Chapter 2 · The Judgment Crisis in a Saturated Environment

Extended content

Liquid part: this text keeps growing after the printed edition.

Ten versions and one hour

Open the last report you asked the tool to produce in several versions and, before reading any of them, write on a piece of paper two discard criteria you can apply in a minute per document. For example: it answers the question that was actually posed, and it contradicts nothing that only you know about the client. Run all ten through that filter, looking at the first paragraph and the figures. Whatever survives gets read in full. The rest is filed away without regret.

Put a clock hour on the exercise and decide when to stop before you start. Without a limit, evaluation expands to fill the day, and that is the permanent state of evaluation the chapter describes. With practice you will see how much filtering your kind of work calls for, and the discard criteria will go from being a piece of paper to being a reflex.

A worked example

Lucía leads a team that maintains software for an industrial plant. A month ago she switched on an agent that announces in the shared channel every change it prepares. At first it felt like transparency. The first Friday she runs the audit, she answers the first question with "quite a few, and I don't know in how many there was an error," which is exactly the answer the chapter marks as a diagnosis. To the second she answers no: not a single full one-hour block. And at the fourth she remembers a decision. She approved a deployment change on Thursday at six in the evening, after eleven alerts from the channel and without having reread the part that touched the plant.

She doesn't switch tools. She agrees with the team that the channel will announce only two kinds of change, that "urgent" means a phone call and not a message, and that on Tuesday and Thursday mornings nobody interrupts anybody until noon. Critical approvals move to those mornings. Three weeks later, her Friday no longer looks like the first one.

Getting the knowledge out of your head once

Pick a task where you gave up on the tool at the fourth version of the prompt. Imagine you have to explain it to a colleague who starts on Monday and won't be able to ask you anything. Write the card: who the result is for, what quality standard is expected, what must never be said, what mistake someone once made and what it cost.

It is tacit knowledge put on a page. It takes an afternoon. From then on every specification starts from the card, and the fatigue stops resetting to zero each morning. In time the card teaches you something else: which parts of your trade you knew without knowing you knew them.

Questions for your Friday review

• Where does my team's judgment reside today, and who exercises it?
• Which output did I accept this week because reviewing it no longer fit in the day?
• Which channel interrupted me most often, and what decision was inside it?
• At what time of day do I make my best decisions, and at what time my worst?
• Which red line did I have last month that I can no longer locate today?

Answer them in writing, in five minutes, and keep the sheets. After a few weeks the pile says more than any single answer.

Your lines, in writing and dated

Take an index card and write four headings: what is serious, the question, the context, your limits. Under each, put a real case from your current work. Under "what is serious": the proposal that goes out to a client with your signature. Under "the question": what this report is really being asked to do. Write today's date and a review date three months out.

The lines move, and a line that moves without your deciding it is delegation without architecture under another name. When the date comes, don't renew it out of inertia. Ask what has changed in the tool and what has changed in you.

Moving on

Once you know where your judgment is and how much it can bear, what remains is to understand what you are handing part of the load to. Chapter 3 looks at the other side of the screen.