How the World Really Works: The Science Behind How We Got Here and Where We're Going

by Vaclav Smil

These are, truly, existential decisions to make—but the conclusion is reasonably clear. If we were to stake longevity (accompanied by healthy and active life) solely on the prevailing diet—which, however important, is but one element of a bigger picture that includes your inherited genes and surrounding environment—then Japanese eating has a slight edge, but an only slightly inferior outcome can be had by eating as they do in Valencia.

The risks of specific activities change over time (driving in the US is now generally much safer than half a century ago, but after 50 years of driving your skills might have deteriorated and you pose a greater risk to yourself and others when you get behind the wheel). And if you want to know if intercontinental flying (which you might do infrequently) is riskier than downhill skiing (which you may have done for many years), you must have a rather accurate comparative yardstick. And how does one compare the risks experienced in different nations—say, driving in the US, being struck by lightning while hiking in the Alps and getting killed by an earthquake in Japan? As it turns out, we can do some remarkably accurate, comparative evaluations of all of these risks.

And the SARS-CoV-2 pandemic elevated these irrational fears to a new level. Humanity’s best hope to end the pandemic was mass-scale vaccination, but long before the first vaccines were approved for distribution, large shares of the population were telling pollsters that they would not get inoculated.[23]

Many people smoke and drive and eat excessively but have reservations about living next to a nuclear power plant, and polling has shown lasting and pervasive distrust of this form of electricity generation despite the fact that it has prevented a large number of air pollution–related deaths that would have been associated with burning fossil fuels (by 2020, nearly three-fifths of the world’s electricity came from fossil fuels, and just 10 percent from nuclear fission).

This is why many researchers have argued that there is no “objective risk” waiting to be measured because our risk perceptions are inherently subjective, dependent on our understanding of specific dangers (familiar vs. new risks) and on cultural circumstances.[27]

Quantifying common risks seems to be a daunting enterprise. How does one compare the risks of dying due to an unusually severe seasonal influenza epidemic to the risk of a mortal injury resulting from occasional weekend kayaking or snowmobiling; or the risk of frequent transpacific flying to the risk of habitual eating of California-grown lettuce that might be repeatedly contaminated with Escherichia coli? And how do we express fatal risks? Per standard number of people (1,000; 1 million) in an affected population? Per unit of hazardous substance, per unit of time exposure, or per unit of ambient concentration?

But the finality of dying provides a universal, ultimate, and incontestably quantifiable numerator that can be used for comparative risk assessment. The simplest and most obvious way to make some revealing comparisons is to use a standard denominator and to compare annual frequencies of causes of death per 100,000 people. When using the US statistics (the latest published detailed breakdown is for 2017) this leads to some surprising outcomes.[39]

Homicides take almost as many lives as leukemia (6 vs. 7.2), a dual testament to the advances in treating that malignancy and to the extraordinary violence of American society.

Motor vehicle accidents take twice as many lives (and, moreover, much younger ones) than does diabetes (52.2 vs. 25.7), and accidental poisoning and noxious substances exact a higher death toll than does breast cancer (19.9 vs. 13.1). But these comparisons use the same denominator (100,000 people) without taking into account the duration of exposure to a given cause of death.

A more insightful metric then is to use the time during which people are affected by a given risk as the common denominator, and do the comparisons in terms of fatalities per person per hour of exposure—that is, the time when an individual is subject, involuntarily or voluntarily, to a specific risk.

