Key Takeaways
Prepare for a labor market where machine complements soar and substitutes sink.
The split is structural. Cowen argues that machine intelligence, globalization, and uneven innovation divide workers into those whose skills amplify computers and those whose tasks computers can replace. In the book's 2013-era figures, young college graduates' real wages had fallen by more than 5 percent since 2000, while advanced-degree holders were among the few education groups whose wages rose. The winners are not only programmers. Sales, design, finance, management, and applied judgment can all become more valuable when they make technology more productive.
The middle thins out. Cowen calls this labor market polarization: employment and rewards shift toward high-skill and low-skill work as many middle-skill jobs shrink. His phrase "average is over" describes widening outcomes, not a guaranteed fate for any individual. The same forces can create new work while making the transition painful for people whose old jobs disappear.
What's striking is the emphasis on the relationship between a worker and a tool, rather than on technology as a simple job destroyer. This resembles the economic idea of complementarity: a tool raises the value of some inputs while reducing demand for others. The distinction helps explain why learning to use software can matter beyond technical occupations. Still, the split is not purely a measure of talent. Access to training, reliable institutions, and the ability to change careers shape who can become a complement. Treating divergent outcomes as individual merit alone risks mistaking unequal opportunity for unequal effort.
Learn to steer intelligent tools, not merely compete against their speed.
Chess becomes a workplace model. In Freestyle chess, competitors may use chess programs and consult other people, but humans still decide which moves to play. In a 2005 tournament, ZackS, a team of two players with modest traditional chess ratings, beat teams that included grandmasters. Their edge came from handling several computer programs effectively, not from being the strongest solo players.
Orchestrate, then judge. The useful human contribution is knowing when to trust a machine, where its analysis may be incomplete, and how to convert its output into a decision. Cowen points to medical screening as a parallel: software can flag suspicious images for specialists, while people provide context and catch errors. A cited study found that Google supplied the correct diagnosis in 58 percent of 26 cases, showing that access to information is not equivalent to dependable expertise.
The chess example makes an important distinction between using a tool and understanding how to use it. In human-factors research, automation bias describes the tendency to accept a system's recommendation simply because it came from a system. Skilled collaboration requires the opposite capacity: calibrated trust, including the confidence to override a tool for a reason. The model is powerful where human judgment can add context, but it is not a universal promise of employment. If a system becomes strong enough, or a human cannot improve its output, collaboration may give way to substitution.
Make reliability and coordination part of your scarce professional skill.
Teams make small failures expensive. As production stretches across specialized workers and technology, a single mistake can disrupt a much larger process. Managers become valuable because they find capable people, coordinate their work, and protect the quality of the team. Cowen illustrates the point with military drones: keeping one Predator in the air for a day required about 168 supporting workers, far more than simply putting a pilot in a cockpit.
Dependability has economic value. In coordinated work, conscientiousness means following through, learning the tools, and not undermining colleagues' efforts. Cowen also identifies marketing as a growing function, including services that improve a customer's experience and compete for the attention of high earners. These are not merely technical jobs. They blend communication, reliability, and an understanding of what people value.
This account usefully broadens the idea of skill beyond credentials and technical fluency. Coordination is itself productive work, especially when specialized tasks depend on one another. Organizational research also suggests that team performance can be limited by weak links, not just raised by exceptional individuals. Yet conscientiousness should not become a polite label for obedience. Reliable execution matters, but good teams also need dissent, error reporting, and room to question a bad plan. The strongest version of Cowen's point is not that workers should never resist, but that cooperation and competence must coexist.
Treat recessions as accelerators of structural job loss, not temporary pauses.
A downturn can reveal what was already changing. Cowen argues that the 2008-2009 recession did more than reduce spending. Employers selectively cut workers they considered less productive, and many of those jobs did not return when the economy began growing again. In the three quarters when unemployment rose most sharply in 2009, measured output per labor hour increased by 8 percent, 7 percent, and 5.3 percent. Cowen interprets this as evidence that firms had removed workers whose pay exceeded the value employers believed they produced.
The recovery changed the job mix. In the figures he cites, about 60 percent of recession job losses came from mid-wage occupations, while 73 percent of jobs added after the recession were in lower-wage occupations. He calls some workers "zero marginal product" to describe employers' view that hiring them adds little value, even at lower pay. The phrase concerns a market judgment, not a person's worth.
