by Serguey Shinder
There was a misspelling on our invoice screen. Recieve, with the letters the wrong way round, on a page every customer saw once a month. It was reported in January by a customer with an eye for it, and it was still there in June.
Nobody had refused to fix it. Nobody had argued that it did not matter. The fix was one word, and the reason it survived five months is that to change that word you had to be able to see the page, and to see the page you had to run the invoicing service, the account service, the service that issued tokens, a message broker, and a front end build, on top of a restored copy of an anonymised database that took six minutes to come up on a good day. I timed it properly one morning, because I had begun to suspect the number mattered. Nineteen minutes from opening the laptop to seeing an invoice, and about seven of those minutes needed me, because a container had gone stale overnight and a development certificate had expired and each of those is its own small errand.
Thirty seconds of work sat behind twenty minutes of approach. So it never happened, and neither did the sixty other items in our backlog tagged small, the oldest of which was fourteen months old. We had not decided to ignore them. The cost of starting had decided for us, quietly, item by item, and every individual decision to do something else first was reasonable.
Two of us spent a fortnight making the front end run against a recorded set of responses, so that a copy change or a layout change is now about forty seconds from a cold laptop to the thing on screen. In the month after that we closed thirty one of the sixty. The same people, the same priorities, the same amount of time in the day.
I had believed, without ever examining it, that what a team works on is decided by its priorities. Mostly it is decided by what is cheap to begin. The cost of starting is a policy, and it is one you are enforcing whether or not anybody has written it down or agreed to it, and it is enforced most strictly on exactly the work that is too small to justify a fight.
So when something obviously worth doing keeps not getting done, I have stopped asking why nobody wants it. I take a stopwatch and I time the approach, from a cold machine to the first change visible on screen, and I write the number on the wall. The number has explained it every single time I have bothered to measure.
– Serguey Asael Shinder
Leave a Reply