Vaclav Smil is a great intellect of our times finding patterns in our technological and societal evolution. He conducts interdisciplinary research connecting energy, environment, population, food production, history of technical innovation, risk assessment, and public policy (1). He is the author of 47 books and 500 research papers on these subjects. He is a Distinguished Professor Emeritus at the University of Manitoba, a Fellow of the Royal Society of Canada, and a Member of the Order of Canada (2).
This book caught my attention after I read about it on Bill Gates’s blog ‘Gates Notes’. Gates claims to have “read all of his 44 books, which cover everything from the role of energy in human life to changes in the Japanese diet… Although sometimes he’s too pessimistic about the upside of new technologies, he’s almost always right … when it comes to the complexities of deploying those technologies in the real world” (3).
As our species evolved, we went through physical, intellectual and behavioural changes closely tied to the invention of tools, which led to farming and weaponry. The earlier simple tools evolved into more complex ones and later into machines. The invention of new materials has been an obvious indicator of civilization’s progress. Another class of inventions consists of new methods of production, operation, and management, which developed into radically new, highly automated ways of mass manufacturing and data collection and processing.
In categorizing innovations, Smil starts with those that were welcomed when they arrived and were rapidly commercialized and launched on a global scale. Eventually, even decades later, they turned out to be so harmful both to humans and to the environment that they were banned outright. An example is the leaded gasoline, which enabled the knock-free running of internal combustion engines. However, the resulting emissions of a neurotoxic heavy metal were widely recognized as an unacceptable trade-off and, starting with the US in 1970, countries began to ban the use of this additive. A ban on DDT widely used as an agent for insect control began shortly afterwards. In 1987, a global agreement outlined the plan for the termination of the use of chlorofluorocarbons, a refrigerant material whose rising atmospheric concentration was linked to the depletion of stratospheric ozone.
The next category of failed inventions includes three important technologies whose initial promise appeared to ensure the eventual domination of their respective market niches: airships for affordable long-distance air transport, nuclear fission for electricity generation, and supersonic aircraft for speedy intercontinental travel. These innovations were commercialized and more or less widely deployed, but it did not take long to realize that they would not reach their anticipated potential. Chronologically, airships were the first practical application to fail, and they did so spectacularly. Hindenburg in flames became one of the most widely seen images of a technological catastrophe. Nuclear fission is a case of missed expectations on a much grander scale in the category of a successful failure.
The third category consists of many highly desirable innovations whose mass-scale commercialization would be truly transformative. Their imminent success has been promised for generations but remains permanently unreachable. The idea of high-speed travel inside tubes with sub-atmospheric air pressure has been around for more than two hundred years. Its recent, highly publicised resurrection under the misleading label of hyperloop offers an excellent opportunity to explain how this generations-old dream still waits for practical, convenient, reliable, and profitable commercialisation.
Suppose the world’s staple grain crops (wheat, rice, corn, sorghum) were able to supply a significant part of their nitrogen demand through symbiosis with nitrogen-fixing bacteria, as leguminous grains such as beans, soybeans, lentils, and peas do. In that case, we would not only increase the global grain harvests but be able to reduce the consumption of synthetic fertilisers, thereby saving a great deal of energy and preventing several categories of environmental pollution.
Another example is the commercial exploitation of thermonuclear fusion for electricity generation, a feat first heralded to be reachable within a decade at the first Atoms for Peace conference in 1955 by its chairman Homi Bhabha. This has been perhaps the most famous and the most publicized example in the category of failing expectations, whose realization seems to always be just beyond the horizon.
Yet another category of inventions dominated their particular sectors for generations before they completely disappeared, or were retained only as marginal curios kept alive by eccentric devotees. The open-hearth furnaces are the best example in the first category. Between the 1870s and the early 1950s all primary steel was made by reducing the carbon level of cast iron in blast furnaces. Then, within a generation, they nearly disappeared in Japan and Europe and lingered a bit in North America. In the transportation sector, Ocean liners dominated intercontinental passenger transport for nearly a century before they disappeared after the introduction of transatlantic jet flights.
Microelectronics has driven the demise or marginal survival of many inventions whose services dominated globally for more than a century. PCs displaced typewriters and smartphones replaced cameras. Music became digital and dematerialized. Records, tapes, and compact discs disappeared with direct digital access marginalising them all.
The final chapter of the book comments on the hype surrounding the reporting of new inventions. Credulous assessments by media as breakthroughs and path breakers have become a practice which raises false hopes and unrequited expectations. The book ends with a brief wish list of much-needed advances whose large-scale adoption would provide long-overdue means to tackle some of our most daunting health, environmental, and economic challenges, from the conquest of malaria to reducing global income disparities. We may succeed in some quests but fail in others.
Smil is sceptical of the belief that as innovations go, our age is unrivalled. Analyzing data from various sectors like agriculture, transportation, and pharmaceuticals, he concludes that the present era is not nearly as innovative as it is generally perceived. To him, there are “unmistakable signs of technical stagnation and slowing advances (3).” This sounds counterintuitive as applications like “generative AI’ and deep learning are galloping forward fast. But to Smil, AI researchers only have managed “to deploy some fairly rudimentary analytical techniques to uncover patterns and pathways that are not so readily discernible by our senses” and produced “impressive achievements on some relatively easy tasks (3).”
Smil believes the period 1867–1914 was the only one of explosive innovation in the past century and a half. Internal combustion engines, electric luminaires, the telephone, cheap methods of steel manufacturing, aluminium smelting, plastics, and early electronic devices were born in this period. It was during this period that great insights in the fields of infectious disease, medicine, agriculture, and nutrition were gleaned. For Smil, the ensuing years have been lacklustre, without inventions that achieve scale and rustle the marketplace.
It appears that the exponential growth in computing power over the past several decades as forecast by Moore’s law has given people a false sense about the growth of innovation in other areas. Smil acknowledges that the growth in microelectronics is unique. However, this led to a belief unfounded reality that everything can grow as fast as computing power.
Smil underplays the achievements of AI. Great achievements in Artificial Intelligence, particularly large language models, have been substantial. AI is now capable of human-like reasoning transcending its ability to provide answers to questions. The transition to ‘smart’ behaviour is imminent. Sam Altmann recently said that “the next transition of AI is to acquire Artificial General Intelligence, which will enable it to perform any task that a human is capable of. This is expected to happen within the next ten years (4).