Coding classes in South Lanarkshire

The school is invented. One teacher knows a lesson and 500 pupils need it. Time passes in rounds, and in each round anyone who already knows the lesson may teach a small number of new people. Every time the lesson passes from one person to another, there is a small chance it gets garbled; the model says each hand-off keeps it intact 98 times in 100. The learner codes several schemes in Python and counts two things: how many rounds until everyone knows, and how many hand-offs separate the last pupil from the teacher.

A lone teacher is faithful but slow: 50 rounds. Letting everyone who has learned teach just one other person is dramatic at first, because the number who know doubles every round, and 500 pupils are reached in only 9 rounds. But the chains get long. The last pupils hear the lesson eight hand-offs away from the teacher, and by then the chance it has survived intact has fallen to 85 per cent. A monitor tree sits in between: 6 rounds and never more than two hand-offs, so 96 per cent arrives unchanged. The fastest scheme here, everyone teaching ten, needs 3 rounds with chains of 3.

The pattern is the key idea of broadcast trees. Rounds fall like a logarithm as each person teaches more, while every extra layer of the tree costs accuracy. The learner checks the arithmetic with a simulation, running the monitor scheme 200,000 times and getting 95.99 per cent against the formula's 96.04. Real computer networks, from software updates to the way a message spreads in a group chat, make exactly this trade-off between how many copies each node sends and how many hops a message travels.

Weighing speed against accuracy is exactly why a South Lanarkshire teenager should still learn to code in 2026. The longer argument is in why teenagers should still learn to code in 2026.