In this post:
Erik Brynjolfsson & Andrew McAfee. The Second Machine Age: Work, Progress and Prosperity in a Time of Brilliant Technologies. W. W. Norton & Company.
If the vision of the future that Brynjolfsson & McAfee paint in their book is right, what lies ahead is technologically exhilerating. And socially troubling.
The book’s central argument is that we are at an ‘inflection point’ in human history, on the cusp of another industrial revolution (hence the title, the ‘second machine age’). Where the first industrial revolution allowed machines to replace manual labor, technologies that exist today (but have yet to realize their full potential) are soon going to make it possible for machines to replace our mental labor, and yet more of the manual labor that machines can’t yet do. The characteristics of these technologies are treated in the first half of the book and the effects that they will have on the economy in the second half. The book, I should say, is highly readable, very accessible and profoundly interesting. Highly recommended.
The brilliant technologies of the title (artificial intelligence, robotics, 3D printing, sensors, etc.) are characteristically digital, exponentially improving and recombinant. The implications of technologies being digital are far-reaching – digital technologies are more like ideas than physical products. To rephrase Brynjolfsson & McAfee, suffice for the present to say that in the same way that black holes (‘where God divided by zero’ *) have strange physics, digital goods (where, for one thing, marginal costs are close to zero) have strange economics. We’ll cover some of those in another post.
Exponential improvement refers to the rate of improvement in technologies, most famously exemplified in Moore’s Law, a prediction made in 1965 that the number of transistors in an integrated circuit will double every two years, implying that computers (and other devices) get twice as fast every other year (today, the trend seems closer to 18 months, but it has been up to 8 years for a short spurt, leading some to predict its end). Moore’s Law has held up amazingly well and we have seen over fifty years of sustained exponential growth. The same applies to other computing technologies, too, such that an iPhone today can do things that were impossible, say, with the technology that sent people to the moon. In time, we’ll all have our very own Watson**.
Recombination refers to how inventions lead to other inventions, new technologies build on previous ones. This relates to a discussion within sociology and economics of technology, where someone like Robert Gordon thinks of inventions as ‘fruit to be picked’ while Brynjolfsson & McAfee (and most other people, to my mind) see inventions as building on other invention, recombining in every more ways. In Gordon’s view, the more we innovate, the closer we get to running out of innovation and the harder it gets. In the recombinant view, the more we innovate, the more opportunities we get for more innovation.
These three characteristics come together to create technologies with incredible potential: more opportunities for invention, more power for those inventions, and economics that enable far-reaching impact. These technologies, the authors hold, are so general and so close to realizing their full potential that they will change our world in the foreseable future, and change not just the digital world, but also spill over into other fields. Someone like Gordon, as representative of the less-than-optimistic observers, would say that much of the potential of these technologies are being channeled into entertainment and that won’t make us better off. He’s right about that for sure, but Brynjofsson & McAfee would reply that it’s a matter of time before the wider effects kick in. And, I (with Clay Shirky), would say that it’s a matter of perspective: something like the Internet gets used for lots and lots of low-value things (e.g. cute cat videos) and some totally amazing ones (e.g. citizen science, knowledge sharing and collaborative creation). Sure, lots of tech goes into wasteful things, but that shouldn’t distract from the fact that the larger trend is so productive. So, in short, I think there’s good reason to be more convinced by Brynjolfsson & McAfee than by the techno-pessimists. Technology will bring great change, possibly even on an ‘industrial revolution’-scale, and we may in principle be near the end of scarcity.
That said, there is a ‘maybe’, a cause for concern about how the gains from that technology will be shared. To use Brynjolfsson & McAfee’s words: new technology will increase the ‘bounty’, but also increase the ‘spread’ in how we benefit from that bounty.
One dimension of this is that when technology replaces human labor (thus boosting productivity and making us better off, on average), somebody probably loses a job. This is what economists call factor substitution: you substitute capital (e.g. a machine) for labor. Since the industrial revolution, machines have made us all richer, but rendered some forms of labor obsolete (my grandfather worked as a carriage driver). This is creative destruction at work and new jobs also emerge to compensate for technological unemployment (he found other work). The alternative of course is that technology complements labor, such that technological advance leads to demand for more workers. Both could happen during this second machine age, but we don’t know how that will play out and what the labor market consequences will be. It will, I promise, be a recurrent theme on this blog.
The other dimension is who gains. Digital goods have the potential to produce ‘superstar economics’ in a way that physical goods don’t, meaning that the winners in the digital economy will win more and that there will be more losers. In itself, accumulating great wealth from inventing great technologies is probably fine. However, if that wealth persists inter-generationally and ends up contributing to reduced social mobility and reduced use of talent in the economy, that’s not fine at all. Rather, it’s the rising inequality that Gordon talked about as a headwind. This is the discussion about the 1% pulling away from the 99%, set in light of technology, explaining for instance why we are seeing reduced and negative real wage growth.
The book closes with recommendations for how individuals and policy can try to keep the social costs of new technology down while still allowing for technology to progress. For individuals, these include getting good at things that computers are bad at (e.g. creativity, idiosyncratic tasks or human care), and going to college – if you actually intend to learn something (because most folks don’t). For policy, the authors call for improving education (working in education, I’m an optimist about what MOOCs will make possible), supporting start-ups rather than established firms (quoting Schumpeter, “it is not the owner of a stage coach who builds railways”), enabling better job-market matches, supporting science and research, upgrading infrastructure (who doesn’t think so?), reforming immigration for both high-skill*** and low-skill workers, and re-designing tax policy.
These are immensely meaningful and helpful, and provide a nice alternative to the regressive desire to slow down technological change (I’d lump that sentiment right in there with the equally regressive anti-trade, anti-immigration sentiments that sweep through Western politics these days). Whether it will be enough, I don’t quite know. They do leave unanswered the question of what we will do to people already in (or at the margin of) the workforce, who will be replaced. The facile assumption that a technologically unemployed truck driver can be re-educated to be a software developer is simply untenable and relying on the argument that the next generation will be better off is similarly untenable in a world of rapid technological change and increased skill obsolescence. The more far-reaching alternatives that Brynjolfsson & McAfee put forth relate to basic income guarantees or negative income taxes, peer economy companies and some ‘moonshot’ concepts. Very much in the spirit that we don’t know specifically what the future will hold, these are more an invitation to thinking. Thinking and practical experimentation will certainly be called for, irrespective of where we end up.
- The economics of ideas and digital goods. This is a huge topic, but I’ll take a look at some of the basic notions to be aware of.
- Substitution or complementarity… A big part of our technological future will depend on whether machines substitute for or complement human labor. Some really interesting work has come out about over the past few years, so we’ll dip our toes in that (maybe it will be several posts). Update: Links to part 1, part 2 and part 3 in this series.
- … Or subversion or resistance. Adding to the above, it’s worth thinking about what humans do to technologies, too. I’m going to write a bit about subversion of technologies as well as what resistance to them might mean.
* The authors attribute this quote to Einstein, and that can’t possibly be the case. But it’s a great quote and he could have said it, if only he hadn’t died nine years before the term ‘black hole’ was first used.
** If you were sceptical of the proposition that technology is an organizational issue of considerable complexity, read this.
*** I once sat in a doctoral course at MIT, when the professor asked the thirty-strong crowd in the room how many were born outside of the US. Only one (a British citizen, no less) was American born. Worth thinking about, I believe.