[28 March 2014]
Human societies change continually, but the changes tend to be slow and incremental. Genuinely revolutionary changes—sudden, radical breaks with the status quo—are rare in human history. Deeper changes, which dissolve and re-form the foundations of the status quo, are rarer still. The great shift from nomadism to village agriculture was one; the emergence of cities, and all they entail, was another. The late 18th century substitution of steam power for wind, water, and muscle—the trigger for the Industrial Revolution and the First Machine Age—was a third. We are living, economists Erik Brynjolffson and Andrew McAfee argue, on the verge of a fourth. The Second Machine Age is their book about it.
The First Machine Age began with the general-purpose steam engine, which was developed by James Watt between 1763 and 1775, but took decades longer to mature, spread, and begin to transform society. The chip-based computer, the basis for the Second Machine Age, emerged in the late ‘50s and early ‘60s and likewise took decades to develop and spread. Computers, however, have grown more powerful, more quickly than steam engines—or any other technology in human history. Gordon Moore, the co-founder of Intel, observed in 1965 that the number of transistors that could fit on a computer chip doubled roughly every two years. That pattern—now dubbed “Moore’s Law”—has held true for the last half-century, and key measures of computer performance (such as processor speed) have followed the same pattern of exponential growth.
The practical results of Moore’s Law are staggering. Computers have, as a result of it, improved at a speed unprecedented in the history of technology, radically increasing in power and flexibility while plummeting in size and price. A state-of-the-art Osborne Executive portable computer from 1982, for example, was 500 times larger, 100 times heavier, and ten times more expensive—but only 1/100 as fast—as a state-of-the-art Apple iPhone from 2007. An iPad 2 from 2011, smaller and lighter than an issue of Vanity Fair and as easily slipped into a briefcase, matched the processor speed of a room-sized Cray-2 supercomputer from 1985. It cost well under a thousand dollars. The inflation-adjusted retail price of the Cray-2 was $33 million, room not included.
Moore’s Law has brought us, Brynjolffson and McAfee argue, to a position analogous to Europe’s in the 1820s and ‘30s. The defining machines of the age (steam engines then, chip-based computers now) have reached technological maturity and spread widely through society, enabling those who possess them to do things that would have seemed utterly fantastic only a decade earlier. Europe stood then, and first-world countries stand now, on the verge of an “inflection point”: a moment in history when the curve of technology-driven progress is about to bend sharply upwards. The result, now as then, will be radical, foundational social and economic change. Today’s computers, however, are far more powerful, versatile, and widespread than steam engines were in the 1820s, and the changes coming in the emergent Second Machine Age will thus dwarf the ones that transformed Europe during the First.
The Second Machine Age spends its 260 pages of text (backed by 30 pages of notes) on three interrelated tasks: documenting the imminence of those changes, sketching their likely effects, and suggesting how individuals and societies can deal with them. Chapters 1-6 explore why innovation in computers and other digital technologies has accelerated in the last decade and will probably continue to do so for the foreseeable future. Chapters 7-11 consider the cumulative impact of all this digital innovation on the shape of the economy, focusing on two overarching concepts that the authors dub “bounty” and “spread”. Bounty is the upside of the emerging revolution: the growing number, diversity, and quality of goods made available, at ever-diminishing prices, by technological progress. Spread is the downside: the ever-growing differences in success—broadly defined as “wealth, income, and mobility” (p. 12)—among individuals. Our goal as a society, Brynjolffson and McAfee argue, should be to maximize the bounty while limiting the spread; the final section of the book, Chapters 12-15, contains their recommendations for achieving that goal.
Across all three sections, the authors advance their arguments with an artfully interwoven mixture of conceptual discussion (both economic and technological) and real-world examples that mix the familiar (the autonomous Roomba vacuum cleaner) and the exotic (Baxter, a multi-purpose industrial robot that can be “trained” by a human operator). They present all of it in clear, conversational prose studded with pop culture references, quotes from experts in a dozen different fields, and unexpected juxtapositions of the real and the imaginary. Chapter 2, for example, includes references—all relevant and illuminating in context—to Jeopardy! champion Ken Jennings, cognitive scientist Steven Pinker, and philosopher Michael Polanyi, as well as the Tricorder from Star Trek, the “Dirty Hungarian Phrasebook” sketch from Monty Python’s Flying Circus, and the Babel Fish from Hitchhiker’s Guide to the Galaxy.
