Part II — Seeing Organizational Flow · Chapter 10 · 4 min read · First Public Draft

Governance

A Tollbooth on a Road With a Bypass

I built a control that people avoided, which meant I had built something else.

Reduces frictionDecision latency

A note on the word, for anyone dropping in here. This paper uses architecture in a wider sense than the technical one — the deliberate shaping of how an organization is put together, of which system design is one part. A reorganization is an architectural decision. So is a funding model, a team boundary and a reporting line. What a Capability Is, and Is Not in Part I sets out the vocabulary this rests on.

This one is mine, and I would rather tell it than have it inferred.

Years ago we had an architecture board. I was part of setting it up, and the purpose was entirely sound: significant decisions should get a second pair of eyes before they become expensive. Anyone who has watched a quiet choice about a database turn into a four-year constraint will recognize the impulse. It was not empire-building. It was care.

It also had good people in it. That matters, because the usual story about governance is that it fills up with people who do not understand the work, and this one didn’t. The reviews were often genuinely useful. When a decision came to us, it usually left better than it arrived.

The trouble was the when.

People stopped bringing decisions.

Not because they disagreed with being reviewed. Nobody ever said that. They stopped because getting on the agenda, preparing the material, waiting for the slot, and then surviving the conversation cost more than the decision was worth to them.

Which is an entirely rational calculation, and one I would have made myself. A two-week wait and an afternoon of preparation is a fine price for a decision that will shape the next five years. It is an absurd price for a decision that needs making on Thursday and could reasonably go either way.

So they routed around it. Some decisions got made quietly and never surfaced — not hidden exactly, just never escalated, because the effort of escalating was the whole problem. Others did not get made at all, and sat waiting for somebody with the appetite to go through the process, which is a slower and more expensive failure that leaves no trace.

The board had been created to reduce risk. It ended up slowing everything down while seeing less than it had before it existed.

A control that people avoid isn’t a control. It’s a tollbooth on a road with a bypass.

I was also part of taking it apart again.

Which is worth saying plainly, because it is easy to tell the first half of a story like this and let the second half stay implied. The boards and review gates came down once we could see what they actually cost, and dismantling them was more uncomfortable than building them had been. Building a forum makes you look responsible. Removing one makes you look like you have stopped caring about risk, and somebody will say so.

The conclusion I drew first was the wrong one.

For a while I thought the lesson was about scope: we had reviewed too much, and a lighter version would have worked. That is wrong, and it is the mistake I see most often when other people tell me the same story. A lighter tollbooth is still a tollbooth.

The board was not the mistake. Making a destination out of something that needed to happen where the work was — that was the mistake, and it would have been the same mistake whatever the board had been called. Design authority, technical council, review forum. The name changes nothing.

Any function an organization turns into a place decisions must travel to becomes friction, however good the people standing in it are. The moment it is somewhere you go rather than something you have, the queue is the product.

What we had built, in the terms An Interface Nobody Can Find uses, was a queue — and a slow one, in front of a capability that people needed constantly. The bypass was not defiance. It was people finding a shorter route to the same answer, which is what people do when the official route costs more than the thing is worth.

And it generalizes further than architecture, which is the uncomfortable part.

Legal. Security. Procurement. Data governance. Every one of those is a real competence that organizations reliably convert into a location, and every one of them then notices that people are going around them.

The people are not the problem. The bypass exists because the road has a toll on it, and traffic is doing what traffic does.

A reliable sign that a control has stopped working is that everybody says nice things about it and nobody uses it.

Cultivating this: Equipping, Not Reviewing, in Part III.