Question Every Requirement

What it is

A requirement in an organisation almost never arrives labelled with its origin. It arrives as a fact about how things are done, which is why it survives. The step asks for two things: the name of whoever imposed it, and the condition it was meant to handle. Both are usually recoverable, and both are usually smaller than the rule they produced.

The lineage is old and the book does not name it. This vault traces the move through Aristotle's Physics and Metaphysics and, in its modern form, Descartes's Meditations on First Philosophy of 1641. The hub calls that omission the strangest one in the book, given how closely first-principles reasoning is associated with the person the book credits as the Algorithm's architect. The step is genuinely useful; the idea that it is new is not supported.

It sits first in the sequence for a structural reason. Everything downstream, deleting, simplifying, accelerating and automating, acts on the set of requirements this step has already interrogated. Run it late and you have optimised things that should have been removed.

In effect

The worked example is mundane and typical. A retailer requires two signatures for every refund. Asked where that came from, the finance team traces it to a single fraud case years ago involving high-value electronics. The requirement shrinks to refunds above a threshold, and the queue at the till disappears. Nothing was discovered. Somebody simply asked.

One caution is worth carrying, and it is the thing that separates this from a slogan. Questioning every requirement is only useful if somebody in the room has the authority to remove one. Without that, the exercise is a ritual, and a demoralising one, because it produces a list of things everyone now knows are pointless and nobody can change.

The force it works against is status quo bias, which is well studied and is not the same thing as this step. The step is a procedure. The bias is why the procedure is needed.

What it does not say

It does not say every requirement is wrong. It says every requirement is a claim with an author, and that the author can be found.

It does not rest on research. There is no study of the Algorithm or of this step, and the language to use is that McNeill argues it rather than that research shows it.

It does not originate with this book. The first-principles lineage runs back through Descartes to Aristotle, and the book names none of it.

It does not work without authority. A team that can ask but cannot remove gets the cost of the exercise and none of the benefit.

---