Numbering From One

The Problem

There is a convention in this industry of starting a serial sequence somewhere other than the beginning. Nobody documents it and everybody understands it: a first customer would rather hold number 0147 than number 0001.

The reasoning is not stupid. A low number signals a small operation, and a small operation signals risk to somebody about to spend money.

The problem is that it makes the number a marketing surface rather than a record, and once a number can be chosen for effect it stops being evidence of anything.

The Idea

Begin at one, and let the number carry only one meaning: this is how many came before.

The decision was easier to make than to keep, because for roughly the first year every number issued was visibly, embarrassingly low, and every one of them went to somebody who could count.

Building The Archive

The scheme itself is deliberately dull. A sequence, a category prefix, and nothing else. No year, no batch, no initials.

Every additional field considered was rejected on the same grounds: it would eventually be wrong. A year is wrong for anything that spans a new year. A batch code is wrong the moment batches are recombined. Initials are wrong when people leave.

A bare sequence has exactly one failure mode — losing count — and that is the only one that can be defended against by writing it down twice.

Challenges

The hardest part was not technical.

It was that the sequence makes gaps visible. A piece withdrawn, a run cut short, a number issued and then voided all leave a hole that anybody can see, and the instinct to close those holes quietly is strong.

The rule that settled it: a gap is explained in the record, never filled. A record that can be tidied is a record that can be edited, and one that can be edited is not evidence.

What Changed

The numbering stopped being a feature and became the constraint everything else is designed around.

Because the number cannot be adjusted afterwards, decisions about what counts as a piece have to be made before it is issued rather than after. That has slowed things down, occasionally by weeks, and it has removed an entire category of argument that used to happen at the end of a project.