From Practice to Instrument
The template is where a standard becomes explicit enough to be enforced by something that has no standards of its own.
What this page is for
The method’s page says what the six steps are. This page says how to do them well.
The rule between the two pages
Anything that decides what a result is lives on the method’s page. Anything that decides how well the steps are executed lives here.
One warning before the craft: if turning work into instruments is not repeatable — if every new act needs improvisation from zero — then everything downstream is luck. This page is tested as hard as any experiment, because it is the Lab’s answer to the direction discipline question: is this a discipline, or a knack?
Seeing the work as it is actually done
Before an act can be named, it has to be seen — and the job description is the last place to look. Watch the work as it actually happens: its inputs, its shortcuts, its unwritten rules. Take the issue-sorting example: the written process says issues are sorted by severity and impact. The real act includes a third input nobody wrote down — who raised it.
Most of what makes work delegable or not lives in what nobody bothered to write down, and it is not found at a desk. This is how the method’s first two steps are done honestly; skipped, the sentence written in step one names the work as imagined, not the work as it exists.
Naming and cutting
One sentence, one act, one owner. Bundles of acts produce bundles of failures that cannot be told apart — if the instrument does three things and the output is wrong, nothing has been learned about which of the three broke.
When the decomposition is done, look where the weight sits:
- an act that is mostly intake and output invites delegation;
- an act that is mostly judgment resists it.
That reading, made before anything is built, is the first sight of the boundary.
Predictions that can be wrong
A prediction that cannot fail is a hope. “The agent will do well” predicts nothing. “The agent will apply the sorting rules more consistently than a person, and will miss the issue that is small on paper but political in fact” — that can be wrong on the record, which is what makes it worth writing.
The discipline: name the part expected to hold, name the part expected to break, and date it. Being unable to predict where it will break means the work has not yet been seen clearly enough to build.
Making the standard explicit
Fix the template before building anything: the structure, the tone, the quality bar of an output worth signing. The template is where a standard becomes explicit enough to be enforced by something that has no standards of its own.
This step has a second use, and it is the deeper one: whatever cannot be written down — the judgment that can be applied but not described — is a first sighting of the boundary, before a single run.
Building the skill
Encode one act as one instrument, with the instructions written the way work is handed to a capable colleague — not the way instructions are written for a machine:
- the inputs, defined;
- the template, attached;
- the context and constraints, stated;
- and what good looks like, shown.
One skill per act. Orchestration — instruments calling instruments — comes later, and only when the single acts hold. An orchestration of untested parts is not an experiment; it is a pile of untested parts.
Handing it back
A dry run may debug the machinery. Only a live cycle — real inputs, real stakes, an output someone was waiting for — produces evidence. From there the instrument is judged like everything else in this Lab, by the method’s dual gate, and its result recorded like everything else. The craft ends where the law takes over.
What accumulates
Every instrument built this way outlives its experiment. The templates, the skills, the decompositions remain — and each new act costs less to instrument than the one before, which is itself a measurement worth watching: it is the transfer question answered in practice.
The Lab’s product is not only verdicts. It is a growing workshop of instruments, and the record of what it took to build each one. And an instrument that survives its verdict does not go back on the shelf — it goes into operation, and that discipline belongs to the method’s sixth step.