The order works because permission comes before production, production before selection, and selection before structure. You cannot edit a blank page, and you cannot spot a useful line until you have generated enough noise to contain one. Each step lowers the stakes just enough for the next to be possible.
Step 1: Give yourself permission to fail first
Opening a blank document is the hardest moment because every option feels permanent. Before you can type freely in step 2, you need to dismantle that permanence. Type the exact phrase “This is going to be terrible, but...” at the top of a new file and then keep going with whatever the sentence demands. For example: “This is going to be terrible, but it might explain why the client keeps stalling.” The phrase gives you a low bar and a running start. What often goes wrong is treating this as a clever writing prompt rather than a disposable trapdoor. People polish the opening line, then stall. Do not fix the sentence, do not choose a title, and do not decide whether the idea is good. The only job is to make the first mark on the page, so step 2 has somewhere to dump raw material.
Step 2: Dump raw words without deleting
With the first line already broken in, the next five minutes are for volume, not quality. Set a timer for 5:00 and type continuously. If you stall, write “I do not know what to say next” or describe the room you are in until a new thought arrives. Do not delete, do not fix spelling, and do not stop to research. A useful example: “the proposal feels stuck because we never named the real decision” can sit beside “my coffee is cold and the neighbor is mowing.” The second sentence is not wasted; it keeps your hands moving until the first one appears. The common failure is treating this as a diary dump and then judging it line by line. Judging during generation is what created the blank page. Your only rule is that the cursor keeps moving forward. By the end, you should have a messy page that contains at least one accidental signal for step 3.
Step 3: Find the accidental sentence worth keeping
After the timer stops, switch from maker to miner. Read the messy page once, top to bottom, without editing. Highlight the one sentence that still makes you lean in: it may be specific, surprising, or spoken in a voice you recognize. For example, in a page about a stalled project, the keeper might be “we keep asking for feedback when we actually need a decision.” It is not the longest or most polished line. A strange pattern: the sentence worth keeping is often the aside you typed while trying to get back on track, not the sentence you planned as your main point. The usual error is picking the most presentable line because it sounds safe. Safe sentences rarely give step 4 enough tension to build from. If two candidates compete, choose the one that implies a conflict, a change, or a next action.
Step 4: Cut everything and develop the keeper
By now the page has served its purpose: it produced one line that can carry weight. Delete all other text, move the keeper to the top, and write the next sentence directly beneath it. Then the next. Build in small moves: define the problem, give an example, name the consequence, ask for the next action. If the keeper is “we keep asking for feedback when we actually need a decision,” the following lines might explain who avoids the decision, what it costs, and what a real decision would sound like. The common trap is saving the garbage “just in case,” which keeps the original mess available and dilutes focus. Another version is rewriting the keeper until it loses its rough edge. Keep the sentence intact. Use it as a spine, not a decoration, and let the rest of the draft answer the question it raises.
A phone reminder or todo app vanishes the second you dismiss it; a paper strip taped near your keyboard gets shorter with each cut and stays there until the final step is gone. That visible subtraction is the point.