Films from the Floor

Fourteen times a gauge stuck. Fourteen times somebody fixed it. That is why nobody knew. Buncefield, England. Five past three in the morning.

The gauge stopped reading. The tank kept filling. Three alarms sat above it. None fired. An alarm knows when a number crosses a line, not when it stops.

One screen ran every tank, with no backup. Tank nine twelve sat behind four other windows. One safeguard was left, a high level switch. It was tested regularly, on schedule. The test needed a padlock put back.

Nobody there knew the padlock mattered. So the testing is what left it dead. By six oh one, a quarter of a million litres had escaped. Forty three people were injured. Nobody was killed.

Here is the part that should stop you. That site had a defect log. The sticking never reached it. Every fix was quick, and a quick fix leaves no trace. The contractor was called out again and again.

And never once asked why they kept being called. The emergency stop on their screen was never wired in. Investigators named that a root cause. What is your plant fixing so fast nobody counts it?

Your plant does not forget what happened. It just cannot reach it in time to matter. You come back from days off to a problem you have not seen. Nobody on now was there when it started. So you call whoever was on before.

Sometimes that works, sometimes he is asleep, and you are back to guessing. The record exists — half on paper, half in a spreadsheet, on somebody else's computer. Troubleshooting is elimination. If you cannot retrieve what was ruled out, nothing is ever ruled out. Why does nobody just go and look?

Finding out costs more than guessing, and the plant set those prices. Four crews, the same search, from the same first branch, every time. None of this is new, and none of it is a secret. So why does every plant still have it?

Your crew does not repeat work because nobody wrote it down. They repeat it because nobody believes it. An operator comes back from days off and finds his job was done again while he was away. The entry said checked. It did not say who, how carefully, or what they actually found.

For the next crew, an unverifiable claim is worth nothing, so checking again is the correct call. Why do logs say so little? They were written to prove work happened, not to hand over what was seen. Now the same job written as what somebody saw. The state, the reading, the number.

Nobody redoes that one. You are not trusting a person, you are reading evidence. Tonight, somebody will read what you wrote. Will they act on it, or do it again?

Two crews ran the same tests on the same fault, a week apart, and neither knew. The record existed — half on paper, half in a spreadsheet, on somebody else's computer. It says checked — not who, not how carefully, not what they found. Write what you saw instead — you will still be told to do it again. Why does the trend outrank the person who was there?

Instructions travel on a channel. What you saw travels on nothing. Do that three times and anyone sensible stops writing carefully. You cannot reach it, you cannot believe it, and it never changes anything. Nobody in that chain is wrong.

Each decides correctly on whatever reached them. Nobody is assigned to the space between two shifts. So who at your plant owns it?

A tool built on the floor is rarely refused for being crude. It is refused for belonging to one person. This is the story of one watch and its principles. How it was built, how it was judged, how it was rewarded. In 1714 Parliament put twenty thousand pounds on a method useful at sea.

Newton told the committee why no watch had yet done it. The motion of a ship, the variation of heat and cold. The reasons sat on the deck, not the sky. John Harrison answered them in wood, at Barrow-on-Humber, by 1735. Two interconnected swinging balances left it unaffected by the motion of a ship.

It compensates for temperature and runs without any lubrication. By his own account, his son carried a later watch to Jamaica. Three appointed men reported; the Commissioners declared it outside the limits. On Barbados the Commissioners certified it within the nearest limits. That settled the watch, not the question.

In 1765 the Board resolved he had fully discovered and explained his principles. A certificate would follow, once he delivered the watch to Kendal. Kendal had undertaken to construct a timekeeper upon the same principles. One clock was not a method until other hands could repeat it. Cook wrote of that copy, so good a guide as Mr Kendall's watch.

In 1773 Parliament voted a further reward for the invention and principles. Every plant runs a sheet like it, built by whoever stands closest. It stops there, because only one person can run it. If you judge one, ask who else could repeat it. If you built it, answer that before anyone asks.

What works on your floor that has never been copied?