↩ ExploRational · A Catalogue of Divergent Rationality
explorational · a divergent move
The debugger does not see the fix-the-bug frame; she sees the bug. Step back far enough and the frame you were seeing with becomes a thing you can see — named, scoped, priced, set beside its alternatives. Robert Kegan called this shift the engine of adult development.
Kegan's account of growing up contains no facts acquired and no skills mastered. Development is a change of address for the frame. Whatever you are subject to organizes experience from behind the eyes: you cannot examine it, because it is what you are examining with. Whatever you hold as object you can inspect, doubt, compare, and take responsibility for. Growth is the transfer — rules you operated by become rules you operate on. The adolescent who was her impulses comes to have impulses; the adult who was his ideology comes to hold one, and can finally hear a second.
The working version, for anyone with a method: you can use the method, or you can see it as one method among several, with a scope and a cost. Nelson Elhage performed the move in public, in both directions, one week apart. First the credo — computers can be understood, because "There is no magic" — a frame he had spent a career inside. Then the counter-essay: distributed systems, big balls of mud, client-side JavaScript, deep learning — systems where pursuing detailed understanding "was either impossible, impracticable, or just a waste of time," and experiment and statistics do better. His closing holds the pair at arm's length: "there's a lot of value in holding both mindsets as tools available for you, depending on the problems and needs of the situation." Two frames on a bench, chosen between per problem. The instrument below stages that flip.
the signature instrument
Three scenes, each written twice. At the subject level you are operating by a frame: it supplies the verbs and appears nowhere. One click steps back — the same scene shrinks to a specimen with a name, a scope, a cost, and shelf-mates. Nothing here is computed; the flip in the typography is a rehearsal for the one in your reading.
One scene, two levels
three scenes · one flip · nothing measured
—
—
the instrument holds no model and takes no measurements: the same authored scene, typeset at two levels. whether the flip happens is decided in your reading, never on this page.
what had you, you now have
A frame in use is transparent. Attention goes through it the way sight goes through glass: while debugging, reproduce, bisect, patch feel like descriptions of the world, and they are the frame's vocabulary. Ask an object-level question from inside — what is this practice for? — and it either parses as obstruction or gets answered in the frame's own terms: for fixing the bug. The move requires somewhere else to stand, a second description in which the first shows up as an item. Kegan's point is that you lose nothing in the step back: the frame keeps working. You gain a hand to hold it with.
The tell that a frame has you is a cost you cannot see. Elhage supplies a confession-grade example: his comprehension mindset hardened into "a need to understand," and one afternoon of standing up a single AWS Lambda endpoint became thirty documentation tabs and a dead end, because the frame graded following a tutorial as failure. "The need to understand can be seductive and harmful when working atop a complex system," he writes — a sentence available only to someone who has set the mindset on the bench. He kept it, on his own terms; the essays are the evidence that he now holds what used to hold him.
what to try
Flip the debugging scene a few times and track the verbs. Inside, they act on the bug: reproduce, bisect, patch. Outside, they act on the frame: name, scope, price, compare. Content and frame trade places — that trade, caught happening in a sentence about your own work, is the move.
Stay at the subject level of the estimation scene and try to ask the object-level question — what is estimation for, here? Watch it fail to parse: from inside it sounds like an attack on the sprint, or gets answered in points. Questions about a frame need somewhere else to stand. The flip is that somewhere.
Take the third scene seriously. Reread the badges in the masthead, flip, and reread them as one filing scheme among alternatives — then find one thing about this move that the five axes cannot file. Timescale is a seed: Kegan's shifts take years, and the catalogue shelves this one beside five-minute techniques.
the cloak, the checklist, the lens
Einar Høst's Death of a Craftsman runs the move on the hardest system there is: the operator. "You might say I tried on the craftsman's cloak, and my god, it fit me!" While worn, the identity did its grading invisibly — "If the software is broken, I simply wasn't good enough." Fifteen years in, "the cloak of craftsmanship was starting to itch and no longer felt right… So finally, I had to take it off." Held out as an object, the identity acquires edges: it explains discipline and pride, and it has no word for the forces his experience said were decisive — "programmers don't make software… Organizations make software." The cloak survives the talk, demoted from skin to garment.
The falsehoods genre industrializes the same move for schemas. Each list takes a rule its readers validate with — a name has a first and a last part, a day has twenty-four hours, an address locates a building — and lays it against the world's counterexamples until it is visibly "an idea that you initially believed was true," no longer the shape of reality. And Chris Krycho aims the move at the profession's master lens: "The act of programming is — fundamentally, inescapably — an attempt to systematize the world: to make it legible to computer systems." A lens seen gets a scope; Krycho borrows Scott's choice — embrace legibility as limited, or force the world to fit it. Only someone who can see the lens gets the choice.
the honest instrument
Both levels of every scene were written by the same hand, so the object level is a portrait of stepping back, done for you — which is exactly what the move never is in the field. Kegan's evidence comes from interviews across decades: subject–object shifts arrive on the timescale of lives, resisted, and with losses, because what is being renegotiated is the self's structure, comfort included. The click rehearses the motion. The years build the muscle.
the move ↔ the instrument
| The instrument | The move |
|---|---|
| the subject level | operating by the frame — it supplies the verbs and appears nowhere in the scene. |
| the flip | Kegan's shift: what had you becomes what you have. |
| the specimen tag | naming the frame — until it has a name, scope and cost have nothing to attach to. |
| scope · cost | a method seen as a method: where it works, and what it hides while working. |
| the shelf | a frame seen is a frame optional — alternatives exist only at the object level. |
| stepping back in | re-embedding: the move ends in a chosen frame, never in a view from nowhere. |
how this opening fails
risk · endless deferral
The catalogue's named risk for this move. The vantage from which you objectified the frame is itself a frame; the move can be run on the mover, then on that mover, indefinitely — every level examined, none operated. Kegan's ladder has a rail: each shift ends in a new embedding, because you are always subject to something. The move's product is a re-entry — the bug fixed, the estimate given, by a frame now held on purpose — never a frame-free view.
risk · object read as refutation
Holding a frame at arm's length says nothing against it. Most debugging sessions should end where they began — patch the race, close the ticket — the same acts, now chosen over named alternatives. And the move flatters its user: naming other people's frames from above is one-upmanship with a pedigree, and it runs on a frame of its own — meta wins — that belongs on the bench with the rest.
What has you, you cannot examine; what you have, you can put down — and what you can put down, you can pick up on purpose.
can you use it?
RECOGNITION — Which is making the system an object? A: pushing harder inside the sprint process to finally hit an estimate. B: pausing to describe the estimation ritual itself — its purpose, its cost, what could replace it. C: quitting estimation in disgust.
B. The move looks at what was being looked through. A is the frame still operating you; C discards the frame without ever seeing it — the opposite failure, and a common counterfeit of this one.
THE NEAREST NEIGHBOR — Stepping out of the system also goes up a level. What separates this move from it?
The deliverable. Stepping out plays the larger game: the current system becomes a move inside it. Making the system an object examines the frame — name, scope, cost, alternatives — and typically ends by stepping back in, to the same frame or a swapped one, now held instead of had.
PRODUCTION — Name one rule you are operating by today — in your codebase, your team, your reading of this page — and write its museum tag: name, scope, one cost, one alternative.
A version: "All changes via pull request — scope: shared branches; cost: ten-minute fixes take a day; alternative: pair-and-push for trunk work." Yours works if the rule appears as a noun with edges, and you can state its purpose without using its own vocabulary.