Online coding and Python classes in Poole

The learner invents a camp of 15 scouts, each with a tracking-test score out of 20: 18, 17, 17, 15, 14, 13, 13, 12, 11, 10, 9, 8, 7, 6 and 4. They add up to 174, so a perfectly fair split into three patrols of five would give each patrol 58. The first attempt simply takes the scouts in list order, five at a time. It produces patrols worth 81, 59 and 34, a gap of 47 between strongest and weakest, the kind of mismatch that would spoil any competition.

Two quick fixes do far better. A snake draft, where the captains pick 1-2-3 then 3-2-1, gives 59, 58 and 57. A greedy rule that puts each scout, strongest first, into the weakest patrol that still has room gives the same gap of 2. But run the very same greedy rule starting from the weakest scout and the gap jumps to 10. The rule did not change; only the order did. That is the most important idea on the page: greedy methods can be excellent or poor depending on the order they see the data.

Can anyone do better than a gap of 2? The learner writes a program that tries every possible way to form three patrols of five: 126,126 splits. It finds that a perfect split, 58, 58 and 58, exists, in fact 342 of them. Then comes the warning. With 20 scouts in four patrols the number of possible splits is 488,864,376, and it keeps exploding as camps grow, which is why real programs use clever shortcuts rather than checking everything. This is the number partitioning problem, one of computer science's famous hard problems.

Testing whether a quick answer is really fair is exactly the sort of thinking that makes coding worth learning for Poole teenagers in 2026. The longer argument is in why teenagers should still learn to code in 2026.