Figure 1

The penumbra of open source. (a) Geographic distribution of hosts and unique email addresses (in parentheses) in our Penumbra sample. (b) Distributions of emails per host and repositories per host. (c) Distribution of unique emails per repositories. (d) Correlation between repositories and emails per host. We see that the number of repositories and email addresses generally correlate, with some outlying hosts with many more repositories than emails. Academic hosts follow the same general trend, except that they tend to be larger than many non-academic hosts. (inset) Hosts are classified as “academic” if over 50 percent of their email addresses end in .edu or come from a manually identified academic domain