Key Takeaways
A richer economy can still fail if ordinary households stop moving forward
Growth is not enough. The book opens with Sarah, a kindergarten teacher with graduate degrees who struggles to support her son on $55,000 a year. Interviews with other working Americans reveal a related frustration: people followed the expected path, yet feel economically stuck and politically unheard. That matters because democratic capitalism depends on broad public consent, not only rising totals on a national balance sheet.
The middle has stalled. From 1947 to 1976, median family income grew about 2.4 percent a year, doubling over the period. From 1976 to 2018, it grew only about 0.6 percent annually, a gain of 31 percent over a longer span. Adjustments to the figures change the scale, Martin argues, but not the direction.
The opening makes a powerful distinction between aggregate prosperity and lived security. Median income is not a complete account of well-being, since household size, public services, debt, and changing prices also matter. Still, a society cannot assume that gains captured by a small slice of people will translate into confidence for everyone else. Political science adds a useful lens: perceived fairness can matter as much as absolute income in sustaining institutional legitimacy. Sarah’s story gives the statistics a human test: can a person who did what society asked build a stable life?
Stop treating the economy as a machine you can perfect
Models guide action. A model is a simplified picture of reality that helps people decide what to do. The machine model treats the economy as parts that can be analyzed, optimized, and assembled into a predictable whole. The alternative is a complex adaptive system: a network whose participants respond to one another and change their behavior as conditions change. Its outcomes cannot reliably be predicted by adding up separately optimized parts.
Forecasts expose the limits. In 2008, all fifty forecasters in the Blue Chip survey still predicted positive US growth for the year in December, despite a steep contraction already underway. In 2013, forecasts of a severe economic shock from the so-called fiscal cliff proved badly wrong. Such failures do not make models useless. They show why confidence in a single model, treated as reality, can mislead.
This argument connects economic policy to a broad lesson from systems science: interactions can produce outcomes that no component displays on its own. The machine metaphor encourages decomposition, but the economy includes people who anticipate policy, copy competitors, and change the very conditions being forecast. That makes humility operationally valuable, not merely philosophical. Still, imperfect prediction alone does not prove that every machine-like model is misguided. Models can clarify specific mechanisms. The practical test is whether their assumptions fit the question, and whether decision-makers remain willing to revise them when reality pushes back.
Never let a convenient metric impersonate the outcome it only approximates
Measures can hijack goals. Surrogation occurs when a measure of a desired outcome is treated as if it were the outcome itself. A standardized test may approximate learning; a customer survey may approximate loyalty. Neither captures the full reality. Once rewards or punishments attach to the number, people can focus on improving the number instead of the underlying result.
Wells Fargo shows the danger. The bank wanted deeper customer relationships and used accounts per customer as a proxy. Aggressive sales targets encouraged employees to open millions of accounts without customers’ knowledge or consent. In Atlanta, some educators facing pressure over student test scores cheated or coached students narrowly for the test. In both cases, a measure intended to represent success became a target that distorted behavior.
The pattern resembles Goodhart’s law, the observation that a measure often becomes less useful once people are rewarded for maximizing it. Metrics remain indispensable: without them, managers and public agencies cannot notice patterns or compare results. The safeguard is not to abandon measurement but to combine indicators, examine how they can be gamed, and keep checking them against direct experience. A customer may score a service highly while feeling pressured to do so; a test may rise while genuine understanding does not. Numbers are evidence about reality, not substitutes for it.
Efficiency plus pressure and connection can turn broad gains into winner-take-most
Outcomes can compound. A Gaussian distribution, often called a bell curve, clusters results around a middle, with fewer cases at either extreme. A Pareto distribution has a long tail: a small number of participants capture a very large share. In connected systems, an early advantage can attract more resources and attention, which creates an even larger advantage. Martin calls this dynamic preferential attachment.
