The Decision Protocol

When facing any significant technology leadership decision, run through this seven-step protocol. Each step forces a descent from analogy and convention to first principles.

Step 1: Deconstruct. Strip the decision to its irreducible components. What are we actually trying to achieve? What do we know to be true? What constraints are real and which are legacy?

Step 2: Identify the Compounding Variable. What capability will this decision make more valuable over time? Favor the option whose primary effect is to increase the rate at which the organization learns and ships.

Step 3: Map the System Coupling. Trace second- and third-order effects. What other systems does this touch? If this succeeds beyond projections, what breaks? If this fails, what else fails with it?

Step 4: Test for the Human-Machine Design. Does this put humans in the position of judgment while machines handle scale? Any system that removes humans from the loop without elevating their judgment is a liability.

Step 5: Apply the Ethics Constraint. Who could be harmed by this? Does this concentrate power or surveillance? Would we be comfortable if this system's decision-making were public?

Step 6: Evaluate Velocity Impact. Does this add a handoff, a gate, or a waiting state? If the decision reduces velocity, it must be justified by a compounding benefit.

Step 7: Commit with a Kill Condition. What would we need to observe to know this decision was wrong? What is the maximum time we will wait for validation? Who has authority to reverse this without a committee?

Every decision ships with a kill condition. If the kill condition is not defined, the decision is not ready.