Coding classes in Mid Ulster

The learner gives each stage two numbers: how many machines it has, from the plan, and how long one machine takes per piece, which we have to invent because the record does not say. A stage's hourly capacity is machines divided by time per piece. The drying loft works differently: it holds 500 pieces for, say, four days, so its capacity is how often it can be filled and emptied. Over an invented working year of 2,500 hours, Python prints each stage's capacity, and then the key line: the whole mill can manage only the smallest of them.

With these guesses the beetling engines hold everything back, and 6,250 pieces a year falls short of the plan's seven to eight thousand. Now the program can answer design questions. To reach 7,500, the beetling engines would need to manage about 3 pieces an hour between them, around 0.67 hours per piece each. Add a third engine and the mill jumps to 9,375. Add a fourth and it stops at 10,000, because now the rubbing boards are slowest: the bottleneck has moved. Add a fifth pair of boards too and three stages tie at 12,500. Improving anything except the slowest stage does nothing at all.

Two slips are worth testing. Add up the stage capacities and you get a nonsense 60,000 a year; average them and you get 15,000. Neither describes a line in which every piece passes every stage. Learners write a check that the mill's answer can never exceed any single stage. They also see how much rests on guessed times: nudge the beetling time from 0.8 to 0.65 hours and the plan's figure suddenly looks reasonable, which is why a model's assumptions belong right beside its answer.

Demanding to see a model's assumptions is a strong reason for a Mid Ulster teenager to keep learning to code in 2026. The longer argument is in why teenagers should still learn to code in 2026.