Online coding and Python classes in Dudley

Table one holds the 2021 census population of each of Dudley's 1,033 output areas, from Nomis. Table two is the ONS lookup saying which built-up area each output area sits in. The first method is the one most beginners write: for each row of the lookup, scan down the population table until the matching code turns up. The second loads the population table into a Python dictionary first, so every lookup is a single step. Both produce identical totals for every built-up area. The only question is speed.

Look at what happens each time the data doubles. The dictionary's time roughly doubles too, which is what "linear" or O(n) means. The nested loops' time roughly quadruples, from 0.76 ms to 3.11 ms to 11.97 ms, because doubling the rows doubles both the number of searches and the length of each search. That is "quadratic", O(n squared), and the comparison counts show it exactly: 534,061 comparisons for 1,033 areas. With the whole borough, the dictionary is about 80 times faster.

At the scale of one borough, 51 milliseconds hardly matters. The trouble comes with growth. Extending the same measured rates to all 188,880 output areas in England and Wales, the nested loops would need about 17.8 billion comparisons, roughly 28 minutes, while the dictionary would need about 377,760 steps, roughly a tenth of a second. Those are estimates, and real timings vary, but the gap between minutes and a blink is exactly what Big O predicts.

Spotting slow code before it runs is a skill Dudley teenagers can use with every AI tool, and a good reason to learn Python in 2026. The longer argument is in why teenagers should still learn to code in 2026.