Quick self-audit (green flags vs. “just loud”)
Intelligence looks like fast learning + accurate predictions + clear explanations + useful outcomes. Run the tests above for two weeks, score yourself, and you’ll know—no vibes, just receipts.
- Learning speed: You pick up new, complex stuff fast and retain it. (Test: learn a topic you know zero about for 60 minutes, teach it back in 5.)
- Transfer: You can use ideas from one field to solve problems in another.
- Calibration: Your predictions match reality. You’re good at saying “70% sure” and being right ~70% of the time.
- Error correction: You notice your own mistakes quickly and fix them without ego.
- Abstraction + detail: You can zoom out to the principle and zoom in to the numbers.
- Model building: You explain why something happened, not just what.
- Writing clarity: You can explain a hard thing simply, without dumbing it down.
- Mind-change rate: You change your mind with new evidence—neither stubborn nor flimsy.
- Attention control: You can focus deeply and also know when to stop.
- Taste: You can tell good ideas/work from mediocre ones before others do.
- Forecast log (10 mins/day, 14 days): Write 10 concrete, time-boxed predictions (e.g., “X stock up/down >2% by Friday,” “I’ll complete Y task by 3pm”). Score them with a simple Brier score (lower = better). If you calibrate well, you’re thinking clearly.
- Fermi set (20 mins, twice/week): Answer 5 back-of-the-envelope questions (e.g., “How many piano tuners are in my state?”). Check rough answers later. You’re testing estimation, decomposition, and sanity checks.
- Transfer sprint (2 hours): Learn a niche method (e.g., Bayesian updating, constraint programming, memory palaces). Apply it to a real problem you have within 48 hours.
- Explainer note (once/week): Pick a hard concept, write a 300-word explanation for a smart 12-year-old. Share with one critical friend. If they get it, your thinking’s crisp.
- Code or calc check (30 mins): For anything numeric, do it two ways (spreadsheet + hand calc). Consistency = rigor.
- Standardized indicators: IQ-type tests can correlate with problem-solving, but they’re noisy and gameable. Treat them as one datapoint, not gospel.
- Peer signal: Ask 3 brutally honest people: “What do I do better than almost everyone? Where am I delusional?” Convergence across them is meaningful.
- Output trail: Smart isn’t talk; it’s artifacts—forecasts that hit, analyses others use, systems that keep working when you’re not there.
- Write down your top 3 biases (e.g., overconfidence, halo effect). Add a pre-mortem to big decisions: “It failed in 6 months—why?”
- Cap your certainty: ban 0% and 100%. Use ranges.
- Separate status moves from truth moves. If you’re optimizing for applause, you’ll get dumber in public.
Intelligence looks like fast learning + accurate predictions + clear explanations + useful outcomes. Run the tests above for two weeks, score yourself, and you’ll know—no vibes, just receipts.
Comment