Mr Gates’s proposal would solve several problems at once. In addition to raising money (which could be used to fund new employment opportunities for people) the tax would probably slow the pace of automation: a good thing, from Mr Gates’s perspective. Economists typically dislike taxes on such investments, since buying and using new equipment raises productivity and growth. But if the pace of automation is too rapid for society to handle, as Mr Gates supposes, then slowing automation could do more good than harm: by prolonging employment for workers who might otherwise fall into long-term unemployment, for instance.

There are reasons to be sceptical of this approach, however. Not all new robots displace human labour; some make existing workers more productive, and deterring such investments would leave workers worse off. Automation can also reduce consumer costs. A robot tax which reduced the use of machines in health care and therefore kept medical costs growing rapidly might hurt as many workers as it helped. Mr Gates’s plan also seems premature; for now, at least, productivity growth remains disappointing, suggesting that automation is occurring too slowly rather than too rapidly. The last thing most economies need is slower economic growth. But the biggest problem with an attempt to squeeze money out of robots is that the machines are not the big winners of the digital era. The gains from growth have flowed to landowners and the dominant firms whose profits have accounted for a growing share of economic output. A tax on land would be an efficient way to capture a share of the benefits of economic growth without discouraging productive investment. And in an economy dominated by “superstar” firms, in which the top company in each industry tends to earn hefty profits, progressive income or wealth taxes might raise lots of money without much harming growth: provided that rich people cannot easily shield their lucre in tax havens.

Of course, tax might not be the best way to manage the massive profits of the superstar firms of the digital economy. Big profits are a sign that companies wield market power. That power might stem from network effects (the value, in a networked world, of being on the same platform as everyone else), or from the competitive advantage a company receives from control over massive amounts of proprietary data, or from government protections or anti-competitive behaviour. To shrink corporate profits and redirect more of the benefits of new technologies to individuals, governments might need to take a page from the early 20th century and break up or regulate hulking digital monopolies.