Coding classes in Oxfordshire

The learner writes a class for a quantity: a number plus the powers of kilograms, metres and seconds that describe it. A joule is kilograms times metres squared per second squared; a watt is the same divided by one more second. Dividing a quantity by a time subtracts one from the power of seconds automatically, so energy divided by time comes out as power without anyone remembering to say so. Adding two quantities checks first that their units match, and if they do not, the program stops with an error. That one rule catches a whole family of mistakes before they ever produce a number.

The third row shows why peak and average are different things. If JET's 1997 run had produced its peak power of 16.1 MW the whole time, 21.7 MJ would have taken 1.35 seconds; since power rises and falls during a pulse, the real run must have been longer. The fourth row is the one that makes people sit up. 3.45 times ten to the fourteen joules per kilogram is a number so large it needs comparison: the same 69 megajoules would run a 3 kilowatt kettle, our example, for about 6.4 hours, and it came from fuel weighing as much as an insect.

There is one more lesson hidden in the sources. GOV.UK's 2023 page gives the 1997 record as 16.1 MW and 21.7 megajoules; a 2022 page from the same agency gives 16 megawatts and 22.5 megajoules. The energy figures differ by 3.7 per cent. We do not know which measurement or definition each page used, so we print both and say where each came from. A program can check units; it cannot tell you which of two official sources to believe, and a careful learner notices when they disagree.

So an Oxfordshire teenager should learn to program in 2026 to be the person who catches the wrong unit in a confident paragraph, whoever or whatever wrote it. The longer argument is in why teenagers should still learn to code in 2026.