The recession-as-accelerator idea helps explain why employment can recover without restoring the old kinds of jobs. It also highlights a measurement trap: average productivity can rise when lower-paid or lower-output workers are dismissed, even if no remaining worker becomes more productive. That statistic alone cannot show why people lost jobs or whether employers assessed their potential fairly. The distinction matters for policy. Training, job creation, and income support address different problems, and a diagnosis that treats structural displacement as individual failure may leave workers with no credible route back.
Build education around practice, feedback, and motivation, not credentials alone.
Digital instruction can spread good teaching cheaply. Cowen points to online courses, Khan Academy videos, and chess software that lets learners practice, receive feedback, and review mistakes whenever they choose. At the time of writing, about 250,000 K-12 students attended full-time virtual schools, and more than two million took at least one online class. Virginia Tech's Emporium model used a large computer-based learning space with instructors available to help students, rather than relying only on conventional lectures.
Motivation remains the human bottleneck. Games can teach complex rules effectively to people who want to master them, but access alone does not make an uninterested learner engage. Cowen sees teachers shifting toward coaching, feedback, and encouragement, while students who need structure may benefit from demanding programs such as KIPP. Education can become cheaper and more flexible, but self-direction determines who gains most.
The chapter's strongest educational insight is that content delivery and learning support are different jobs. Software can provide repeated practice and immediate feedback at scale, while people help learners persist, interpret setbacks, and set goals. That division also offers a better test of educational reform than asking whether a screen can replace a classroom. The important question is which learners benefit under which conditions. Low-cost online access can widen opportunity, but it can also reward students who already have time, support, and self-discipline. Good design must account for that uneven starting point.
Use machine advice to test intuition, while retaining judgment over context.
Even experts miss better options. Computer chess engines routinely find moves that seem ugly or risky to human players, yet prove stronger after deeper analysis. Researcher Ken Regan compared human games with computer recommendations and found that even top grandmasters matched the engine's preferred move only about 55 percent of the time. The point is not that human judgment is useless. It is that confident intuition can overlook possibilities, especially in complicated situations.
Algorithms can widen a person's choices. Online dating services sometimes recommend matches users would not have selected on their own. Cowen describes Cambry and Karrah, who met through Match.com despite not fitting each other's stated preferences; their first date went poorly, but they continued seeing one another. A recommendation can interrupt habitual filtering, but a match score cannot establish chemistry or long-term compatibility.
This is a case for using computation as a corrective to confidence, not as a replacement for judgment. Decision science often distinguishes between calibration, knowing how accurate one's beliefs are, and raw confidence. Comparing predictions with outcomes can improve calibration in domains where feedback is available. Romance and career choices are less tidy: the desired outcome is partly subjective, and preferences can change through experience. Algorithms are most useful when they surface overlooked options or expose an assumption. They become risky when their measurable proxy is mistaken for the whole human goal.
Expect convenience to arrive bundled with self-service and constant evaluation.
Customers increasingly do part of the job. Cowen's frustrating call to a cable company illustrates how automated menus, PINs, and repeated data entry move work from employees to customers. Self-checkout and ATMs make service cheaper or faster when they work, but they also leave customers to solve problems that a worker once handled. GPS offers a similar trade: it can reveal a better route, yet stumble in ambiguous settings such as complicated intersections.
People become measurable too. Cowen predicts more machine-assisted ratings of professionals, from lawyers to doctors, as a way to help customers compare quality. Such ratings could make expertise easier to find, but they are incomplete and may change behavior: professionals could avoid difficult cases, while customers might dismiss sound advice from someone with a mediocre score. Greater legibility can improve accountability while narrowing privacy and discretion.
The examples anticipate a central issue in service design: efficiency for the provider can mean transferred effort for the user. A system's average performance may improve while a person with an unusual need faces a dead end. Ratings create a related problem. Once a measure affects rewards, people have incentives to optimize the score, sometimes at the expense of the underlying service. The promise of machine evaluation is better information, not perfect neutrality. Human appeal, explanation, and the option to reach a person are not decorative extras when an automated decision affects health, work, or access.
Follow talent clusters, and treat immigration as a way to strengthen them.
Location still matters in a digital economy. Cowen argues that educated workers and innovative firms increasingly gather in a limited number of cities, where proximity helps people learn from one another and connect to employers. The internet's wage benefits were concentrated in counties containing about 42 percent of the US population, according to a study he cites. Regions with more college graduates tended to attract more ambitious workers, while less prosperous areas struggled to catch up.
