Still dreaming of hoverboards

Slowing productivity and innovation growth are a sign of corporate decline

Growing up in the 1970s, my friends and I really believed that flying cars, fusion reactors, domestic robots, and colonies on Mars would be a part of our future. This wasn’t just because we were naive readers of comic books and avid watchers of The Jetsons and Space: 1999. The inevitability of ever-faster technological progress was also impressed on us from other much more credible sources: the educational TV programs we watched, the New Scientists we read, and even the planetaria we were forced to visit on school trips. Jules Verne had successfully predicted the moon landing, submarines, videophones, and solar panels, and H. G. Wells had successfully predicted the atomic bomb, tanks, lasers, and genetic engineering. It seemed only natural, therefore, that the predictions of Isaac Asmimov and Arthur C. Clarke were probably also going to be right. And yet, how wrong these predictions turned out to be. As Peter Thiel, who also grew up in the 1970s pointed out: we wanted flying cars, but ended up with text messaging. Why were the dreams of my childhood dashed by a slowdown in technological progress?

From the 1940s to the 1970s the global economy witnessed unprecedentedly strong economic growth, driven by the gains in efficiency from the widespread adoption of American mass production. This business cycle was kick-started by the massive US government investments into private companies like General Motors, Ford, Chrysler, and Lockheed to produce the tanks, planes, and guns needed to win the Second World War, and resulted in American economic model of mass production and mass consumption becoming enshrined as global orthodoxy. The cycle started to fizzle out in the mid-1970s as the marginal gains from improved efficiency tailed off in both the US and in Europe. Since then, world growth has been sustained by the expansion of global trade and the reemergence of China, which benefited from deploying the technologies of mass production originally pioneered in the West.

Underlying these declining economic growth rates in the US and in Europe is the slowing pace of meaningful innovation. US productivity growth, an underlying driver of economic growth, has slowed from three percent per year in the 1950s, to two percent in the 1980s, to just one percent today. These productivity declines may sound small, but their impact compounds over time. Had America’s productivity growth remained at the the level it was in the 1950s until the present, the US economy would be almost twice its current size.

Anecdotal evidence of slowing innovation has been accumulating for some time: research productivity has reduced in many fields, papers and patents are less novel than they used to be, and the gap between the year of discovery and the awarding of a Nobel prize has increased, suggesting that today’s contributions just aren’t living up to those of the past.

What is the cause of our slowing rates of productivity and innovation? In a recent paper ‘Are Ideas Getting Harder to Find?’, Nick Bloom and his colleagues at Stanford University observed that the US's decline in productivity growth had occurred even as the total number of scientific researchers had increased. By studying the relationship between US productivity growth and the amount spent on research and development, Bloom estimates that an average researcher’s work today is only ¹/_40th as impactful as it had been in the 1930s. Or, to put it another way, the US would need forty times as many researchers today as it had had in the 1930s just to keep productivity growing at the same pace. As the actual number of researchers today is only 24 times higher than it had been in the 1930s, it is inevitable, he argues, that productivity growth in the US has slowed. Bloom concludes that either today’s researchers are having fewer ideas than their predecessors or the ideas they do have are just less impactful, or, most worryingly, both!

To understand better why the average scientist is producing less impactful work, Michael Park, a PhD student at the University of Minnesota, decided to look for evidence of the slowing rate of innovation in the content of scientific papers and patents. Park reasoned that a paper or patent was more ‘disruptive’ when the subsequent work that cites it was less likely to also cite its predecessors: that is, when future researchers consider the ideas in the paper or patent to have in some way superseded the work that had gone before, making it less relevant. By contrast, he assumed that a paper or patent was more ‘consolidating’ when the subsequent work that cites it was also more likely to cite its predecessors: that is, when future researchers consider that the knowledge on which the paper or patent builds remains just as relevant, even after its publication. By analysing a paper or patent’s citations, Park was able to give it a disruptiveness ranking between +1, where the work was completely ‘disruptive’, to −1, where it was completely ‘consolidating’. He called this measure of a work’s disruptiveness the ‘CD index’.

Looking at the changes in the CD index across forty-five million papers and almost four million patents, Park found that, in every discipline, science and technology papers and patents had indeed become steadily less disruptive, with the disruptiveness of papers declining by ninety-two percent between 1945 and 2010 and the disruptiveness of patents declining by seventy-eight percent between 1980 and 2010 (see graphs above, courtesy of M. Park).

Park wanted to understand the reason for these widespread and precipitous declines. Perhaps, he hypothesised, disruptive ideas were just getting harder to find: that is, perhaps scientists had already picked the low-hanging fruit and were now being forced to climb higher to find the disruptive ideas. He dismissed this idea, however, for two reasons. First, the declines in disruptiveness could be seen across every field and every time period and it seemed unlikely that every field was consuming its low-hanging fruit at a similar rate. And second, even though the proportion of disruptive works has declined, the absolute number of disruptive ideas has remained stable over the period, which suggested that the fundamental issue was not the depletion of a limited stock of ideas but instead something in the way scientific research itself was being done.

The most disruptive papers and patents, Park found, were the ones with the most diverse citations. The declining rate of innovation, he concluded, was, at least in part, due to the increase in scientific specialisation, which had resulted in scientists and inventors using an ever-smaller proportion of the knowledge available to them. While the stock and diversity of scientific knowledge had continued to increase, the effective stock and diversity of knowledge being used by individual scientists had reduced. Park’s work suggests that one of the key causes for our declining rates of innovation is that science is ghettoising into narrow academic silos. Scientists and inventors are using an ever-smaller and ever-more homogeneous share of the global pool of available ideas. Perhaps the expansion in the total stock of knowledge has now made it almost impossible for any scientist to be an expert in more than the narrowest of disciplines, or perhaps the pressure to publish has caused scientists to prioritise the consolidation of existing work rather than to risk forging off in new directions. Either way, whatever its cause, the current trend for scientists to rely on ever-narrower and more homogeneous slices of knowledge is likely benefitting individual careers but it is doing little for scientific progress and innovation more generally.

Making an organisation more innovative is really hard. There is almost no company, university or government today that won’t claim ‘increasing innovation’ as a strategic priority. The problem is, therefore, not one of intent, but rather one of understanding. Organisations consistently struggle to innovate because having spent the last 100 years focusing on increasing efficiency, they have imbibed an organisational logic of specialised and independent silos that stifles innovation.

This bureaucratic logic started in companies around the turn of the nineteenth century led by the likes of Ford, Taylor and Sloan and has since spread to infect government and universities. The increasing inter-penetration of governments, universities and companies has led each party to adopt the language, sensibilities, and functional hierarchies of the corporate world. The resulting focus on efficiency as a goal in its own right has driven an increase in paperwork and a reduction in original research. Universities now too often serve as institutions that provide the elites with the credentials needed to gain access to the upper echelons of our bureaucracies rather than, as had been the case historically, as institutions that expand the boundaries of knowledge.

Our slowing rate of productivity is a sign that the managerial bureaucracies that enabled the mass production paradigm initiated 100 years ago by Taylor, Ford, and Sloan and the machine metaphor that underpins it have run out of steam. The substantial gains in efficiency that it has delivered have slowed, and the hierarchy and standardisation needed to drive further efficiency are stifling our ability to innovate.

To get out of this impasse, we need to start by reimagining what companies are for and how they work.

Graph courtesy of M. Park. Image courtesy of Pampling.