# Decision Principles

The canonical principles behind how Sparker.ai evaluates uncertainty, runs experiments, ends work, and turns decisions into reusable assets.

By Brian Sparker. Published 2026-07-28, updated 2026-07-28. https://sparker.ai/principles/decision-principles-v0-1

## 1. Information is cheap. Judgment is scarce.

Access to information does not resolve a decision. Value comes from knowing which evidence matters, what uncertainty remains, and what action is justified.

## 2. Decisions are products.

A strong decision has a user, an outcome, inputs, constraints, quality criteria, and a moment when it must be delivered.

## 3. Every investment needs success and failure criteria.

Success criteria define what must become true. Failure criteria create permission to stop before sunk cost becomes strategy.

## 4. New technology begins as a Bet.

Novelty is not evidence. Emerging capabilities should receive a bounded investigation before they alter the dependable Platform.

## 5. Evidence determines promotion.

A Bet becomes Platform only when its problem repeats, its method creates value, and its operation can be maintained.

## 6. Ending work is a product skill.

Killing a weak initiative protects attention, capital, and credibility. A clear ending is often more valuable than indefinite continuation.

## 7. No work is wasted when its learning is captured.

Code, frameworks, artifacts, retrospectives, and changed opinions can compound after a project ends.

## 8. Practical artifacts are stronger than abstract commentary.

A benchmark, prototype, decision memo, evaluation set, or working tool makes reasoning inspectable.

## 9. Consulting should create reusable intellectual property.

Client work should improve the next engagement while respecting confidentiality and ownership.

## 10. Platforms should emerge from repeated need.

A platform is the residue of problems solved repeatedly, not a product category selected in advance.

## Version note

This is version 0.1, published July 28, 2026. Future revisions will preserve the prior version and explain what evidence changed the principle set.