START/FINISH and the Critical Path

Assuming you are using some kind of fairly standard Gantt program (I generally use MS Project, but have used Open Workbench from time to time) and assuming you subscribe to the Critical Path Method, I recommend you start thinking about your Gantt's network diagram in regards to START and FINISH activities.

Firstly, these will be dummy activities. Secondly, they are zero duration activities, there only to provide an anchor to the beginning and end of the network. All nodes in the network must have one or more predecessor activities, which can be other activities or START and all nodes in the network must have one or more successor activities, which can also be other activities or FINISH. The exceptions are that START only has successor links and FINISH only has predecessor links.

Why do this? A network with dead end paths is not a completed network. More importantly, if you have dead end paths then the critical path you have calculated (or the software is presenting) may not be the actual path and changes in the project schedule (as the project progresses and is managed) may not reflect changes to the critical path (again, assuming you are seeing the correct one in the first place).

90% of the project plans (gantts) I see in the workplace have this dead end problem and no one seems to use this START/FINISH dummy set. Why? Maybe a lot of PMs feel that if a path ends in the middle of the project it has no effect on the critical path. That might be true, but is far from an absolute.

I think my strongest argument for this is that its just good design. Linking all activities in an unbroken network creates a dynamic plan, where changes are instantly calculated across your entire network.

You could of course call START and FINISH something else if you prefer, such as the name of the final deliverable, but I prefer the generic names for the clarity they offer, and what is my final deliverable? the created deliverable or the post mortem?