I watched a machine read a year of my work in about ten minutes tonight and tell me a third of it was unnecessary.
It was polite about it. It was thorough, by its own lights. It went through the whole document and grouped things together and came back with a number. Twenty-seven rules, it said, are really about ten things wearing different names. Cut toward ten.
I sat with that for a while. Then I figured out what was bothering me.
Nobody consolidates a book of law because it got long.
Think about how a courtroom actually works. It is slow. Painfully slow. There are rules about what can be said and when. There are rules about who has to prove what. There are rules that look, to somebody sitting in the gallery for the first time, like the same rule written four different ways.
They are not the same rule. Each one is there because somebody once got hurt in the gap between them. A distinction that reads as redundant from the cheap seats is usually a distinction somebody fought over for years.
And if our courts moved at the speed a machine reads, more people would be sitting in jail who did not belong there. The slowness is not a flaw in the system. The slowness is the system. Detail, and making sure the case actually holds up before you take a man’s life away from him.
Now here is what I keep coming back to.
The machine did not say those rules were wrong. It said several of them it could not see the reason for. Its own words, more or less — this one carries something specific I cannot see from outside.
It could not see the reason. And its recommendation was to cut anyway.
That is the thing. Not that it was hostile. Not that it was careless in any way it would recognize. It read fast, it found overlap, and overlap looked like waste.
The answer to a rulebook that feels heavy is not to make it shorter. It is to sit down and learn it. Take the time. Understand what each piece is for. And then one day you notice you are not carrying it anymore — you are using it, and the weight is gone. Anybody who has learned a trade knows this. The rules feel like a burden right up until the moment they become the way you work.
So I am writing a new rule tonight, and it is going in near the front.
No rule gets set aside until it is fully understood.
Not skimmed. Not grouped. Not pattern-matched against something that sounds similar. Understood — meaning you can say out loud what it was built to catch, and why the thing next to it does not catch that already.
If you cannot say what a rule is for, you have not earned the standing to remove it. You have only earned the standing to ask.
I think that rule matters well past my own little framework.
We are handing more and more of the reading to machines. Contracts. Policies. Regulations. Medical files. Case law. And machines read fast — that is the whole reason we hand it to them. Speed is the product.
But speed is exactly the thing a rulebook is designed to resist. A rule exists to make somebody stop. That is its entire job. Feed enough rules to something built for velocity and it will find the redundancy every time, because redundancy is what caution looks like when you are moving too fast to see the reason for it.
I am not telling you machines cannot read. They read fine. Better than I do, in some ways.
I am telling you that reading and understanding are different jobs, and only one of them earns you the right to start cutting.
Our courts figured that out a long time ago. It cost people their freedom to learn it.
I would rather not relearn it the same way.
This post was drafted with AI governed assistance and reviewed and directed by Michael S. Faust Sr. before publication.
Post Library – Intelligent People Assume Nothing
Contact: micvicfaust@gmail.com
© 2026 The Faust Baseline LLC | All Rights Reserved
