This one is about the others. There are three steps, each with its own post, and this is the high-level map of how they fit together plus the two properties that make the arrangement work at all. It deliberately does not go into the detail of any one step. It is the post about how to read the other three.

The Three Steps

Why: the Primary Directive. The standing order that everything downstream is ranked against. It comes in two nested forms: a specific one that is personal and particular to your situation, and a general one that is structural and shared (Explore, Play, Joy). Answered first.

Who, What, When, Where: the Contextual Core. The four questions that locate you against the directive. Four separate questions, four separate ways to be lost, bounded by the directive so that the search actually terminates. Answered second.

How: the Action Phase. The execution, the driving force, the part where the work gets done and the evidence gets produced. Last, and smaller than it usually is, because most of what makes execution hard was decided in the two steps above it.

Then it cycles. The evidence How produced is what the directive gets audited against, and the audit issues the next directive.

The ordering is the whole practical payload: Why → Who/What/When/Where → How → Why. It runs against the default, which is to start at How, because How is where the typing is and typing feels like progress. Everything else here is an argument for that order, or an explanation of why the order is harder to keep straight than it looks.

The First Meta Property: It Is Self-Referencing

The three steps are not three different kinds of activity. They are one shape, and the shape shows up again inside each of its own parts.

Take the Contextual Core, which looks like a single step from the outside. Start actually doing it and it unfolds into the same three: you need a directive for the search (what am I trying to locate, and when is it enough), you need to locate yourself within the search (who knows this, where is it written, when do I need it by), and then you need to execute the search. Why, then the four Ws, then How, one level down, nested inside a step that was itself the four Ws.

The same is true of the Action Phase, which has its own directive (what does done mean here), its own locating (what does this code touch), and its own execution. And it is true going up: a career is one cycle whose How is a decade of projects.

Two things follow, and they are the reason this property is worth naming instead of just admiring.

The shape is what recurses, not the content. Nothing about a code review resembles a career. The three-beat shape of both is identical. When this framework seems to claim that everything is everything, that is the claim being made badly: only the structure repeats.

Levels connect through a single handoff. Each level's directive is the output of the audit one level up. Life-level Why issues the career-level directive; career-level Why issues the project-level directive; and so on down. Which means the layer "above" the loop is not a different kind of thing from the loop. It is the same loop running slower, and it only looks stationary because its cycle time is measured in years.

The recursion has to stop somewhere or it is circular rather than nested. It stops at the general directive: Explore, Play, Joy are terminal, which is to say nothing underneath them explains them. That is the base case, and it is what makes this a well-founded structure instead of a set of ideas justifying each other in a ring.

The Second Meta Property: It Is a Loop

The steps are not a pipeline with an end. The audit at the close of one cycle is the directive at the open of the next, which is the same event described from two sides. That single identity is what makes the whole thing compound rather than merely repeat.

Information moves in both directions, and both directions are load-bearing:

  • Downward: directives. Each level hands the level below it a rule for what counts as better. A structure with only this direction is a bureaucracy: orders with no evidence that they are working.
  • Upward: evidence. Executing reveals whether the directive was any good, and that evidence is the raw material of the slower audit above. A structure with only this direction is drift: lots of learning, no direction to apply it to.

Both running at once is the compounding. Each cycle improves what you do, and also improves what you are optimizing for.

This is also the honest answer to a reasonable objection: if Why comes first, how do you ever answer it the very first time, with no evidence? You do not, not well. The first directive is a guess. The loop is what upgrades a guess into a directive, and expecting the first pass to be correct is a misreading of what kind of structure this is.

The Consequence: Half the Vocabulary Is Relative

Here is the part that trips people up, including me, and the reason this post exists rather than being a paragraph in another one.

Once the structure is both self-referencing and a loop, a lot of the words used to describe it stop being properties of things and become positions relative to wherever you are currently standing. "Before" is not a property of the search. It is where the search sits relative to the act you happen to be considering. Change which act you are considering and the labels move, even though nothing about the structure changed.

The clearest case: the four Ws are described as the "before" step, because they come before the action. But when you sit down and actually do the locating work, that work is now the current activity. It has its own before (its directive) and its own after (did I find what I needed). The step that was "before" is now the "during." Nothing moved except the frame.

So the terms split cleanly into two kinds, and keeping them apart resolves most of the confusion:

Fixed terms. These do not move. They mean the same thing at every level and from every vantage point.

TermWhat it is
Why / Primary DirectiveThe rule that ranks; always answered first
Who, What, When, WhereThe four locating questions; always the middle
HowThe execution; always the last step of a cycle
Search, Drive, RenewThe three forces, in that order
Mastery, Autonomy, PurposeThe motivations attached to those forces
Explore, Play, JoyThe general directive; terminal, same at all levels
The order itselfWhy → four Ws → How → Why

Relative terms. These are coordinates, not properties. Each one is only meaningful once you say which act and which level you are standing in.

TermRelative to whatWhat moves it
Before / during / afterThe act under discussionChanging which act you mean
CurrentWhatever you are doing nowStarting the next step
Above / belowThe level under discussionZooming out to a slower cycle, or in
Faster / slowerThe neighbouring levelWhich two levels you are comparing
First / lastOne cycleWhere you cut the loop (it has no true end)
The whyAltitudeWhether you mean the whole layer or one step

That last row deserves its own sentence, because it is the single worst ambiguity in this whole series. "The WHY" sometimes means the entire layer above the operational loop, all three forces of it, and sometimes means one specific step: the directive. Both usages are in the posts. Neither is wrong. They are different altitudes, and a self-similar structure will do this to your vocabulary every time if you let it.

The practical rule is short: when using a relative term, say the frame. Not "the search comes before," but "the search comes before the action." Not "the why sits above," but "the why sits above this cycle." It is one extra clause and it removes the entire class of confusion, because the ambiguity was never in the idea, only in the sentence.

How to Read the Other Three

So the series reads like this. Start with the directive, because without it the next step has no edges. Then the four Ws, because without them the action is executing against knowable unknowns. Then the action, which is smaller and better posed than it would otherwise have been. Then back to the directive, carrying evidence.

And read all three knowing the two properties: the same shape is inside each of them, and the end of the cycle is the beginning of it. Those two ideas are not decoration on the model. They are what makes it a model rather than a list of three good habits. A list of steps tells you what to do next. A structure that is self-referencing and closed tells you what to do next at any scale, from any starting point, including the one where you are already halfway through and do not know which step you are in.

Which, in practice, is most of the time. The useful question is rarely "what is the first step." It is "which step am I currently standing in, and did I skip the one before it?" That question only has an answer because the structure loops, and it has the same answer at every scale because the structure repeats.


This is the overview of the three steps in the Quest Engine framework: Why, Who/What/When/Where, and How. For the forces underneath the question words, see Quest Engine: The Why and Why Search Maps to Mastery, Not Autonomy.