Coding classes in Moray

Roman numerals use seven letters, I, V, X, L, C, D and M, worth 1, 5, 10, 50, 100, 500 and 1,000. Most of the time you add them up, but a smaller value placed before a larger one is subtracted, as in IV for 4. The learner's first program does exactly that and nothing more: it skips any character it does not recognise, then walks along the letters, adding or subtracting. It is short, and it is dangerously polite, because it never refuses an input.

The naive parser turns the transcription into 1695, or 1694 if the 1 is read as an I, even though the listing itself says the lighthouse is dated 1844. The problem is the L before the C, which it treats as "subtract 50". Standard Roman numerals never put L before C, and a strict validator knows that. The learner writes one as a pattern: up to three M's, then a hundreds part, a tens part and a units part, each from a short list of allowed forms. Anything else is rejected with a clear message instead of a confident wrong number.

Then the checks. Converting every number from 1 to 3,999 into Roman numerals and back gives zero mismatches, so the two directions agree. Of all 19,607 possible strings of one to five Roman letters, only 735 are valid numerals, fewer than 4 per cent, yet the naive parser gives a number for every one of them. It even reads IV, IIII and the nonsense IVX as 4. The longest standard numeral below 4,000 is 3,888, MMMDCCCLXXXVIII, fifteen letters.

That is why a Moray teenager should still learn to code in 2026: to build programs that know the difference between an answer and a guess. The longer argument is in why teenagers should still learn to code in 2026.