The economy shows the shift. The richest 1 percent of Americans own nearly 40 percent of the country’s wealth, while the bottom 90 percent own about 23 percent. Martin links this concentration to pressure for ever-greater efficiency and cheaper connections between participants. Waste Management’s acquisitions illustrate how scale can reinforce itself: greater purchasing power and centralized repairs made the company more efficient, helping it acquire still more competitors.
Preferential attachment helps explain why competition does not always produce a stable spread of winners and losers. In platforms, fame, or industries with strong scale economies, the reward for being ahead can itself become a cause of further advantage. The ecological comparison is useful: a uniform system may perform well under ordinary conditions yet become exposed to a shared shock. But concentration is not automatically evidence of social harm, and not all industries behave like platforms or garbage collection. The key question is whether accumulated power blocks entry, weakens bargaining, or makes failure more damaging.
Build resilience with friction, regular revision, and deliberate separation
Balance beats extremes. Resilience is a system’s capacity to adapt to changing conditions while continuing to function. Martin argues that it must balance efficiency, rather than simply replace it. His design principles pair pressure with friction, the pursuit of perfection with ongoing improvement, and connection with separation. Productive friction is a constraint that slows risky optimization enough to preserve the system’s ability to recover.
Small limits can protect capacity. Baseball teams limit pitchers’ workloads because the final pitches in an overlong outing may raise injury risk and cost the team a valuable player. Stock-market circuit breakers pause trading after sharp declines, interrupting a feedback loop in which falling prices can trigger more selling. These constraints reduce efficiency in the moment, but can help prevent a larger breakdown. Since adaptive systems change, rules need continued review and adjustment.
The central insight is that slack and safeguards can function like insurance: their value is most visible when conditions turn bad. Resilience research in ecology and engineering likewise emphasizes diversity, buffers, and the ability to reorganize after disruption. Such protections do carry costs. Excessive redundancy can make a system expensive, slow, or resistant to useful change. The challenge is therefore not to add friction everywhere, but to identify where one failure could cascade, and where a modest pause or buffer protects essential capacity. Regular review matters because yesterday’s safeguard can become tomorrow’s obstacle.
Give workers and customers enough slack to make the whole business work
Optimize the experience, not a silo. Joe’s Stone Crab in Miami Beach has thrived for more than a century while closing for part of the year and keeping a low-priced chicken entrée that financial staff considered unprofitable. The owners see these choices as part of a broader system: employees, customers, suppliers, and the community all affect the restaurant’s long-term success. Staff stay far longer than the industry norm, and the restaurant makes room for a range of customers.
Slack can pay. Retailers such as Costco, Trader Joe’s, and QuikTrip pay more, train more, and schedule more staff than a bare-minimum model recommends. Research discussed in the book finds that these practices can support higher sales and profitability through better service. Southwest Airlines similarly balances cost with customer and employee satisfaction, rather than allowing one measure to dominate every decision.
The business cases challenge the assumption that low labor cost is equivalent to efficiency. In service work, knowledgeable employees and available time can prevent errors, improve trust, and make repeat business more likely. This connects to the idea of high-quality jobs: workers are not merely a cost input but part of the organization’s capacity to learn and serve. The caution is that Joe’s is a distinctive, high-demand restaurant, not a template that transfers unchanged to every firm. The generalizable lesson is to measure the whole service system, including the costs created by understaffing and turnover.
Write policy as a revisable prototype, and design it near real people
Rules need maintenance. Canada’s Bank Act has required periodic review since 1871, allowing regulators to revise it rather than treat one legislative design as permanent. During the global financial crisis, Canada’s large banks avoided the US-style bailout. Martin points to a broader regulatory approach that included one principal financial supervisor, regular contact with bank leaders, and resistance to mergers that would have concentrated much of the market in two banks.
Observe before prescribing. The UK’s government digital service studied how citizens actually used government websites. Researchers found that people searching for a stolen passport might type “lost,” so a system with separate pathways could send them in the wrong direction. Combining the entry point made the service more usable. The same principle applies to economic policy: test assumptions with the people affected, then revise rules as behavior and conditions change.
