Use Frequent Branches to Tell a Story and Simplify Code Reviews

My team was discussing branching today, and how doing all development in feature branches makes sense when you have a decent version control system.

We had 4 different tasks going on at once, on different parts of the system, yet all of them in the same repo. While my task was branched, the other three were being edited in place in a develop branch. After the code is put into a serious test environment (cloud app and all that), we see instability. But can we tell exactly what caused it? Not really, because we can't easily take out each task.

Had everyone else been working on a branch, we could easily have tested feature by feature, and see what actually introduced the instability with a few simple commands. Instead, we have a bunch of little commits that are mingled with each other, and the best we can do is try to imagine how the branches should have looked like by username.