Ask a management team whether they make architectural decisions and they’ll say no. Architecture is technical. It happens somewhere further down, among people with a different vocabulary.

Then they spend the morning deciding which departments exist, who reports to whom, and where one team’s responsibility ends and another’s begins.

That is the architecture. Not a factor shaping it — the thing itself. Split one capability across two departments and you have guaranteed a handover, a negotiation and a delay, permanently, in everything that capability touches. The decision took four minutes. It will outlive everyone who made it.

Reorganizations are architecture, performed by people who were told architecture was somebody else’s specialism.

Which means the most consequential design work in most organizations is done by the group least likely to review it as design work. Nobody sketches the alternative. Nobody asks what the new shape makes slow. The boxes move, the announcement goes out, and the consequences show up eighteen months later as a culture problem.

Once you see it in the org chart, you start noticing it in the other commitments leadership makes, and budgeting is the clearest second case.

Look at what a budget is actually attached to. Almost always a project or a system — something with a name, a scope and an end date. Which is entirely reasonable: those are the things that can be scoped, costed and closed, and finance quite properly needs something it can close.

But a project ends and a system gets replaced. The capability underneath them doesn’t. Claims handling outlives every claims system; onboarding outlives every onboarding project. So the money attaches to the temporary thing, and the enduring thing — the one the organization actually needs to keep getting better at — has no line of its own and no one accountable for it across the years.

That has a second effect worth noticing. When funding is granted to a project, the project becomes the thing that gets reported on: delivered, on time, within budget. All three can be true while the capability is no better than it was. Nobody misrepresented anything. The question of what the investment was meant to produce simply wasn’t the question the process was built to ask, because it was answered once at allocation — the one moment when no information exists yet.

Add the annual rhythm and you get the mismatch from The Cost of Value Never Created in its purest form: the work needs a decision this week, and the funding mechanism can produce one in April.

None of this is anyone’s failure. It’s a set of arrangements that made sense when what you built stayed built — and that quietly stopped matching how the work actually behaves.

And then there is the assumption underneath both: that a manager is where decisions go. Not someone who creates the conditions for decisions to be made well, but the person who personally makes them. It’s rarely stated, it’s frequently rewarded — we still measure a leader’s standing partly by how many people report to them — and it produces a specific, predictable trap.

If every decision must travel upward, everyone above needs enough information to decide. Getting the right information to the right level takes people whose job is moving and summarizing it, so you add a layer. But each layer sits further from the detail than the one below, so the summaries thin out while the decisions stay just as consequential. Add enough layers and you arrive somewhere nobody intended: the people with the authority to decide have the least direct knowledge of what they’re deciding, and the people with the knowledge have none of the authority.

Nobody designs that. It’s simply what you get by default when decisions travel upward and information has to chase them.

That trap has always had a quiet exception built into it: the summarizing itself needed a person, because nobody else had the time to do it. That exception isn’t permanent. A system can already read what ten teams produced last week and hand a decision-maker the three things that actually matter, without two days spent assembling the deck first.

I want to be careful here, because this isn’t an argument that the layer was pointless. Reading the shape of a room, translating between how a team talks and how leadership needs to hear it, protecting people from noise they don’t need — most of what a good manager in that position does was never really about moving information. It was judgement, exercised while moving it. That part doesn’t go anywhere.

What does thin out is the part of the role that was purely relay — summarizing because summarizing was the only way the information could travel at all. Almost nobody was hired for the relaying. Most people in that seat were put there for their judgement, and the relaying was simply what the structure demanded of them alongside it, quietly, until it looked like most of the job.

So if this describes a seat you’re in, the question worth asking isn’t whether you’re replaceable. It’s whether the organization around you has ever actually seen the judgement you carry, separate from the reports that judgement got wrapped in to travel upward. Most organizations haven’t, because the reports were the only part anyone could see. That’s worth raising on its own, regardless of what technology does next — not because your position is at risk, but because the best part of what you do has been invisible for reasons that had nothing to do with its value.

Every one of these is a design decision that was never recognized as one — which is the whole argument of this chapter. What to do about it is a matter of practice, and Part III takes it up.