Policy design benefits from the same user-centered discipline used in good product development. A rule can be logically coherent yet fail because its designers misunderstand how citizens, firms, or officials will interpret it. Scheduled review creates a formal opportunity to learn, while direct observation can reveal problems that aggregate statistics miss. The Canadian banking comparison is suggestive, not definitive proof that its regulatory structure alone explains different crisis outcomes. Outcomes reflect many interacting factors. Its strongest lesson is institutional: build processes that notice emerging problems and make adjustment ordinary rather than politically exceptional.
Teach students to test models, combine perspectives, and learn from failure
Learning is model-building. Beth Grosso, a teacher in Hamilton, Ontario, hid paper clues around her classroom. Student groups found different clues and formed different explanations for why she had arrived at school in a bad mood. The exercise showed that conclusions depend on which evidence people encounter and how they interpret it. Grosso used the ladder of inference, a framework describing how people select data, interpret it, and build conclusions from it.
Failure can improve the model. Grosso’s students proposed a new way to support younger English-language learners. She encouraged them to test the plan with the people it was meant to help. When the first idea failed, they revised it instead of treating the setback as proof they could not solve the problem. Martin argues that education should reward thoughtful inquiry, integration across subjects, and the ability to revise conclusions, not only certainty and narrow correct answers.
This approach develops intellectual humility without asking students to give up judgment. It resembles scientific practice: hypotheses become useful when they are exposed to evidence that could change them. The classroom fish exercise also models perspective-taking, since disagreement may reflect different information rather than bad faith. There is a real constraint: schools must teach established knowledge as well as uncertainty, and not every question has multiple equally sound answers. Strong teaching distinguishes settled evidence from open questions, then gives students practice reasoning appropriately in each case.
Use buying, organizing, and voting to add friction where institutions won't
Citizens have leverage. When Heinz closed its Leamington, Ontario, plant, Canadian tomato growers and workers lost a major buyer. Consumers rallied behind French’s ketchup, which used Canadian tomatoes. Grocery chain Loblaws initially dropped the product, then restored it after public pressure. This buycott, a coordinated choice to reward a company, helped an alternative brand gain market share and gave consumers a way to influence the market.
Participation can reshape rules. Citizens can support alternatives across providers, a practice called multihoming, rather than relying on one dominant platform or seller. They can organize around shared concerns, as Mothers Against Drunk Driving did, or take part in participatory budgeting, where residents propose and vote on local projects. Campaigns for independent redistricting commissions and automatic voter registration offer further ways to make political participation matter.
The examples show that citizenship includes more than casting an occasional ballot. Purchasing choices, local organizing, and participatory budgeting can create counterweights when large organizations or political institutions have little incentive to change. Collective action is especially powerful when people share a clear goal and can sustain involvement. Yet consumer action is not a substitute for regulation: choices are constrained by income, access, and information, and individual purchases rarely alter structural rules alone. Civic tactics work best as complements, linking everyday pressure to durable changes in institutions.
Analysis
Martin’s most persuasive contribution is not simply that efficiency can go too far. It is the chain he describes: a metaphor shapes a model, a model selects measures, and measures alter behavior once attached to incentives. That sequence explains why apparently sensible reforms can produce outcomes opposite to their stated aims. The book also joins distribution to political legitimacy. Stagnant median incomes matter not only as an economic result but because people who feel excluded may withdraw consent from the system that produced it.
The natural-system perspective usefully replaces fantasies of a final fix with an iterative stance. Its strongest prescriptions follow from that premise: retain buffers, revise rules, study users, and avoid relying on one indicator. This is a coherent bridge from diagnosis to action, and the examples make the agenda tangible across firms, government, schools, and civic life.
The argument is less conclusive when it assigns broad economic changes to efficiency pressure and connectivity. The shift in income distribution is well documented in the text, but causation is harder to isolate. Bargaining power, tax policy, labor institutions, technology, trade, and political choices may interact, and the book’s own complex-systems framing cautions against single-cause explanations. Likewise, Canada’s banking experience offers a useful comparison but cannot establish that its regulatory arrangements alone prevented crisis.
