So that you can easily rebuild and re-run each one with difference in build settings, instrumentation, etc, and get comparisons between the baseline and the modified version, without having to check out a different branch each time.

Sometimes it's just quicker to work with two separate working directories than switching branches in one.

`git worktree` is pretty neat btw!

Because my git-fu is weak and I have a fast network connection and a big disk! Doing things the dumb way works fine for me :)

I cannot speak for OP, but I work on the Pyret compiler. Building a bootstrapping compiler is a lot of work and creates many (intentionally) untracked files. Pulling the repo twice allows me to hack on my fork without doing a full-rebuild every time I need to try something or change something on master.