Trace dependencies before adding durations.
Review parallel branches, slack, whole-task capacity and descriptive averages. The fictional models do not promise actual project dates.
The count describes six selected tasks under their stated rules. Read the missed item and its explanation before choosing a revision step; it is not a professional level.
What the six tasks cover
The selected tasks use fixed durations, stated dependencies and declared arithmetic. They can reveal a missed maximum, an incorrect sum or an unreachable combination in those models. They do not measure your whole scheduling skill or explain why a real team missed a date. Read the assumptions beside the numbers before using the result to choose practice. No speed score is computed here.
Walk a fresh dependency example
Suppose preparation takes two days. Afterwards, translation takes four and diagram work takes six in parallel. A one-day check waits for both. Under fixed durations, unlimited resources and no gaps, completion is 2+max(4,6)+1=9 days. The sum of all four durations is thirteen, which describes combined task duration, not elapsed completion time when two tasks overlap. Write the waiting condition explicitly.
Slack belongs to a model
In that example the shorter four-day branch can finish two days later without moving the final check, provided no other constraint changes. That allowance disappears or changes when resources, calendars or dependencies differ. A task is not harmless to delay simply because it was once called noncritical. Recompute against the current model and distinguish a late finish from a late start when the question specifies one.
Distinguish capacity from combinations
A ten-unit limit does not mean every integer up to ten can be assembled from whole tasks. With sizes one, four and eight, the available totals are zero, one, four, five, eight, nine, twelve and thirteen. Nine is the largest feasible total. Enumerating this small set avoids assuming an unavailable two-unit fragment. Real plans also need value, dependencies and risks; filling capacity is only the objective supplied in this exercise.
An average describes past records
Add comparable observations and divide by their count to compute a mean. Check equal iteration lengths and consistent unit definitions before interpreting it. The arithmetic mean is not a guarantee for the next iteration, and size-units from different teams are not automatically comparable. A future commitment also depends on available people, the actual work and uncertainty. This practice asks for a descriptive calculation, not a universal velocity policy.
Change one assumption at a time
Redraw the fresh example and change the diagram branch from six days to three. The four-day translation now controls the join and total completion becomes seven days. Explain why the controlling branch changed rather than memorizing the original answer. Then add a shared-resource restriction and identify what information is missing before claiming a new completion date. A correct formula cannot repair an underspecified model.
Keep the learning claim bounded
All explanations and all five guides remain available at zero, partial and perfect counts. Repeating this bank can improve familiarity with its exact networks and numbers, without proving transfer to an unseen project. Use fictional schedules, not private team commitments, in follow-up examples. These author-checked models are not a scheduling certification, independent bilingual validation, a real resource plan or a probability estimate of finishing on time.