Key takeaways
- Control means deciding what is fixed, what method runs, and what feedback is accepted.
- Fixed points are planner commitments the solver builds around.
- Staffing pins narrow a run to placement questions only.
- A run you cannot explain afterwards was not under control, whatever the result.
Three decisions that stay with the planner
Scheduling method, fixed points, and the changes allowed per run. Manual, semi-automatic, and fully automatic methods all have a place; the choice is a policy decision about where judgment adds value, and it can differ between periods in the same school.
A solver earns trust by respecting locked commitments and showing issues before changes are applied. Both properties are checkable on any run: did the fixed points survive, and did anything change that was not reviewed?
The control levers
These are what a planner actually pulls during a run, and the reason a result can be explained afterwards.
- Scheduling method: manual, semi-automatic, or fully automatic
- Fixed points locked before the run
- Staffing pins that limit assignment changes
- Feedback shown and reviewed before changes apply
- The reason each blocked period is unusable, visible on the scheduling screen
- A record of what was locked and what was accepted
Working with fixed points
Fixed points work best few and deliberate.
- Name the placement that must not move, and why.
- Check it against known constraints before locking; a fixed point that contradicts availability is a guaranteed conflict.
- Lock it before the run that must respect it.
- Generate around it.
- Read the feedback for conflicts the lock caused.
- Move a fixed point only as an explicit trade-off, not as a workaround.
How to do this in Smootables: pins, edits, and versions
Fall 2026 · Teacher view
Maths 1
Room 12
Workshop
Lab B
Health
Room 4
Maths 1
Room 12
Workshop
Lab B
Health
Room 4
Maths 1
Room 12
Workshop
Lab B
Health
Room 4
On Timetables, pin any lesson to keep it fixed: pinned lessons are not moved when the solver regenerates. Manual control covers dragging a lesson to another slot, swapping two lessons by dragging one onto the other, splitting an occurrence in half, and parking a lesson in the waiting area to place later. Undo and redo cover every step.
The larger lever is versions. Each timetable state is kept on a branch in Versions and changes, so a regeneration is never a leap of faith: if the run went the wrong way, Restore this version brings the earlier grid back, and the Edit Log shows what changed in between.
Feedback is the control surface
Feedback protects control only when it arrives before the change. If the scheduling screen shows why a period is unusable, the planner can choose between a swap, a staffing pin, or a different method, with the trade-off visible. The same information after the fact is an audit, not a decision.
When feedback keeps flagging the same lesson, stop treating it as a placement problem: that is the signal to move to conflict resolution and read the clash properly. Weight-tuning for the preferences the solver optimises is covered in soft constraints.
Questions planners ask about control
Is fully automatic scheduling a loss of control?
Not if the locks and the review step are yours. Control lives in what you fix before the run and what you accept after it, whichever method places the lessons in between.
Do more fixed points mean more control?
No. Each lock removes options from every other lesson, so a grid full of fixed points can become infeasible without any single lock looking wrong. Lock real commitments; leave preferences to the soft constraints.
When should staffing be pinned?
When the question is placement, and who teaches the lesson is settled. Pinning staffing turns a run that could reshuffle assignments into one that only moves lessons in time.