This order works because a treasure must exist before clues can point anywhere, and clues must exist before sources can hide them. The map stays rough, the expeditions stay short, and the final build exposes gaps that become the next hunt.
Step 1: Name the treasure first
Start by choosing a treasure that can be displayed, explained, or used—not a vague goal like 'learn Python.' A treasure might be a one-page explainer of how neural networks learn, a working Arduino traffic light, or a ten-minute talk on why Rome fell. This artifact gives every later clue a direction. Without it, you will collect facts that never connect, and the hunt becomes scrolling with extra guilt. Write the treasure on the top of your paper strip and date it. Then add one sentence describing who will see it and when. That deadline is not pressure; it is the edge of the map. People often pick a treasure so broad that no single clue can confirm progress. Narrow it until you can imagine the final item sitting on a table.
Step 2: Draw your rough map
Before you search for sources, sketch the territory on one page. Put the treasure at the center, then draw five to seven blank zones around it—questions, skills, or periods you cannot yet explain. For a Rome talk, zones might be 'republic vs. empire,' 'army loyalty,' 'bread and circuses,' and 'why 476?' The point is not a beautiful diagram. It is to mark what is unexplored, because blank zones are where clues hide. A map that only lists what you already know is a summary, not a treasure map. Keep the map loose enough to redraw when a clue changes the shape. A common error is making the map too neat or too complete; that turns it into a poster you admire and never revisit. Tape it where you will see it during the hunt.
Step 3: Write clues as questions
Now turn the blank zones from step 2 into three clues written as questions you cannot answer yet. Good clues are specific enough to hunt and open enough to surprise: 'Which Roman emperor first paid the army from the treasury?' or 'How does a neural network adjust weights without being told the rule?' Write each clue on its own line, leaving space beneath for the answer and the source. This order matters because questions created before sources keep you from collecting random highlights. When you open a book or video first, you tend to accept whatever it emphasizes as important. A frequent shortcut is writing questions you already know the answers to; those feel productive but teach nothing. Keep at least one clue that makes you slightly uncomfortable. Fold the strip so only the clues show, not your current guesses.
Step 4: Hide clues in sources
With your three questions written, assign each one to a real hiding place. For a book, that means a chapter title plus page range; for a video, a timestamp range; for a conversation, a person and one question you will ask. Write the hiding place next to the clue. This turns vague research into a trail you can follow when motivation dips. For example, clue one might hide in chapter 4, pages 71–89 of a Rome history, while clue two hides in a 12-minute neural network explainer starting at 3:20. What derails this step is choosing sources by convenience alone—only free PDFs or only short clips. A source that cannot answer the clue wastes an expedition. You do not need the perfect source, just one plausible hiding place per clue. If you cannot name any, the clue is probably too broad and should be split.
Step 5: Run one clue expeditions
At this point the clues and hiding places are fixed, so the hunt becomes mechanical in the best way. Set a timer for twenty-five minutes and open only the source assigned to one clue. Your job is not to finish the chapter or watch the whole video. It is to return with one answer, one quote, and one new question. Write the answer under the clue, add the page or timestamp, and mark the map from step 2 if the territory shifted. Then stop. The counterintuitive part is that leaving a source unfinished keeps the hunt alive; finishing everything often ends curiosity. A common failure is opening three tabs and letting the algorithm steer. If the assigned source truly lacks the answer, note that and reassign the clue, but do not start a second expedition in the same sitting. One clue per session is the pace.
Step 6: Assemble the treasure
Finally, use the finds from your expeditions to build the treasure named in step 1. If the treasure was a ten-minute talk, arrange your answers and quotes into three points with a beginning, turn, and ending. If it was a working circuit, wire the version your notes describe. The build exposes gaps: a missing quote, a step you cannot explain, a date you guessed. Those gaps become the first clues of the next hunt, so the process loops instead of ending. A frequent misstep is waiting until every gap is closed before making anything; that postpones the feedback that tells you what to hunt next. Show or use the treasure within a week, even rough. Then write three new questions on the back of the strip and start again. The treasure is proof the map worked, not the finish line.
A phone reminder disappears the moment you dismiss it and leaves no trace of how far you have come. A paper strip taped where you will see it visibly gets shorter, stays put until the last clue is found, and turns the sequence into something you can hold.