The HN thread began on Oct 16 at 20:22 Z. If you visited that github page instantly and took that screenshot instantly, even accounting for 21 hours due to rounding, the commit in that screenshot had to be after Oct 15 23:22.
The repo as I and many other people cloned it has the first commit ("first commit", not "initial commit") at Oct 12 23:20 Z, and the "done" commit at Oct 15 19:37 Z.
A likely explanation is that pixelmelt squashed both commits at or after they put up the blog post, but didn't force-push the rewritten history to github until it hit HN and blew up.
The repo as I and many other people cloned it has the first commit ("first commit", not "initial commit") at Oct 12 23:20 Z, and the "done" commit at Oct 15 19:37 Z.
A likely explanation is that pixelmelt squashed both commits at or after they put up the blog post, but didn't force-push the rewritten history to github until it hit HN and blew up.