Reframing Complexity: Designing Solvable Problems (book summary)

Complex problems do not survive because we lack effort or intelligence. They survive because the fit between reality, the way we describe it, and the way we work on it is mismatched. We reach for speed when the map is wrong. We polish dashboards while the seam that matters — names, units, time, IDs — remains ambiguous. We optimize a proxy because it is visible, and then watch the target drift out of reach. The world moves while our learning loops lag; a rare, adversarial tail arrives and poisons cadence; a silent dependency flips a bit and truth turns into rumor. Under pressure, handoffs become the real bottleneck: one team’s assumption leaks into another’s interface, everyone is busy, and progress stalls without a single villain to blame.There is also the physics of the thing. Some landscapes are nonconvex or combinatorial, hostile to wishful thinking and brute force. Safety and fairness add constraints that will not negotiate with a tight deadline. Incentives pull in different directions: local KPIs reward throughput, global outcomes require care, and the organization ships green metrics into a red reality. Memory fails to compound. We rediscover the same subproblems because witnesses, templates, and guards are not kept where the work lives. Ethics live in policy instead of mechanism, so, at the first hint of stress, they vanish.This book exists because these forces can be tamed by craft rather than heroics. The craft is modest and repeatable. Choose a map that makes local truth legible — one timestamp that means “when this happened,” one identifier that survives refactors, units that travel through the pipeline, features and constraints that expose structure instead of hiding it. Name the few invariants you refuse to trade and place them where the mechanism must pass, not in a document nobody reads at two in the morning. Carry receipts: a gap beside every plan, a bound beside every claim, a witness beside every result, lineage that makes yesterday’s truth replayable tomorrow. When the world tilts, prefer reversible repairs to grand gestures; make the smallest safe change that preserves the invariants and resume. Let success compound by storing oracles so that tomorrow’s work is verification, not search. Work at a cadence where feedback arrives before consequences.