How to Pick What to Learn — When Everything Looks Interesting

The short answer: pick by useful-and-curious, then time-box, then kill on data

Cal Newport's So Good They Can't Ignore You (source) demolishes the "follow your passion" advice with a sharper alternative: identify rare and valuable skills, build them deliberately, and let career capital purchase autonomy later. For the practical question of "what should I learn," this reframes the choice. The right learning project isn't the one that lights up the most interest in week one — it's the one that compounds rare skill, sits in the curiosity zone long enough to survive the boring middle, and has a clear way to verify (or kill) the bet without years of sunk cost. Pick by usefulness AND curiosity, time-box the experiment, decide what would convince you to stop.

Why "interesting" alone fails for the picking decision

Many learners pick what to study by interest in the abstract — they're curious about the topic, they want to know more, the field looks fun from the outside. That gets you started but doesn't carry you through the boring middle, which arrives at month two for almost every subject. Pure interest also produces breadth without depth: ten half-learned topics rather than one well-learned one. The usefulness layer is what motivates you through the dull stretch; the curiosity layer is what keeps the daily practice tolerable. Both have to be present; either alone fails predictably.

A four-step picking method

• Write down the use case. Not the vague "this will be useful in my career" — the specific. "I want to ship a small mobile app for my own use." "I want to have business conversations in Spanish within a year." The use case is the anchor that survives the motivation drop.

• Check the curiosity signal. When you read an introductory article in the area, do you want to read another one? When you watch an introductory video, do you finish it? If the curiosity flicker isn't there at the start, the boring middle will be impossible. If it is there, the project survives the dull stretch.

• Time-box the bet. Three months for a small skill, twelve months for a substantial one. Inside the window, the system runs without questioning the choice. Outside the window, you evaluate whether to renew, redirect, or stop. Without the time-box you'll either quit early (felt motivation drop) or grind too long (sunk-cost compounding).

• Pre-decide the kill criteria. What evidence would convince you to abandon this learning project? "After three months I can't build the smallest version of the use case" is a kill criterion. "I haven't enjoyed a single week of it" is one. Pre-written criteria are how you avoid stopping under emotional duress or grinding under emotional inertia.

Why this pays double for ADHD

ADHD learners face the multi-potential trap: every adjacent topic looks fascinating, and the natural impulse is to start them all. The four-step method narrows the choice to one project per cycle, gives it enough scaffolding to outlast the dopamine-novelty drop in week three, and provides explicit permission to stop on data rather than guilt. The kill criteria are particularly important — without them, ADHD learners either quit during the boredom middle (and conclude "I'm bad at finishing things") or push through long after the project stopped earning. Both outcomes damage future picking; the structured pick avoids both.

Where it fails (and the repair)

• A use case that's actually three. "I want to be a designer who codes and writes and consults" is four projects, and none of them gets the focus needed to compound. Pick one specific use case; the others queue. Multi-track ambition is the failure mode the four-step method is designed against.

• Vague kill criteria. "If I'm not making progress" is too soft to enforce. "If after three months I haven't built version 0 of the app" is enforceable. The whole point of pre-writing the criteria is that you write them when calm and check them when foggy; soft criteria allow the foggy version to override the calm one.

• Treating curiosity as enough. If you've abandoned three projects in a row because the boring middle hit and there was no use case anchoring them, the diagnosis is missing usefulness, not missing willpower. The picking method exists to add the anchor that pure curiosity can't provide.

FAQ

What if I'm not sure what's useful for my future?

Few people are. The four-step method works even on partial usefulness signals — a plausible use case, not a guaranteed one. The two-year version of your career is unknowable; the next-twelve-months version usually has a few visible levers. Pick the use case that's plausible at the twelve-month mark and let the longer arc emerge from the wins of that one project.

How do I avoid the multi-potential trap of starting everything?

Keep a parking lot. Every time you want to start a new learning project, write it on a list and date it. When the current project's time-box expires, return to the list and pick the most-still-relevant item. Most of what felt urgent at month one stops being interesting by month three; the parking lot lets you discover this without committing.

Is it okay to abandon a project mid-time-box?

Not on motivation drop alone. Yes on data — kill criteria were met, the use case became clearly false, the project produced unexpected evidence that the path was wrong. The mid-window kill on real signal is fine; the mid-window kill on "I don't feel like it" is the failure mode the time-box exists to prevent.

How do I tell genuine usefulness from career-anxiety projection?

Useful learning has a recognisable concrete outcome (a built thing, a measurable skill, a specific qualification). Anxiety-driven learning chases the field everyone seems to be talking about and has no concrete outcome. The use-case test catches this — if you can't name what changes when the learning is done, the project might be running on social pressure rather than real usefulness.

What's the smallest version to try today?

Pick one candidate topic. Write down the use case, the curiosity check, the time-box, and the kill criteria — twenty minutes. Don't start studying yet. Tomorrow re-read the page. If it still reads as compelling, start. If not, the picking already paid for itself by saving you the three months.