Trade is a pressure, not the whole explanation. Cowen notes research linking Chinese import competition to weaker manufacturing employment and wages in exposed US regions. But he argues that technological change and weak domestic skills are more fundamental causes of broad wage stagnation. He also contends that immigration can keep related businesses and support jobs at home, strengthening North American economic clusters rather than simply competing with existing workers.
The argument connects labor economics to agglomeration, the advantage firms and workers gain by locating near one another. Knowledge moves through formal systems and informal contact, so online access does not erase the value of place. The immigration claim is especially useful as a corrective to treating offshoring and immigration as identical threats. Still, cluster growth can benefit a region while raising housing costs and pushing out workers who make that region function. A strategy that attracts talent without addressing housing and mobility can deepen the same geographic inequality it seeks to solve.
Prepare for science whose predictions work before its explanations make sense.
Knowledge is becoming harder to hold in one mind. Scientific fields rely on deeper specialization, larger research teams, and more complex tools. Cowen notes that the average age at which Nobel-winning physicists made their prize-winning discoveries rose from 37 in 1905 to 50 in 1985, which he links to the growing preparation required to reach the frontier. Even mathematical proofs can depend on so many separate results that specialists must assess different parts before the whole is accepted.
Machines may deepen the divide. Data analysis can reveal reliable patterns without producing a theory that people can easily explain. Cowen sees economics moving toward stronger empirical work and larger datasets, while machine-assisted research could make scientific results even less intuitive. He also points to citizen science, where members of the public contribute observations or computing resources without needing to master the full theory.
The distinction between knowing that a model predicts well and knowing why it predicts well is increasingly important across science and public policy. Prediction can be valuable even when explanation is incomplete, but it does not settle questions of causation, scope, or acceptable risk. This creates a new responsibility for institutions: communicate uncertainty and test performance outside the conditions where a system was built. Citizen participation may broaden the evidence base, but participation is not the same as shared authority. Trust depends on who chooses the questions and who can challenge the results.
Plan for inequality to reshape where people live, learn, and receive care.
The social contract faces a fiscal squeeze. Cowen expects aging and rising health costs to pressure public budgets, while political resistance makes deep cuts to benefits for older voters difficult. He predicts that more of the adjustment will fall on poorer people, workers, and communities through tighter eligibility, employment costs, and unequal access to services. He also expected that about 19 percent of the US population would be 65 or older by 2030.
Housing becomes an economic adjustment. Cowen points to migration into Texas, where more than 147,000 people arrived through migration in 2008-2009, and contrasts Houston's relatively low home prices with Brooklyn's much higher ones. He argues that some households will trade public services or proximity to opportunity for cheaper housing and more cash. His broader forecast is a society with greater material inequality, alongside abundant low-cost entertainment and education.
Cowen's scenario is valuable because it treats housing, healthcare, and work as connected parts of household security. If wages stagnate and public costs rise, cheaper places can function as a private safety valve. But this is not a neutral adjustment: moving may separate people from jobs, care networks, and schools, and low prices can reflect limited public investment. Migration patterns also reveal constraints and trade-offs, not a single national preference. The political question is therefore not only how people adapt to inequality, but which costs society chooses to make private and which it chooses to share.
Analysis
Cowen's central achievement is to replace the blunt question, "Will machines take our jobs?" with a more useful one: Which human abilities become more valuable when paired with machines, and which become redundant? His chess examples make that distinction vivid. The best human-machine teams are not simply humans obeying software. Their advantage comes from knowing what the software can do, spotting where its output needs interpretation, and making decisions under pressure. This model gives the book a coherent center across work, education, science, and daily life.
The book is also strongest when it connects technical change to institutions. A worker's prospects depend not just on aptitude, but on hiring costs, management capacity, education, housing, and the geography of opportunity. Its less comfortable insight is that productivity can rise while broad employment and wages deteriorate. An economy can become better at producing value without distributing that value to the people displaced by its methods.
The analysis becomes less secure when a structural tendency turns into a forecast about character or politics. Cowen's claims about who is conscientious, who can adapt, and how a more unequal society will behave are plausible hypotheses, not laws. Performance measures also reflect the systems that produce them: a low score may indicate weak ability, poor training, unsuitable evaluation, or a constrained opportunity to demonstrate skill. The book recognizes some of these measurement limits, but its meritocratic frame sometimes risks making market outcomes sound more deserved than they are.
