What is a near-critical path?
A near-critical path is a chain of dependent tasks whose total slack is small relative to the uncertainty of the estimates in it: close enough to the critical path that ordinary estimate error can promote it to critical. A chain two days shorter than the critical path, containing tasks that routinely vary by three, is not "safe by two days". It is the project's next critical path, waiting on a coin flip.
The probabilistic definition#
The crisp version requires treating durations as distributions rather than numbers. Simulate the project many times (Monte Carlo forecasting), and in each simulated future, find which chain actually decided the finish date. Each chain then has a criticality rate: the fraction of futures in which it was the binding constraint.
| Chain | Paper slack | Criticality rate | Honest reading |
|---|---|---|---|
| A | 0 days | 55% | The headline critical path, but only a 55% favourite |
| B | 2 days | 30% | Near-critical: one bad week from deciding the project |
| C | 9 days | 2% | Genuinely slack; stop watching it |
"Near-critical" is the middle row: meaningful criticality rate, no headline status. A useful rule of thumb without simulation: any chain whose slack is less than the combined spread of its tasks deserves near-critical treatment.
Why near-critical chains cause the surprises#
Attention follows the labelled critical path: it gets the standing agenda item, the daily questions, the best people. Near-critical chains get none of that, by definition, which produces a repeatable failure pattern: the project slips because of a chain nobody was watching, and the retrospective says "but that work had slack". It had slack on paper; the paper assumed the estimates were exact.
This is why the binary critical/not-critical label is the most dangerous simplification in classical scheduling. Risk does not live on one path; it is distributed across the contenders in proportion to their criticality rates, and managing it means watching the top few chains, not the top one. The full essay covers the management side, including the inverse question (which task, if accelerated, actually moves the finish date, weighted by how often its chain is in charge).
In practice#
Topolog computes criticality per simulated future as a by-product of the same Monte Carlo runs that produce the completion forecast, so the critical-path view shows the modal path highlighted and the near-critical contenders ranked by how often they seize the lead, continuously, with no extra modelling work. The critical path docs show the surface.