On Tuesday, 30 January 2018 at 12:35:21 UTC, jmh530 wrote:
Sure, but I don't think there are enough D github-repositories
to get decent quantitative analysis... Maybe a qualitative
analysis.
Small sample size problem makes me think of Bayesian
analysis...though I suppose there's a bigger problem if D
doesn't have enough github repositories.
Yes, I think you need quite a lot of users to get enough github
repositories. Not sure what the percentage would be, maybe 1 out
50 users or less are active on github?
Of course, the Rust analysis might be a bit off too, because it
is common for enthusiasts to try out and leave young languages.
There might be a significant difference between hobby and
professional users too.
Anyway, I think the "pagerank" transition analysis is
interesting. It provides some insights into dynamics, and also
which transitions don't happen give us something to think about.