FRAMEWORK

The Law of 2

Time-Boxed Conviction Scaling. Not about speed. About earned conviction per unit time.

Increasing Investment
Increasing External Exposure
Increasing Production Hardness
Increasing Organizational Gravity
Phase 1
2 hrs
Prototype

Signal Generation & "Spin Up"

The first gate. Can we generate any signal at all? This is the minimum viable experiment — not a product, not a pitch. A probe.

Build the simplest possible thing that can answer: is there real signal here? Not user research. Signal. Engagement, willingness to pay, something that moves.

Gate Question
Is there real signal?
Decision
Kill / Pivot / Escalate
Phase 2
2 days
Labs

Pressure Testing & Torque

The signal was real. Now pressure test it. Two days of concentrated execution to determine if the signal is durable enough to justify two weeks of work.

Apply force from multiple angles. Talk to potential users. Build rough models. Look for the break points. Conviction is earned through stress, not comfort.

Gate Question
Is conviction strong enough for 2 weeks?
Phase 3
2 wks
Design Partner

The Shape of the Box

Conviction earned. Now find the shape. Two weeks with real design partners — people who will tell you when you're wrong because they want you to be right.

The test is not "did they like it?" The test is "did they actually use it?" Usage is signal. Enthusiasm is noise.

Gate Question
Did the customer actually use it?
Phase 4
2 mos
Full Commit

Production Hardness

The shape is right. Now build it to last. Two months of full commitment to production hardness — the point at which the venture can stand on its own physics.

Scale requires durability. Durability requires commitment. At this phase, the factory allocates maximum resources and expects maximum signal in return.

Gate Question
Is this durable and scalable?
Decision
Scale / Sunset
Kill Criteria

Stopping is Success

A kill decision at Phase 1 is better than a zombie at Phase 4. The framework exists to surface kills early, not to justify continued investment in weak signal.

Failure is Signal

Every failed experiment generates data that sharpens the next one. The factory doesn't lose when a venture dies early. It recalibrates and reloads.

The Ant Colony Model

No single venture is precious. The colony survives through distributed effort and rapid adaptation. Individual experiments are expendable. The system is not.

Cultural Soil

Everyone Wears All Hats

Specialization is for insects. At the factory, every team member is expected to engage across functions. Depth is earned. Breadth is required.

The Ant Colony

Decentralized intelligence. No single point of failure. Everyone understands the mission well enough to make good decisions without permission.

Inner Simon & Inner Rick Rubin

Be rigorous about signal (Simon). Be curious about possibility (Rick Rubin). Hold both. The factory lives in the tension between them.