Connection circles · Two workshop repair challenges
Loop Repair
Follow the arrows all the way around. Treat more than the symptom, without cutting off a helpful loop.
Do more than clear today's backlog
The old workshop receives only 3 work units per round and handles 4 routinely, yet backlog keeps returning. The previous shift's diagram may be wrong. With a budget of 2, verify the effects, then keep backlog at or below 3 in both final rounds.
Arrows are your hypotheses; nodes show measured values. + means change in the same direction; − means opposite directions. Test one variable at a time before choosing which loop to interrupt. Dashed × links are cut by this intervention. Redrawing an arrow does not change the real workshop.
Test your proposed connections first
The initial diagram is the previous shift's guess. Links may be missing or have the wrong direction. An isolated test increases just one variable by 2, holding other conditions fixed, then measures the next round. Tests start from the snapshot 6 / 4 / 2 / 2 and do not change your main run.
- Connection from Backlog to Reminders; effect ＋
- Connection from Reminders to Backlog; effect ＋
After a run, compare your diagram's backlog prediction with the actual result. You need not finish the diagram before running; a run can expose gaps in the model.
Interventions stay active for six rounds; one-off clearing changes only the start. Runs are unlimited. Each new plan starts from the same initial conditions.
Connections and simulation rules
＋ More backlog creates more reminders.
＋ More reminders create more rush jobs that interrupt the schedule.
＋ Rush jobs interrupt routine work, increasing backlog in the next round.
Each round brings 3 work units; routine capacity is 4. Rush jobs add half their previous-round value to backlog; support removes three quarters of its previous value. Other arrows pass the previous value one-for-one. All nodes update together, so effects are delayed. Values are clamped to 0–30 and rounded to one decimal place.
This simplified model makes relationships observable. These are the simulated workshop's full rules, provided to check your experiments; real workshops need not behave this way.