Calculating the baseline, the average population-wide or sex- and age-specific risk of overall mortality, is easy. In 2019, overall mortality (crude death rate) of well-off (developed) countries clustered at around 10/1,000, with actual rates ranging from 8.7 for North America to 10.7 in Japan and 11.1 for Europe. That annual mortality of 10/1,000 (with 1,000 people subject to dying for 8,766 × 1,000 hours) prorates to 0.000001 or 1 × 10-6 per person per hour of exposure. Cardiovascular diseases are the leading cause of mortality in all affluent countries and they account for nearly a quarter of that total (3 × 10-7). Seasonal influenza carries a risk an order of magnitude lower (usually about 2 × 10-8 and up to 3 × 10-8), and even in the violence-prone United States the risk of homicide has recently been just 7 × 10-9 per hour of exposure, half the risk of death attributable to falls (1.4 × 10-8). But, as already noted, the frequency of the latter kind of accidental death is highly skewed, with people over 85 years of age having a risk of 3 × 10-7 compared to just 9 × 10-10 for people 25–34 years old.[43]

Men and women had similar rates, but states differed significantly, with California as low as 0.84 AEMT deaths per 100,000. In absolute terms this averages to about 4,750 deaths a year, less than 2 percent of the lowest estimate published in 2016.[49] Translated into a comparative risk metric, this results in about 1.2 × 10-6 fatalities per hour of exposure, which means that any elderly male reader of this book (whose general mortality risk is between 3 × 10-6 and 5 × 10-6) will increase his risk of demise due to AEMT by no more than about 20–30 percent during the few days of an average stay in an American hospital—and that, I would argue, is a very encouraging risk finding!

But many risky exposures cannot be so easily assigned, because there is no clear dichotomy between voluntary and involuntary risks: driving to work may be a matter of choice for a family that built a dream exurban house, but it is a matter of unavoidable necessity for millions of people in North America with its notoriously poor mass transit systems.

Assuming an average combined speed of 65 km/hour (about 40 mph) gives us annually about 80 billion driving hours in the US, and with 40,000 fatalities this translates exactly to 5 × 10-7 (0.0000005) fatalities per hour of exposure. Neither the fact that traffic fatalities also include pedestrians and bystanders killed by vehicles nor the deployment of other plausible average speeds (say, 50 or 70 km/hour) would change the order of magnitude. Driving is an order of magnitude more dangerous than flying, and during the time a person is driving the average chance of dying goes up by about 50 percent compared to staying at home or tending a garden (as long as that does not include climbing a tall ladder or working with a large chainsaw).

Finally, a few key numbers concerning one of the most dreaded modern involuntary exposures: the risk of terrorism. Between 1995 and 2017, 3,516 people died in terrorist attacks on US soil, with 2,996 fatalities (or 85 percent of that total) on September 11, 2001.[58]

In less fortunate countries, the recent toll of terrorist attacks has been much higher: in Iraq in 2017 (with more than 4,300 deaths) the risk rose to 1.3 × 10-8, and in Afghanistan in 2018 (7,379 deaths) to 2.3 × 10-8, but even that rate raises the basic risk of being alive by just a few percent and it remains lower than the risk people voluntarily assume by driving (particularly in places with no lanes and ad hoc traffic rules).[59]

As a result, the lower exposure rates to terrorist threats carry an unquantifiable accompaniment of dread, qualitatively so different from being concerned about possibly slippery roads during a morning commute.

Remarkably, calculations of exposure risks to other commonly encountered natural disasters around the world converge on the same order of magnitude (10-9) or yield even lower rates. Again, these low average fatality exposure rates help to explain why entire countries come to terms with the ever-present risks of earthquakes.

But for a population that grew from 71 million in 1945 to nearly 127 million in 2020, that works out to about 5 × 10-10 (0.0000000005) fatalities per hour of exposure, four orders of magnitude lower than the country’s overall mortality rate: obviously adding 0.0001 to 1 can hardly be a decisive factor that changes the overall assessment of life’s risks.

At the same time, there are reasons for concern, as both the annual worldwide frequency of natural disasters and their economic cost have been increasing. We can say this with a high degree of confidence because the world’s largest reinsurance companies (whose profits and losses depend on the unpredictable occurrences of earthquakes, hurricanes, floods, and fires) have been carefully monitoring their trends for decades.

