Coding and AI classes in Harlow

The learner starts with the three facts from the Nobel Foundation. 1960s fibre: 1 percent of the light left after 20 metres. Kao and Hockham's goal: 1 percent left after 1 kilometre. Today: 95 percent left after each kilometre. The first attempt at today's fibre goes wrong in a very common way. "95 percent left means 5 percent lost per kilometre, so after 20 kilometres nothing is left." That is linear thinking, and it is false. Each kilometre keeps 95 percent of whatever arrived, so after two kilometres 0.95 times 0.95 remains, and the light never quite reaches zero.

Engineers avoid the trap by using decibels. The loss in decibels is minus ten times the logarithm of the fraction left, so 1 percent left is a loss of 20 decibels and 95 percent left is about 0.223. The great advantage is that decibels add: a 100 kilometre fibre at 0.223 decibels per kilometre loses 22.3 decibels, which is about 0.59 percent of the light arriving. Adding up losses like this for a whole link, fibre plus joints plus connectors, is called a link budget. It also shows how extreme the 1960s glass was: two 20 metre lengths in a row leave 1 percent of 1 percent, just 0.01 percent.

The final check ties the numbers back to the press release, which says purest glass could carry signals "over 100 kilometers, compared to only 20 meters for the fibers available in the 1960s." Using the same rule, 1 percent left, the learner's program gives 20 metres for the old glass and about 90 kilometres for today's, the same scale as the Nobel Foundation's words. A figure calculated independently that agrees with a published statement is worth far more than one simply copied, and that habit of checking is exactly what we teach alongside AI tools.

Harlow teenagers who can test an AI's sums against their own will stay in charge of the tools they use, which is why learning to code still pays off in 2026. The longer argument is in why teenagers should still learn to code in 2026.