Its lasting contribution is therefore not a precise prediction of which jobs survive. It is a practical lens: look for complementarity, follow where coordination and judgment matter, and ask who bears the costs when systems become more efficient. The future described here is neither a simple machine takeover nor a smooth rise in prosperity. It is a distributional struggle shaped by technology, human adaptation, and choices about which forms of value society rewards.
Review Summary
Average is Over receives mixed reviews, with praise for its thought-provoking ideas about technology's impact on the economy and job market. Cowen's predictions about growing inequality and the importance of working with intelligent machines are seen as insightful by some. However, critics find the book's heavy focus on chess analogies tedious and argue that Cowen's arguments lack sufficient evidence. Some reviewers appreciate the book's engaging writing style, while others find it superficial and overly speculative.
People Also Read
Glossary
Freestyle chess
Chess with computer collaborationA form of chess competition in which players may consult computer programs, databases, and other people while making their moves. Success depends not only on chess knowledge but on rapidly interpreting machine analysis, recognizing its limits, and choosing a move under time pressure. Cowen uses it as a model for human work alongside intelligent machines.
meta-rational
Knowing when to deferCowen uses this term for the ability to reason about the reliability of one's own reasoning relative to a tool or expert. A meta-rational person generally defers when a computer is stronger, but can identify situations where its assumptions, search limits, or evaluation may be misleading. It is judgment about when to trust judgment.
zero marginal product
Little added value to employersCowen applies this economic phrase to workers whom employers believe add little or no value to a business, even if they accept lower wages. The argument is that some applicants are not hired because integrating, supervising, or correcting their work costs more than it contributes. The term describes an employer's assessment, not a measure of a person's inherent worth.
labor market polarization
Jobs shifting toward both endsA pattern in which employment grows more at high-wage and low-wage ends of the labor market while many middle-skill occupations decline. Cowen links the pattern to technology, organizational change, and global competition. It describes a changing distribution of jobs and pay, not the disappearance of every middle-class worker.
Generation Limbo
Young adults delaying career stabilityCowen's label for young people whose entry into stable adult work is delayed by weak job prospects. They may live with family longer and rely on part-time work, freelancing, or service jobs rather than beginning a conventional career. The term captures economic uncertainty and delayed independence, while recognizing that some young adults value the flexibility.
threshold earners
People content with getting byCowen's term for workers who aim to earn enough to meet their needs but do not continually pursue higher wages, credentials, or career advancement. He associates the choice with the availability of relatively affordable ways to live and enjoy leisure. It describes an economic orientation, not necessarily dissatisfaction or lack of ability.
hyper-meritocracy
Precisely measured competitionCowen's description of a labor market in which firms can measure and compare workers' performance with increasing precision. Better measurement can reward people who produce scarce, valuable results, but it can also make employment more demanding and exclusionary. The term is deliberately ambivalent: greater productivity does not automatically mean a fairer or happier society.
Emporium model
Computer-guided learning spaceAn educational approach in which students work through computer-based instruction in a shared learning space, with instructors available to answer questions and provide help. Cowen describes Virginia Tech's large, flexible mathematics learning center as an example. The model combines scalable digital lessons with human support rather than relying solely on lectures.
citizen science
Public participation in researchScientific work that draws on contributions from people outside professional research teams. Participants may collect observations, share personal data, or lend computing power to a project without fully understanding every theory being tested. Cowen presents it as a way for the public to contribute to scientific knowledge even as research becomes more specialized.
factor price equalization
Trade pressures wages toward similarityAn economic idea that trade and investment can push the prices of productive inputs, including labor, closer together across countries. When workers in one location are much cheaper and sufficiently productive, firms may move capital and production there rather than move workers across borders. Cowen uses the idea to explain how outsourcing can place downward pressure on some domestic wages.
FAQ
What's Average Is Over about?
- Economic Changes: Average Is Over by Tyler Cowen explores the shifting economic landscape in America, focusing on how intelligent machines and globalization are transforming job markets and income distribution.
- Hyper-Meritocracy: The book introduces the concept of a "hyper-meritocracy," where success is increasingly tied to one's ability to work with intelligent machines and leverage unique skills.
- Winners and Losers: Cowen highlights the divide between those who will thrive in this new economy and those who will struggle, leading to a widening gap between high earners and those with less marketable skills.
Why should I read Average Is Over?
- Future Work Insights: The book provides valuable insights into the future of work and the skills needed to succeed in an automated world.
- Economic Predictions: Cowen offers a compelling analysis of economic trends, essential for understanding the implications of technology on employment and wages.
