Randomness and safety

I have recently become intrigued by the phenomenon of cumulative error – error building gradually until it suddenly becomes quite gross. It's got me thinking.

By the mid-20th century, quality assurance (QA) had taken over from quality control. The latter involved armies of inspectors at the end of the manufacturing process looking for faults and removing faulty products from the line before they could reach the customer. It was effective but very labour intensive and wasteful.

The inventor and guru of QA, William Edwards Deming, believed that if you could identify the upstream error(s) which had contributed to any fault, you could adjust the detailed specification of the manufacturing process so that such errors would gradually be reduced – and you could be gradually be assured that virtually all products coming off the line would be fault-free. You could then dispense with the armies of quality inspectors and reduce wastage, reduce costs and thus beat your competitors. Deming’s ideas contributed to the somewhat simpler ‘Plan-Do-Check-Act’ cycle that today is drummed into every manager (and not just in manufacturing) and which also underlies things like certificated health and safety and environmental management systems.

It was said that Deming hit upon the idea of QA during his time as a lodger who liked his toast and boiled egg to be ‘just so’. When it wasn’t, he asked his landlady to write down exactly what she did (freshness and thickness of bread, toasting time, size of egg, length of time in boiling water), noting down any minor variations in all the variables so that eventually he could be assured that his daily breakfast would always be consistently perfect. 

That was the theory. And it took off in a big way, particularly in post-war Japanese manufacturing.

What happened in practice however, particularly when everyone went QA mad from the 1980s onwards, was that armies of consultants emerged writing massively thick QA manuals for companies (for equally-massive fees), full of process detail which not even the cleverest manager could read and understand, let alone the folk on the line who were actually building and delivering the product.

The process had to be humanised by having things like ‘quality circles’ or just pre-shift ‘huddles’ in which you attempted to involve the front-line workers and tap into their detailed knowledge of what was actually happening out on the line, as well as further upstream.

What intrigues me is that it feels as if QA is an attempt to defeat randomness and things like cumulative error taking over – in its own way a sort of morality play about the human condition. But a little humility and caution are required.

Von Moltke the Elder said (and I paraphrase): “No plan survives first contact with the enemy”. And Rabbie Burns said: “The best laid plans of mice and men gang aft agley”.

The point being that the randomness of the future cannot be predicted in its entirety and in absolute detail, and thus our attempts to suppress its effects can only ever be partially successful.

That said, the fact that in so many areas of safety we are able to actually improve failure rates and cut casualties, suggests that the great God of Randomness does not have it all his own way. He can indeed be kept at bay most of the time but we must never be tempted to believe we have finally got the better of him.

Thoughts on a post card, please.

 

Roger Bibbings, partnership consultant