The Second Time Something Breaks

The same thing failed again. It was fixed six weeks ago.

It gets fixed the same way, by the same person, with the same relief — and nobody asks the only question that matters.

The question

"Why did the fix not hold?"

⚠️ Almost never asked, because the second failure is treated as a fresh event rather than as evidence about the first repair.

Three answers, and they need different responses:

The fix was a workaround. It addressed the symptom. Which was correct at 22:00 and was never followed up in daylight.

The cause is upstream. The thing that breaks is not the thing that is wrong — and repairing it will produce a third failure.

It is wear. Genuinely repeating, and the answer is a schedule rather than a repair.

Why the second time is where the information is

The first failure has no pattern. ⚠️ Nothing to learn from a single event — and reacting to one is frequently over-reacting.

The second gives you an interval. Six weeks, or a night of the week, or after a particular event. The interval is the diagnosis.

And a third is now predictable. ⚠️ Which makes it preventable, and that only becomes true at the second occurrence.

What stops it being noticed

Different people fixed it. ⚠️ Neither knows the other did — the most common reason, and it is a records problem rather than an attention problem.

Nothing was written down the first time. So there is no second time — there are two first times.

The gap was long enough to forget. Six weeks is long enough, and it is the typical interval for exactly the faults that matter.

The small habit

Write down what broke and what was done. ⚠️ Two lines — and the second line is the one people skip.

Check for it before fixing. ⚠️ Ten seconds: has this happened before? If the answer is yes, the response is different.

Mark the second occurrence explicitly. "Second time — needs a real look", so it is not quietly closed like the first.

Escalating properly

At the second occurrence, the response should change:

Stop patching. ⚠️ The third patch is more expensive than the diagnosis, and by then it has failed on a busy night.

Ask who else it affects. Frequently something else is failing for the same reason, and nobody has connected them.

Decide the interval you can tolerate. If it will keep happening, a schedule is a legitimate answer — planned, in a quiet week, rather than at 22:00 on a Saturday.

Three to hold

Faults recurring within three months. ⚠️ The list is short and it is the priority list.

Whether the previous fix was recorded. If not, recurrence is invisible by construction.

Time between occurrences. ⚠️ Shortening is the signal to stop patching — and it is only visible if both were written down.

Where the record has to sit

Recognising a recurrence requires the first occurrence to be findable by whoever handles the second — which is frequently a different person on a different shift. In one person's memory, every failure is the first one.

tasteck keeps tasks and notes on the records staff open during a shift, so what broke last time is visible when it breaks again.

It is not bad luck twice. It is the same fault twice, and the only reason it looks like luck is that nobody wrote down the first one.