- Practical Advice: It contains practical advice on adapting to these changes, making it relevant for students, professionals, and policymakers.
What are the key takeaways of Average Is Over?
- Skill Complementarity: Individuals who can complement intelligent machines will have better job prospects and higher wages, while those with easily replaceable skills will struggle.
- Education Reform: Cowen advocates for rethinking education systems to prepare individuals for the new economy, emphasizing critical thinking and adaptability.
- Inequality Trends: The book highlights growing income inequality, suggesting a future with a greater divide between high earners and those in stagnant sectors.
What are the best quotes from Average Is Over and what do they mean?
- "Average is over.": This encapsulates the book's thesis that mediocrity will no longer suffice in the job market; only those who excel will thrive.
- "The key questions will be: Are you good at working with intelligent machines or not?": Emphasizes the importance of adaptability and skill in working alongside technology for future employment.
- "Lacking the right training means being shut out of opportunities like never before.": Highlights the necessity of continuous learning and skill development in a rapidly changing economic landscape.
How does Tyler Cowen explain the relationship between technology and job markets in Average Is Over?
- Technological Displacement: Intelligent machines are displacing traditional jobs, especially those involving routine tasks that can be automated.
- Creation of New Jobs: While some jobs are lost, new opportunities will emerge in sectors requiring human creativity, emotional intelligence, and complex problem-solving.
- Skill Mismatch: There is a growing mismatch between the skills workers possess and those needed in the new economy, emphasizing the need for retraining and education.
What is the hyper-meritocracy described in Average Is Over?
- Definition: Hyper-meritocracy is a system where success is determined by one's ability to leverage technology and specialized skills.
- Impact on Employment: Individuals who can effectively collaborate with intelligent machines will rise to the top, while those without such skills will face challenges.
- Societal Implications: This leads to greater income inequality, as economic growth rewards those with the right skills and education.
How does Average Is Over address income inequality?
- Growing Divide: Cowen points out that income inequality is likely to increase as the economy splits into high earners and low-wage workers in stagnant sectors.
- Impact of Globalization: Globalization exacerbates this divide, with workers in developing countries competing for jobs once secure in the U.S.
- Social Implications: The book raises concerns about the social consequences of rising inequality, including potential unrest and a decline in social mobility.
What role does education play in the future economy according to Average Is Over?
- Need for Reform: Cowen advocates for significant reforms in education to align with the skills required in a technology-driven economy.
- Focus on Critical Skills: Emphasizes teaching critical thinking, adaptability, and technical skills that complement intelligent machines.
- Lifelong Learning: Promotes the idea of lifelong learning, suggesting individuals must continuously update their skills to remain competitive.
How does Average Is Over explain wage stagnation?
- Historical Context: Cowen provides a historical overview of wage stagnation in the U.S. since the 1970s, attributing it to globalization and technological advancements.
- Impact of Outsourcing: Outsourcing has contributed to wage stagnation by shifting lower-value jobs to countries with cheaper labor.
- Technological Displacement: Technology has had a more significant impact on wages than foreign competition, as machines replace human labor in many sectors.
How does Average Is Over discuss the concept of meritocracy?
- Meritocracy and Income Inequality: Meritocracy will become increasingly important in a world of rising income inequality, with self-motivated individuals finding it easier to succeed.
- Challenges for the Underprivileged: While some will rise through meritocracy, many will struggle due to systemic barriers.
- Social Implications: Emphasis on meritocracy may lead to a lack of empathy for those who do not succeed, with the wealthy class gaining more influence.
What does Average Is Over predict about the future of social contracts?
- Redefining the Social Contract: The social contract will need to be rewritten in response to economic changes, particularly wage stagnation and technological displacement.
- Increased Individual Responsibility: Individuals will be expected to take more responsibility for their economic well-being, leading to a shift in societal values.
- Emergence of Two Nations: Cowen foresees a future where society is divided into the wealthy elite and the struggling majority, highlighting potential for increased inequality.
How does Average Is Over address the impact of technology on jobs?
- Job Displacement: Intelligent machines are capable of performing tasks traditionally done by humans, leading to job displacement in various sectors.
- New Skill Requirements: Workers need to develop new skills that complement technology, such as creativity and emotional intelligence.
- Polarization of Employment: Cowen predicts a polarized job market where high-skill jobs flourish while low-skill jobs diminish.
Download PDF
Download EPUB
.epub digital book format is ideal for reading ebooks on phones, tablets, and e-readers.