There is also a productive tension between advocating more friction and avoiding protectionism or needless bureaucracy. Friction can protect resilience, but it can also preserve incumbents, raise prices, or block beneficial innovation. The book’s deeper standard is therefore not efficiency versus inefficiency. It is whether a constraint preserves broad participation and adaptive capacity over time. That question is more demanding than a simple metric, but it is also the book’s most valuable decision test.
Review Summary
When More Is Not Better argues that America's obsession with economic efficiency has created a dangerously unbalanced economy lacking resilience. Martin proposes viewing the economy as a complex adaptive system rather than a machine, and offers solutions to restore balance. He emphasizes the importance of slack, adaptability, and avoiding proxy measures becoming goals. Reviewers praise the book's accessibility, practical solutions, and balanced approach to problem-solving. While some find certain sections less compelling, most agree it provides valuable insights into improving democratic capitalism and addressing inequality.
People Also Read
Glossary
complex adaptive system
interacting, changing networkA system made up of participants whose interactions produce outcomes that cannot be understood simply by adding up each participant’s behavior. Its members adapt to changing conditions and to one another, so cause and effect can be nonlinear and difficult to predict. The economy is treated in the book as such a system, rather than as a machine with independently optimizable parts.
surrogation
measure replaces the goalThe process in which a proxy used to measure a desired outcome comes to be treated as the outcome itself. For example, a company may use accounts per customer to estimate customer loyalty, then focus on opening accounts rather than earning trust. Surrogation is dangerous because incentives tied to a measure can encourage behavior that improves the number while damaging the real goal.
Pareto distribution
concentrated, long-tail outcomesA distribution in which a relatively small number of participants receive a large share of an outcome, while many others receive much less. Unlike a bell-shaped distribution, its long tail can extend far toward extreme high values. In the book, connected advantages can reinforce themselves, helping economic rewards and market power become more concentrated.
Gaussian distribution
bell-shaped distributionA symmetrical distribution in which most observations cluster near the middle and fewer appear toward either extreme. The mean and median are typically near the center. Martin uses this familiar bell-shaped pattern as a contrast to concentrated economic outcomes, while noting that assuming economic rewards follow it can obscure how advantages compound.
resilience
capacity to adapt and functionThe capacity of a system to adjust as its context changes while continuing to function and deliver its intended benefits. The book treats resilience as a necessary partner to efficiency. Buffers, alternative suppliers, limits on risky behavior, and separations that prevent cascading failure can preserve resilience, even when they reduce short-term efficiency.
productive friction
constraint that protects capacityA deliberate limit, pause, or cost that moderates the pursuit of efficiency when unrestrained pressure could damage a system. Examples include limits on a pitcher’s workload or a pause in stock trading after a sharp decline. Productive friction is not the rejection of efficiency; it is a way to preserve safety, adaptability, and continued operation.
ladder of inference
steps from evidence to conclusionA framework associated with Chris Argyris that describes how people move from selecting some available information to interpreting it, drawing conclusions, and acting on those conclusions. Because people may choose different evidence or interpret the same evidence differently, they can form different accounts of one event. Examining the steps can make reasoning more conscious and open to revision.
multihoming
using multiple providersA consumer practice of dividing use across competing providers rather than relying entirely on one. A person might use more than one ride service, news source, or retailer. By avoiding complete dependence on a dominant provider, consumers can reduce the self-reinforcing advantage that comes when popularity and scale attract still more users.
buycott
buying to reward a companyA consumer action that supports a company or product because its behavior is valued, rather than withholding purchases to punish another company. In the Canadian ketchup example, consumers favored French’s after it used Canadian tomatoes following Heinz’s closure of its Leamington plant. A buycott can help an alternative supplier gain enough support to become a credible competitor.
Download PDF
Download EPUB
.epub digital book format is ideal for reading ebooks on phones, tablets, and e-readers.