Open Ends and Logic Gaps Explained

Why an Open End Breaks the Math, Not Just the Picture

CPM scheduling calculates every activity’s early dates from its predecessors and its late dates from its successors. An activity with no predecessor gets an early start defaulting to the project start (or the calendar’s earliest available date) regardless of what actually has to happen first in reality. An activity with no successor gets a late finish defaulting to the project finish, giving it artificially generous float that has nothing to do with its real position in the work. Either way, the calculated dates and float for that activity -- and the accuracy of the critical path itself -- can no longer be trusted at face value.

Worked Example: One Missing Successor

A six-activity mechanical room build-out sequence, where “Duct Insulation” was added late to the schedule and never got tied to the activity that actually depends on it finishing:

M30’s fire-and-life-safety inspection only requires ductwork to be installed (M10), not insulated -- so on paper, the ceiling can close in right behind the inspection while insulation sits open-ended, contributing nothing to the calculated critical path at all. In reality, most specifications require duct insulation to be complete before the ceiling grid closes it in -- meaning M20 should have been tied as a real predecessor to M40, not left dangling with no successor.

Frequently Asked Questions

Is an open end always a mistake?

Not always -- the very first activity in a project genuinely has no predecessor, and the very last one genuinely has no successor. Those two are expected. The problem is an open end in the MIDDLE of the network, on an activity that should be logically tied to the rest of the schedule but isn't.

Why does DCMA target under 5%, not zero, for the logic check?

A handful of genuine start/finish milestones and administrative activities can legitimately sit outside strict predecessor/successor logic without undermining the schedule as a whole. The 5% threshold accepts that reality while still catching a schedule where open ends are widespread enough to be a real structural problem.

Can an activity have a predecessor but still be effectively disconnected?

Yes -- a predecessor that itself has no path back to the project start, or a chain that loops back on activities already accounted for elsewhere, can leave an activity technically 'linked' while still not meaningfully tied into the schedule's real critical path calculation. Logic gaps aren't only about a completely blank predecessor or successor field.