Coding and AI classes in Widnes

The Census divides Widnes into 193 output areas, the smallest units it publishes. From the Nomis API the learner downloads five counts for each one: residents, households, one-person households, households with no car or van, and households living in purpose-built flats. For each count the program does the same three things. It sorts the areas from largest to smallest, takes the top 20%, which is 39 areas, and works out what share of the town's total those 39 hold. If the 80/20 rule were a law, every answer would be close to 80%.

Only the last row looks anything like 80/20. Purpose-built flats are built in blocks, so they pile up in a few places: 36.8% of the areas have none at all, the top 5% of areas alone hold 29.2% of them, and four fifths of all the flats sit in the top 23.8% of areas. Residents are the opposite case, and for a reason worth noticing. Output areas were drawn to contain roughly similar numbers of people, so the unit itself stops residents from concentrating; reaching four fifths of them takes 74.6% of the areas. (The 59,938 is our own total across the 193 areas and differs by three from the ONS built-up area figure, which is rounded.) The rule is a description of lumpy things, and the learner now has a test for lumpiness.

For a Widnes teenager, being able to check a popular claim in ten lines of Python is a strong reason to keep learning to code while AI writes more of the routine parts. The longer argument is in why teenagers should still learn to code in 2026.