Every tool teaches a different model

Across the practice years and then the corporate ones I worked in six accounting systems. CaseWare, ProFile, Sage 50, QuickBooks, Adagio, and later SAP. That was not a plan. It is what happens when you move between firms of different sizes, because the system a business runs is decided by its size, its age, and whoever was there when the decision got made.

The first few times, I experienced this as a tax on changing jobs. New menus, new shortcuts, a fortnight of being slower than everyone, and a great deal of storage spent on keyboard paths that expired within two years.

What it actually was, and I did not see this for years, is the single most useful accident of my training.

Each one has an opinion

Open any of them and within about ten minutes you can tell what the designers believed the work fundamentally is.

One is built as though the central object of accounting is a file of working papers, and the numbers are things that live inside documents. Another is built as though the central object is an entity’s year, and everything hangs off a period. Another assumes the world is a list of customers and a list of bills, and that is such a comfortable assumption for a small business that the software almost disappears. A large one assumes the world is a set of configurable processes with approvals attached, and it is not wrong, and it will take three months of your life.

None of those is a lie. They are all describing the same domain. But each one has decided what the noun is, and once the noun is decided, everything else follows: what is easy, what is a workaround, what is technically possible but socially unthinkable in that shop.

What six of them buys you

Learning several systems for one domain is the fastest way I know to separate the domain from its interfaces, and it is fast because it is subtractive.

Whatever appears in all six is not the tool’s. It is the field’s, or close enough to it that the distinction stops mattering. Whatever appears in one is a vendor’s opinion, no matter how emphatically that vendor’s interface presents it as a rule.

The tell, in a person who has only ever worked in one, is the sentence “you cannot do that”. Sometimes it is true. Usually it is a claim about a screen delivered in the grammar of a claim about accounting, and the person saying it has no way of knowing which they mean, because they have never had a second data point. I said it myself for years. I was not lying. I had confused the shape of a menu with the shape of the work.

The cost, which is not small

Here is where I have to be fair to the people who stayed put.

Being fluent in six systems and expert in none is a real condition and I have had it. Breadth gets you the abstraction and it does not get you the thing that actually saves a month-end, which is the person who has run one system for fifteen years and knows precisely where it lies to you. Which report is subtly wrong under a particular setting. Which posting looks reversible and is not. That knowledge is not transferable, it does not survive a migration, and it is worth more on any given Tuesday than my comparative view of what accounting really is.

There is also a snobbery that comes with the abstraction, and I have caught myself in it. Dismissing the interface as surface, as though the real work were happening underneath somewhere. The interface is not the surface of the job. For the person doing it, eight hours a day, the interface is most of the job.

The thing I still find genuinely surprising is which items turned out to be which. Several things I had assumed were accounting were one company’s decision about a screen. And two or three things I had filed away as arbitrary vendor choices turn out to be in every single one of them, which is a much more interesting fact than it sounds, and I do not have a satisfying account of why.