The authors excel at explaining complex ideas clearly. They move with ease from purely technological concepts like Moravec’s Paradox (teaching robots complex analytical tasks is easy, but teaching them simple physical activities is hard) to purely economic ones like winner-take-all markets (those in which the product judged to be of highest quality crowds out all lower quality products, even if its relative advantage over them is very slight).
The discussion of technological unemployment—an idea coined by John Maynard Keynes in the ‘30s—shows how the shift from a mechanical to a digital world turned it from a debatable concept into an inescapable side-effect of technological progress. Brynjolffson and McAfee’s explanation of Moore’s Law, and the concept of exponential growth that underlies it, is particularly well done. It uses the familiar analogy of grains of rice on a chessboard (one grain on the first square, two on the second, four on the third, eight on the fourth, and so on), but draws an unfamiliar lesson. The 32nd square, the authors note, holds a (relatively) comprehensible two million grains of rice. It is in “the second half of the chessboard” that the effects of exponential growth rapidly outrun the ability of the human imagination to grasp them.
The same eye for unexpected interpretations and telling examples is evident when they break new interpretive ground. Pointing out the fact that the digital economy is capable of creating value without creating jobs, they contrast Kodak and Instagram, two businesses built on enabling millions of customers to share billions of photographs. Kodak, created in an age when photos were emulsion-on-paper objects, employed over 145,000 people directly (a third of those in its Rochester, New York headquarters alone), and supported thousands more who were part of its network of dealers and distributors. Instagram, created in an era where photographs were collections of digital information, had only15. Later, arguing for educational reform, they draw a striking, suggestive parallel between large government bureaucracies (which the existing system was designed to staff) and vast simulated computers in which humans are the components.
Brynjolffson and McAfee falter only when they shift, in the third section of the book, from analysis to advice and description to prescription. Here, and only here, their scientific approach to the material—building their argument around powerful explanatory concepts, and taking real-world data seriously—fails them. “Politics,” as Bismarck famously said, “is the art of the possible.” So, too, is that subset of politics we call social and economic reform. We have decades worth of evidence, both academic and anecdotal, that partisan political views are essentially impervious to data. In light of that, the authors’ proposed, data-driven policies—offered without regard to the political complexities of implementing them—feel both irrelevant and naïve.
Instituting additional tax brackets at the $1 million and $10 million income levels almost certainly would help to fight “the spread” created by the digital economy. The data suggesting that they would not diminish high earners’ incentive to work probably is solid. Both points are well taken, but of little use without some insight into how to sell such policies in a political climate dominated by anti-tax zealotry, accusations of “class warfare”, and the politics of irrational aspiration dominate campaign and Congressional rhetoric.
Likewise, large numbers of immigrants may well be—and have been shown to be—beneficial to the economy, boosting innovation without driving down wages, but immigration remains one of the most politically and culturally fraught issues in the 21st century United States. An acknowledgement that “we don’t pretend that the policies we advocate here will be easy to adopt” (p. 227) is not a solution. Indeed, it recalls Sidney Harris’ famous cartoon of two scientists contemplating a chalkboard-filling equation that, as one of its central terms, contains the bracketed phrase: “Here, a miracle occurs.”
The Second Machine Age is part of a well-established and well-populated nonfiction genre: the big-picture book about computers and society. Brynjolffson and McAfee till analytical fields adjacent to those worked by Steven Thompson (Future Perfect, 2012), Douglas Rushkoff (Present Shock, 2011), and Clay Shirky (Cognitive Surplus, 2011), among others. The Second Machine Age is part of a techno-optimist thread within the genre that reaches back through Shirky’s Here Comes Everybody (2008) and David Weinberger’s Everything is Miscellaneous (2007) to Esther Dyson’s Release 2.0 (1997) and Nicholas Negroponte’s Being Digital (1995). Vannevar Bush’s 1945 essay ” As We May Think”, which described a device equivalent in function (though not form) to a desktop computer and digitized library, is the distant literary ancestor of them all.
Brynjolffson and McAfee’s work embodies the virtues of the genre: It deals with the biggest of big ideas, describes exhilarating changes that the reader will (implicitly) live to see, and offers of the prototypes of tomorrow’s technology. It also, however, reflects the genre’s penchant for glossing practical details and downplaying the friction that results when new technologies meet established institutions. It is more effective, and more valuable, as a call to action than as a specific blueprint for action. The virtues of The Second Machine Age far outweigh its limitations, however, and it is well worth the time—and the careful attention—of anyone interested in computers, the economy, and their deeply intertwined futures.