Saturday, 21 August 2021

A is for Algorithm

A recipe for a soup or a method of solving long division are both simple examples of algorithms. An algorithm is basically any mathematical method of solving a human problem. More complex algorithms are computer implementable to speed up decision making. Jonathan Everett (Royal Statistical Society) has become concerned about a recent tendency of UK politicians to blame 'mutant algorithms' for poor policy outcomes (https://www.theguardian.com/commentisfree/2021/aug/17/a-levels-pensions-algorithms-easy-targets-blame-mutant-maths). Everett points out that 'mutant algorithms' have been blamed for unfairly allocating 'A' level (pre-University) grades; badly thought out plans for new housing and even the decision to increase UK pensions inline with specific economic indices (the so-called 'triple lock'). The actual choices, however, are made by people in power. They tell the algorithm writer what they want. There's an old computing adage. 'Garbage in; garbage out'. Computing 'errors' are basically down to the operator. There's another, older adage. 'It's a poor workman who blames his tools'. Sloppy thinking seems to underpin the 'mutant algorithms'.

No comments:

Food For Thought?

The link between global heating and food prices is clearly illustrated in a recent CarbonBrief ( https://www.carbonbrief.org/five-charts-ho...