Think something like he started the project at some point, never planning for it to become this big. It was probably some code written in a random directory, with no regard for whether test fragments are around, whether it contains test data that maybe shouldn't be public, different pieces that are closely tied to a server config that itself is very specific to what he already had running and would need some documentation to be useful etc. pp.
For something like this, keeping things private has allowed Troy to work out the best way to do things with complete autonomy. That's really useful if you have the time and resources to get to market on your own.
Now it's stable and best security practices are nailed down, he can open it up to a bit of scrutiny and feature bloom.
You'd do it to prevent feedback.