I am currently starting my PhD in physics. I like the work and the working environment, thus I chose the project. The onliest thing I am a bit worried about is that my supervisor is known for changing tasks during the project, regardless of deadlines. I already saw that during my master thesis in the same group, where he told me to start and finish another (rather large) lab task two weeks before my final dead line. I managed to do it, but I would prefer not to have such events again.
Thus I proposed the usage of milestones or project tasks for the project. He responded that he would prefer not to use them, due to possible (yet unknown) new discoveries during the research phase, which then could change the topic completely. That argument makes sense for me, too. But is there still a way to set up at least some borders in order to prevent surprises?
The lab task was basically designing, building, testing and running measurements with a new measurement setup for optical properties. It was discussed when I began, but never mentioned later, thus I may have forgot it (I am not without fault here, I have to admit). That lead to achieving the first real measurements ~30 minutes before submission deadline.
Some professors I knew and have seen the thesis afterwards made the remarks that that definitely increased the total scope of the thesis way out of the range of a common thesis.
Furthermore, I discussed the topic with him, and he agreed to make at least a rough draft of expectations, after he understood (as far as I could see) my concerns. So, I'll see...
You can set multiple plans for milestones, such as the main plan A, second alternative B, and the third one C. If you fail in A, just move quickly to alternative B. This is an standard way for doing PhD work, because it is an unknown world for us before doing exactly what should we do to reach the results.
My proposition is to have a overall milestone including details for each step of your project and also have multiple plans(aims) for doing the same project.