When we think about rare but truly extraordinary risks that have global effects, and even more so when we contemplate catastrophic events that could severely damage or even end modern civilization, we do so on an altogether different mental plane: those real (albeit very low) risks belong to a very different perception category. As with every event that might take place in a possibly quite distant future, we strongly discount their impact and, as demonstrated yet again by the 2020 pandemic, we are chronically unprepared to deal even with those risks whose recurrence is measured in decades, not in centuries or millennia.

Americans should not worry either about supernovas or asteroids, but if they want to scare themselves by thinking about an inevitable natural catastrophe (and one that would emanate from one of the country’s cherished places!) then they should consider another mega-eruption of the Yellowstone supervolcano.[72] Geological evidence shows nine eruptions during the past 15 million years, with the last three known eruptions occurring 2.1 million, 1.3 million, and 640,000 years ago. Of course, the dating of just three events offers no basis for predicting any periodicity, but still a thought intrudes: taking the mean interval of 730,000 between eruptions we would have still 90,000 years of waiting left, but if the first interval was 800,000 years and the second one was 660,000 years then a similar shortening would indicate the next span being some 520,000 years—and a new eruption would be already more than 100,000 years overdue!

than 62 years, and they had all lived through three viral pandemics in a single lifetime: 1957–1959 (H2N2), 1968–1970 (H3N2), and 2009 (H1N1).[81] The best reconstruction of the total mortality for the 1957–1959 pandemic was 38/100,000 (1.1 million deaths; global population 2.87 billion), the 1968–1970 pandemic had a mortality of 28/100,000 (1 million deaths; global population 3.55 billion), while the 2009 event had low virulence and mortality no higher than 3/100,000 (about 200,000 deaths; global population 6.87 billion).[82]

Obviously, we will be able to quantify the total COVID-19 mortality only after this latest pandemic ends. Meanwhile, the best way to assess the recurrent pandemic burden is to compare it to global seasonal influenza-associated respiratory mortality. The most detailed assessment for the years 2002–2011 found a mean of 389,000 deaths (ranging between 294,000 and 518,000) after excluding the 2009 pandemic season.[85] This means that seasonal influenza accounts for about 2 percent of all annual respiratory deaths, and that its mortality rate averages 6/100,000—or 15–20 percent of the death rates recorded in the two late 20th-century pandemics (1957–1959, 1968–1970). Inversely stated, the first pandemic exacted a more than six times higher and the second one a nearly five times higher relative death toll than seasonal influenza.

Provisional COVID-19 data from the CDC make that clear: during the week of the peak US COVID-19 mortality (ending on April 18, 2020), people over 65 years of age accounted for 81 percent of all deaths, and those younger than 35 years for a mere 0.1 percent.[88]

Where risk is concerned, many truisms seem to be permanent. As individuals, we can exercise some control. Many people do not find it difficult to abstain from smoking, consuming alcohol and drugs, and prefer to stay home rather than sharing a cruise ship with 5,000 passengers and 3,000 crew in the midst of a coronavirus or norovirus outbreak.

Another set of truisms applies to our risk assessment. We habitually underestimate voluntary, familiar risks while we repeatedly exaggerate involuntary, unfamiliar exposures.

As I already noted, about a billion people have lived through three pandemics, but when COVID-19 struck references were made overwhelmingly to the 1918 episode, as the three more recent (but less deadly) pandemics—unlike the widely remembered fear of polio during the 1950s or AIDS in the 1980s—have left no or only the most superficial impressions.[95]

And the lessons we derive in the aftermath of major catastrophic events are decidedly not rational. We exaggerate the probability of their recurrence, and we resent any reminders that (setting the shock aside) their actual human and economic impact has been comparable to the consequences of many risks whose cumulative toll does not raise any extraordinary concerns. As a result, fear of another spectacular terrorist attack led the US to take extraordinary steps to prevent it. These included multitrillion-dollar wars in Afghanistan and Iraq, fulfilling Osama bin Laden’s wish to draw the country into stunningly asymmetrical conflicts that would erode its strength in the long run.[99]

Terrorists have always exploited this reality, forcing governments to take extraordinarily costly steps to prevent further attacks while repeatedly neglecting to take measures that could have saved more lives at a much lower cost per averted fatality.

