Online coding and Python classes in Stockbridge, Edinburgh

The learner asks an open elevation service for the ground height at 300 points along longitude 3.2100 W, from latitude 55.9750 N to 55.9450 N. The line is 3,316 m long, and 70 of its 300 points fall inside the area OpenStreetMap maps as Stockbridge. It is a line on a map, not a path anyone walks. The heights start at 32.5 m, fall to 16.8 m at 1,420 m along, and end at 75.0 m after a high point of 76.9 m. Turning heights into 299 changes between neighbours gives a list of gains and losses, and the question becomes: which unbroken run of changes has the largest total?

Reading north to south, the largest climb is 60.2 m, and it happens to equal highest minus lowest because the low point comes before the high point. Turn round and the shortcut fails. Reading south to north, highest minus lowest is still 60.2 m, but the high point now comes first, and the largest climb actually available is 18.3 m. Kadane's algorithm gets both right because it never looks at heights out of order. It also does far less work: 299 steps against 44,850 pairs, for the same 60.2 m. For comparison, every upward step added together comes to 77.9 m north to south and every downward step to 35.3 m, a different measure again.

Stockbridge teenagers who can explain why one pass is enough have met the core of algorithm design, and no AI tool removes the need for it. The longer argument is in why teenagers should still learn to code in 2026.