Summary of "Upstream: The Quest to Solve Problems Before They Happen"

Core Idea

Prevention beats crisis management: Most "inevitable" problems are preventable; the challenge is seeing them and taking ownership before they escalate.
Systems thinking solves root causes: Change the environment and incentives, not just individual behavior—small leverage points shift entire systems.

Problem Blindness

Make invisible problems visible through data visualization, naming, and pattern-breaking (e.g., reframing "normal" behavior as abnormal).

Lack of Ownership

Identify who can own the problem and extend psychological standing to stakeholders without direct skin in the game.
Ask: "How am I responsible for this situation?"

Tunneling (Resource Scarcity)

Create structured slack (standing meetings, huddles) to force systemic work into focus alongside daily crises.
Build prevention into routines so upstream thinking becomes automatic (e.g., fluoride in water).

Unite diverse stakeholders with meaningful roles; use data-driven meetings focused on specific cases (by-name lists), not abstractions.
Map systems strategically: Identify leverage points; sketch "food webs" of interconnections before intervening.
Deploy early warning sensors: Use technology to detect problems early; pair detection with quality metrics to avoid "alarm fatigue."
Anticipate unintended consequences: Test at small scale; ask "what else will change?" before full rollout.
Pair metrics thoughtfully: Combine quantity metrics with quality measures to prevent gaming (e.g., arrests + community trust).
Build closed feedback loops: Ensure data feeds back into continuous improvement, not static reports.
Align financial incentives: Use pay-for-success models and integrated payment systems so prevention benefits the payer, not just downstream services.

Identify one preventable problem in your domain; make it visible through data or naming (document the "water" you're trying to change).
Map stakeholders and leverage points: Who owns it? Who should? What small change creates the biggest ripple?
Assemble a diverse team with structured meetings (e.g., weekly huddles) and accountability mechanisms; give everyone a meaningful role.
Design paired metrics: Define what success looks like (quantity + quality); anticipate how the system might game your measures.
Run a small-scale pilot before full implementation; monitor feedback loops and adjust based on leading indicators, not just outcomes.