language patterns to that of another leader within the Linux community, Greg Kroah-Hartman, to determine both which words best differentiate Torvalds from this other leader, as well as assess how well automatic classification of these two leaders within the Linux community can be conducted. Finally, since a Code of Conduct was put into place in March 2015 for the LKML in response to these allegations, we study salient features of Torvalds' emails before and after the Code of Conduct to see if any noticeable changes occurred due to the stricter guidelines for speech. These findings may have implications for understanding leadership styles on the LKML as well as learning what impact a Code of Conduct can have on email communication.