It's hard to know sometimes which is the greater threat to prosperity — headlong technological progress that's destroying decent jobs and hollowing out the middle class, or fading technological progress that's causing persistently slow growth. With a little ingenuity, you could no doubt combine these ideas and worry that technological progress is both too fast and too slow.

On the other hand, you could consider both sets of arguments and be guardedly optimistic.

One of the main authorities on the diminished prospects for growth is Robert Gordon of Northwestern University. His view is straightforward: By now, he says, all the really valuable inventions have been invented. The amazing surge of progress from the age of steam to the arrival of cheap computers was a historical anomaly (for century after century before, there had been little or no growth), and this blip of momentous ingenuity may be over. Everybody is enchanted by the Internet and related innovations, but these aren't in the same league as electricity and the internal combustion engine — breakthroughs that powered growth for the better part of a century.

If you doubt this, Gordon says, look at the numbers. The productivity benefits of the “new economy” had more or less disappeared by 2004. And to Gordon that makes perfect sense. “Attention in the past decade has focused not on labor-saving innovation, but rather on a succession of entertainment and communication devices that do the same things as we could do before, but now in smaller and more convenient packages.”

In Gordon's view, replacing human labor with machines is the key to growth in productivity and hence in living standards, and this is happening too slowly. The more common fear, of course, is just the opposite — that networked automation is destroying jobs across an ever-widening front of occupations. Yesterday, factory workers; today, office workers; tomorrow, teachers, pilots, doctors, you name it. Before long robots will be doing everything. The capitalist elite will do fine, because they own the machines. How will the rest of us make a living?

On the face of it, one of these views must be wrong.

Gordon is an outstanding economist, and I hesitate to question his reading of the figures, but his dismissal of the Internet and the devices that connect to it is puzzling. Perhaps these inventions are less consequential than electricity — a pretty high bar — but like earlier world-changing innovations, they aren't merely products or services in their own right. They're also platforms for new disruptive innovations. They open new worlds of economic possibility.

Then why don't you see this revolution in the productivity numbers? For one thing, the crash and its unusually prolonged aftermath are clouding the picture. For another, it's still early days. The productivity revolution seeded by electricity took years to gather momentum. The same goes for the impact of other great innovations, often dismissed as curiosities to begin with. In 1876, an executive at Western Union famously wrote that the telephone would never catch on. The Internet revolution is still quite young. As recently as 1999, you probably hadn't heard of Google.

Remember too that output and productivity in the Internet economy aren't easy to quantify. Gross domestic product measures output by what people pay for goods and services (or by what the products cost to make). Much of what people consume over the Internet is free. Google and Facebook reach hundreds of millions of people and deliver enormous value — but what they produce requires relatively little labor and is supplied at no charge. As far as GDP is concerned, this output barely exists. That says more about the defects of GDP than about the shortcomings of digital innovation.

In “The Second Machine Age,” an excellent new book on technology's economic impact, authors Erik Brynjolfsson and Andrew McAfee of the MIT Sloan School of Management emphasize this measurement problem: Something huge is happening, and the figures aren't capturing it. They also argue persuasively that Gordon and other productivity pessimists seriously underestimate the power of innovation still to come. Digital technologies, they argue, are “general purpose technologies” — like steam power and electricity. They're “pervasive, improving over time, and able to spawn new innovations.”

Too much so? That's the other fear: Will there be enough decent jobs in this new world — or will the new technologies just make the rich richer while impoverishing everybody else? Brynjolfsson and McAfee meet this second kind of pessimist halfway. Just like its predecessors, the next phase of innovation will eventually destroy a lot of jobs, they say. But it will create a lot as well. This won't happen all at once. There'll be time to adjust, as economies have in the past. If policymakers rise to the challenge, they can make the adjustment easier and the gains more widely shared.

Much of the needed agenda is familiar: To support growth in the second machine age, the government needs to retool the education system (digital technologies can help), foster a favorable climate for entrepreneurs and new business startups (where most of the new jobs will be), encourage research and development, build modern infrastructure, and liberalize immigration to help meet the demand for skills. In all these areas, the U.S. has advantages it's in danger of surrendering, especially if undue pessimism drives it into a defensive crouch.

To support workers' earnings, Brynjolfsson and McAfee also propose a deliberate shift in taxation from labor to consumption, to economic “bads” such as pollution and congestion, and to high incomes inflated by failures of corporate governance rather than well-directed market forces. They'd like the earned income tax credit — a subsidy for low- wage labor — expanded to more workers and made more generous; in the longer term, they say, it should be recast as a negative income tax along the lines advocated by Milton Friedman. All good ideas, and I'd add to them reforming Social Security to give beneficiaries a stake in the economy's capital stock — privatization, but with guarantees.

In its own right, the second machine age is good news not bad. For the United States, the biggest danger is a floundering system of government that will fail both to help the victims of economic disruption and to reap the fullest benefits. Don't worry about robots, worry about Washington.

Clive Crook is a Bloomberg View columnist.