Thinking & Models

On Intelligence: Key Ideas


Premise

Two Definitions.
One Breaks. One Holds.

A definition of intelligence is only useful if it survives its edge cases. Naval Ravikant's outcome-based version meets a monk and breaks. Elon Musk's prediction-based version holds.

Definition One

"Intelligence is getting what you want."

Naval Ravikant proposes a pragmatic test: the only true metric of intelligence is whether you get what you want.

  • It rewards outcomes — reading the terrain, not scoring well on tests.
  • But it is a ratio: wants met over wants held.
  • That makes the score hostage to the denominator.
Raise desire. Watch the same wins score as less intelligence.
CALCULATED RATIO 50% Wants met: 50% Desire: 50

Intelligence = wants met ÷ desires. The score hangs on the denominator.

The Edge Case

Tuning the denominator to zero.

A monk cultivates total equanimity. He wants nothing — and so gets everything he wants.

  • By Naval's definition, the monk scores perfect intelligence.
  • But tuning the denominator to zero is transcendence, not computation.
  • The monk exits the game; he does not win it.
Mental Mode:
Toggle the modes: active computation vs. desire zeroed.
Transcendence (Desire = 0) Predictor: Processing variables & mapping paths

Zeroing wants solves the equation. It never runs the computation.

The Conflation

Desire management vs. reading the terrain.

Tuning wants is a real skill — but a different one. Intelligence is the processing instrument; desire management is the throttle.

  • Lower your desires and you find peace; your intelligence is unchanged.
  • High intellect coexists with failed peace all the time.
  • Separate the faculties and the computational one comes into focus.
Switch faculties: one dials wants down, the other reads the world.
Shrinking Wants Market Physics Behavior Outcome Mapping Terrain

Desire control collapses wants. Predictive capacity charts trajectories.

Definition Two

"Intelligence is predictive ability."

Elon Musk proposes a definition that survives the monk: intelligence is the ability to predict the future.

  • An intelligent system reads present noise and triages what matters.
  • Its output is a forecast: states not yet realized.
  • The measure is computational capacity, independent of desire.
Raise capacity. Watch the cone narrow and reach further.
Past Now Future

More capacity narrows the cone of uncertainty and reaches further out.

Examples

Traversing time horizons.

Predictive intelligence runs under every skilled act. Pick an archetype to trace its path:

Each compresses different factors into one forecast.
CHESS HORIZON Checkmate (T+12) BRIDGE TRUSS LOAD MODEL Predicted Stress Fracture Point Load Vector MARKET TREND FORECAST Weighted Value Expectation

Each discipline weights different factors toward one forecast.

Compression

Intelligence is reality compression.

Reality offers infinite noisy parameters. Intelligence strips them to a model small enough to act on.

  • Noise is high-dimensional; signal is low-dimensional.
  • To predict well is to have compressed reality into rules.
  • Compression capacity sets the ceiling on forecast reliability.
Raise compression. Watch noise resolve into signal.
RAW DATA (NOISE) Lens COMPRESSED MODEL

Compressing noise into a few rules is the core of intelligence.

The Future

Intelligence without consciousness.

Artificial systems fold proteins with startling precision. They experience nothing, want nothing, feel nothing — and predict.

  • No desires at all — yet unmistakably intelligent. The monk paradox dissolves.
  • What they run is pure compression: physical rules extracted from data.
  • Intelligence is computational, not emotional.
Protein Fold Model:
Run the fold. Structure falls out of pure prediction.
STATUS: LINEAR UNSTRUCTURED

Folding a protein is deep prediction — with nothing that feels.

Conclusion

The Monk &
The Predictor.

The monk is at peace. The predictor is intelligent. They may be the same person, but they exercise different faculties — and only one of them is intelligence.