(I always think first of the 26 people, including 20 six- and seven-year-old children, shot in 2012 in Newtown, Connecticut) have been able to change the laws, and during the second decade of the 21st century about 125,000 Americans were killed by guns (the total for homicides, excluding suicides): that is the equivalent of the population of Topeka, Kansas or Athens, Georgia or Simi Valley, California—or of Göttingen in Germany.[101]

In contrast, 170 Americans died in all terrorist attacks in the US during the second decade of the 21st century, a difference of nearly three orders of magnitude.[102]

Understanding the Environment The Only Biosphere We Have

Returning to the real world, if our species is to survive, never mind to flourish, for at least as long as high civilizations have been around (that is, for another 5,000 or so years), then we will have to make sure that our continuing interventions do not imperil the long-term habitability of the planet—or, as modern parlance has it, that we do not transgress safe planetary boundaries.[4]

Providing systematic reviews of all of these concerns—and setting them in their appropriate historical and environmental perspectives—is a task for a major book, not for a single chapter (unless it consisted of superficial summaries). Instead, I have decided to give this chapter a decidedly utilitarian tilt and focus on just a few key existential parameters, starting with the environmental circumstances of three irreplaceable existential requirements—breathing, drinking, and eating. Provision of these three preconditions of our existence depends on natural goods and services: on the oxygenated atmosphere and its incessant circulation; on water and its global cycle; and on soils, photosynthesis, biodiversity, and flows of plant nutrients. In turn, their provision affects natural goods and services.

Appraisals of the environment are perhaps even more prone to unwarranted generalization, biased interpretation, and outright misinformation than those of energy and food production. This tendency must be condemned and resisted: we will not succeed if our actions are based on myths and misinformation. Admittedly, the underlying science is often complex and many verdicts are uncertain and resolute judgments are unadvisable—but not in this particular case.

And that is an understatement—we waste water enormously and, so far, we have been slow to adopt many effective changes that would reverse undesirable habits and trends. As we will see, water supply is thus a perfect example of an almost universally mismanaged resource, with the added complication of highly uneven access.[15]

No rehydration for a day is a trying experience, for two days it becomes perilous, for three days it is usually fatal. Beyond this existential necessity, translating to a per capita average of about 750 kilograms (or liters, or 0.75 cubic meters) of water a year, there are several other—and much more voluminous—water needs: for personal hygiene, cooking, and laundry (even without an indoor toilet, these categories add up to the minima of 15–20 liters a day, or about 7 cubic meters a year), for productive activities, and, above all, for growing food.[17]

This is why national per capita consumption is the best (the most exhaustive) way to assess water footprints: it adds green, blue, and grey water components as well as all virtual water (water that was required for the growth or production of imported food and manufactured goods).[18] Domestic blue water use (all values are in cubic meters per year per capita) ranges from just over 29 in Canada and 23 in the USA to about 11 in France, 7 in Germany, and about 5 in China and India, and to less than 1 in many African countries.[19]

As a result, economies with very different climates and sectoral consumption—Canada and Italy, Israel and Hungary—have similar consumption totals (in all of these cases, between 2,300 and 2,400 cubic meters/year/capita). Food imports incorporate considerable amounts of green water, and hence the two countries with the highest dependence on imported food—Japan, South Korea—are also the highest users of virtual water.

