PRAGMA · Marshall, Sebastian

Finish the part that lets reality answer

An original essay drawing on PRAGMA’s arguments about unfinished work, meaningful measures and identities that remain open to correction.

The last step changes the kind of information you have

Marshall and Kai Zau had prepared an advertising experiment: landing pages, keywords, technical setup and an account. After roughly three days of work, the project was almost ready. Marshall then left it for five days while other priorities intervened. He does not describe a catastrophe. He describes a small delay that kept returning to mind and postponed the first evidence from the market.

The distinction is easy to miss. Before the experiment ran, the team had evidence that it could prepare an experiment. After it ran, it could begin learning whether the experiment worked. Those are different achievements. A project can be full of genuine effort while still missing the step that exposes its central assumption to a response.

This is an editorial reading of the anecdote, not a diagnosis of why Marshall delayed. He acknowledges other important work. The question is more useful than a judgment about willpower: which unfinished step is preventing the kind of information the project exists to obtain?

Progress has to be connected to its purpose

PRAGMA repeatedly warns against mistaking an attractive intermediate quantity for the desired result. In its Magic example, a player can accumulate cards and still lose before using them. In its burger-stand example, faster order-taking does little for completed meals when the grill is the constraint. Both examples ask what a local improvement changes downstream.

That question is useful around almost-finished work. A polished outline, a large research folder or a beautifully arranged task board may be valuable. Their value comes from the work they help make possible. If the next useful test requires a rough prototype, another hour of sorting research may leave the actual uncertainty untouched.

The prototype is an invented illustration, but the reasoning follows Marshall’s distinction. A good measure should help explain whether the objective is becoming more attainable. It should not merely provide a number that can keep improving while the objective waits.

Make the test small enough to finish well

‘Finish’ can sound like an instruction to push an entire large undertaking through exhaustion. Marshall’s more practical suggestion is to define complete stages. He separates writing into activities such as outlining, drafting and editing, and describes scheduling a full outline when the available session is too short for the whole essay. A stage becomes useful when its completion changes what the next stage can do.

The same idea changes the cost of learning. A small complete attempt can answer a bounded question while leaving room to revise. An enormous half-built attempt may keep every assumption entangled. This does not mean that all work should be rushed into public view. The needed result might be an internal review, a checked calculation, a private rehearsal or a carefully scoped trial.

Marshall’s reserve-capacity argument belongs here. A finishable unit still needs enough time for an awkward case, a correction or a failed component. Filling the calendar with optional work and then calling every delay unexpected is a way of spending the reserve before the uncertainty arrives.

Build a routine that survives the next repetition

The tour-media story offers a modest model. One tour produced an unwieldy pile of photographs and video. Later tours used agreed storage locations, publication destinations and a processing cycle after each event. The change was partly about knowing what ‘done’ meant and partly about having a reliable occasion to do it.

Automation extends this idea beyond machines. A checklist can preserve a decision; a template can make an acceptable result repeatable; a well-chosen automatic step can remove a recurring intervention. Marshall argues that the obstacle is often the cost of pausing to improve the method when doing one more piece of work feels more urgent.

His account of Zau supplies a useful habit: review the previous run before repeating it. Retain what worked, identify the recurring emergency and change enough to make the next attempt easier. The routine need not become perfect to stop wasting the same lesson.

Let the result revise the story about yourself

There is a quieter reason feedback can be difficult. A project is sometimes carrying an identity as well as a practical objective. A person who is trying to be seen as insightful, creative or reliable may find an unfinished promise easier to protect than a finished result that can be assessed. PRAGMA’s Being And Doing chapter explores how favorable labels can become shields against examining conduct. This is a general possibility, not an inference about the advertising anecdote.

Marshall also supplies an alternative: let identity name a continuing obligation. The writer’s claim becomes a practice of writing and revising; reliability becomes a record of handled commitments. A disappointing result can then tell the person what to improve without deciding their entire worth.

The language essays make the same demand at the level of explanation. Calling an idea revolutionary, a person brilliant or a plan strategic does not reveal what changed, what happened or why it should work. Specific descriptions make room for specific corrections.

Choose the next contact with evidence

One way to apply these ideas is to look at an open project and write a plain sentence: ‘The next complete result will let us find out whether…’ If the sentence cannot be finished, the immediate problem may be the objective. If it can, identify the smallest responsible unit that produces that evidence, the quality it requires and the margin needed to complete it.

This is an editorial practice assembled from Marshall’s chapters, not a named method in the book. It also has limits. Sometimes waiting is necessary; sometimes incubation is useful; sometimes the evidence says to stop. The point is to reduce avoidable uncertainty and repeated handling, not to make completion a moral obligation independent of the work’s value.

The advertising example remains powerful because it is ordinary. The final step did not need to look heroic. It needed to let the project begin answering the question for which it had been built.