The amount of land used in food production could be reduced with the combination of better farming practices, reduced food waste, and the widespread adoption of moderate meat consumption. As already explained in chapter 2, reversion to preindustrial farming is inconceivable in a world of nearly 8 billion people, but getting higher yields with existing inputs (agricultural intensification) conforms to a long-established trend, and the elimination of many wasteful practices could produce higher yields even with reduced fertilizer or pesticide use. A convincing large-scale, decade-long (2005–2015) demonstration of this included nearly 21 million farmers cultivating about a third of China’s cropland: they were able to raise staple grain yields by 11 percent while reducing the application of nitrogen per hectare by 15–18 percent.[24]

Losses of nitrogen from fertilized farmland (and from animal and human waste) also cause eutrophication, but aquatic photosynthesis is more responsive to phosphorus additions. Neither primary sewage treatment (sedimentation removes 5–10 percent of phosphorus) nor secondary removal (filtration captures 10–20 percent) prevents eutrophication, but phosphorus can be removed by using coagulating agents or by microbial processes, then turned into crystals and reused as fertilizer.[33]

We are concerned about too much of something without which we could not be alive: the greenhouse effect. This existential imperative is the regulation of the Earth’s atmospheric temperature by a few trace gases—above all by carbon dioxide (CO2) and methane (CH4). Compared to the two gases that make the bulk of the atmosphere (nitrogen at 78 percent, oxygen at 21 percent), their presence is negligible (small fractions of a percent) but their effect makes the difference between a lifeless, frozen planet and a blue and green Earth.[35]

“Thus human beings are now carrying out a large scale geophysical experiment of a kind that could not have happened in the past nor be reproduced in the future. Within a few centuries we are returning to the atmosphere and oceans the concentrated organic carbon stored in sedimentary rocks over hundreds of millions of years.”[47]

This means that climate sensitivity is extremely unlikely to be so low that it could prevent substantial warming (in excess of 2°C) by the time the atmospheric CO2 concentration rises to about 560 ppm, twice the preindustrial level. And yet, so far, the only effective, substantial moves toward decarbonization have not come from any determined, deliberate, targeted policies.

Because oceans have an enormous capacity to absorb atmospheric heat, it takes a long time to raise the average temperature of the lower atmosphere by an appreciable margin. During the late 2010s, after a couple of centuries of accelerated burning of fossil fuels, the temperature averaged across global land and ocean surfaces was almost 1°C above the 20th-century mean. It has been documented on all continents, but it has not been evenly distributed: as Arrhenius rightly predicted, higher latitudes have seen much higher average increases than the mid-latitudes or the tropics. In terms of the global average, the five warmest years in the past 140 years have happened since 2015, and 9 of the 10 warmest years have been experienced since 2005.[54]

Many regional, national, and global models have examined future water availability. They assume different degrees of global warming, and while the worst-case scenarios offer a generally deteriorating outlook, there are substantial uncertainties depending on necessary assumptions about population growth and therefore water demand. With a warming of up to 2°C, populations exposed to increased, climate change–induced water scarcity may be as low as 500 million and as high as 3.1 billion.[61]

And yet, fundamentally, this is a self-limiting natural phenomenon that we have experienced on a global scale three times since the late 1950s: even without any vaccines, every viral pandemic eventually subsides once the pathogen infects relatively large numbers of people or once it mutates into a less virulent form. In contrast, global climate change is an extraordinarily complex development whose eventual outcome depends on far-from-perfectly understood interactions of many natural and anthropogenic processes. As a result, we will need, for decades to come, more observations, more studies, and far better climate models in order to get more accurate appraisals of long-term trends and of the most likely outcomes.

At the same time, we do not need an endless stream of new models in order to take effective actions. There are enormous opportunities for reducing energy use in buildings, transportation, industry, and agriculture, and we should have initiated some of these energy-saving and emissions-reducing measures decades ago, regardless of any concerns about global warming. Quests to avoid unnecessary energy use, to reduce air pollution and water, and to provide more comfortable living conditions should be perennial imperatives, not sudden desperate actions aimed at preventing a catastrophe.

Is it necessary to airlift blueberries from Peru to Canada in January, and green beans from Kenya to London? The vitamin C and roughage these foodstuffs provide can be secured from many other sources with much lower carbon footprints. And could not we, with our immense data-processing capabilities, price food better and more flexibly in order to make a major dent in the 30–40 percent waste rate? Why not do what can be done, profitably and immediately, rather than waiting for more modeling exercises?