Key Takeaways
Treat human control as a feedback loop, not a one-time fix
The planet is already managed. The Anthropocene is the proposed name for an era in which human activity has become a major force shaping Earth’s systems. People have transformed more than half of ice-free land and dammed or diverted most major rivers. Kolbert opens with Chicago, where engineers reversed the river to carry sewage away from Lake Michigan, the city’s drinking-water source. The project solved an urgent public-health crisis, but it also connected two river basins that had once been separate.
Each fix inherits a past. After invasive fish began using the canal, engineers installed electric barriers, then considered adding noise and bubbles. The task is no longer simply to control nature, but to manage the effects of earlier control. The history cautions against treating infrastructure as a final answer: interventions reshape the conditions that future interventions must address.
The canal is a vivid case of path dependence: once a city, economy, and water system are built around a choice, reversing that choice becomes far harder than making it. Systems engineering often distinguishes a local fix from a whole-system solution, because optimizing one part can shift costs elsewhere. Chicago’s sewage problem was real, and the canal’s public-health benefits matter. The deeper lesson is not that engineering should stop, but that major projects need monitoring, revision plans, and attention to consequences beyond their original boundary.
A fix can turn yesterday's waste into tomorrow's invasion
Good intentions traveled downstream. Asian carp were imported to the United States to control aquatic weeds and, later, to consume algae in sewage-treatment ponds. Those purposes echoed Rachel Carson’s argument in Silent Spring that biological controls could replace harmful chemical treatments. But carp escaped from farms and treatment ponds into the Mississippi system. The four fish are exceptionally effective at feeding and reproducing. On the Illinois River, Asian carp came to make up nearly three-quarters of the fish biomass, crowding native species and threatening freshwater mussels.
Containment became its own project. Barrier defense means removing carp from waterways before they reach the electric barriers near Chicago. One campaign removed more than 50,000 pounds of fish in four days, then ground them into fertilizer. Reconnecting the Great Lakes and Mississippi basins made the invasion possible; separating them again was estimated to take 25 years and up to $18 billion.
This case shows why ecological risk depends on context, not simply on whether an organism is useful or harmful. A species that performs well in one managed setting can become disruptive when it reaches a different food web. The same principle appears in public health: interventions can change incentives and exposure patterns, sometimes moving rather than eliminating risk. The carp story also complicates blame. Their arrival followed a sequence of official decisions, not a single reckless act, which makes accountability and long-term evaluation essential parts of environmental policy.
Protecting a place can destroy the processes that sustain it
A defended river stopped building land. The Mississippi once spread sediment over southern Louisiana during floods, adding soil to a delta that naturally sinks as its wet ground compacts. Levees and spillways reduced catastrophic flooding, but they also kept much of that sediment from reaching the coast. Since the 1930s, Louisiana has lost more than 2,000 square miles. This loss brings the Gulf closer to New Orleans and removes wetlands that help buffer storm surges.
Rebuilding has its own limits. One project pumped river sediment into a coastal basin, creating 186 acres of marsh for $6 million. Yet the new ground began sinking after the pumps stopped. New Orleans faces a related loop: draining water helps keep neighborhoods dry, but drying marsh soils increases subsidence, or the gradual sinking of land, which in turn increases the need for drainage. Protection can preserve lives and property while undermining the landscape it depends on.
The delta illustrates a central problem in infrastructure planning: defenses often change the hazard they are meant to contain. A levee can lower routine flood risk while increasing dependence on a maintained barrier and interrupting the sediment supply that sustains wetlands. This does not make levees or pumps pointless. It means their value must be judged alongside the systems they displace, including who receives protection and who bears the remaining risk. Louisiana’s planned sediment diversions attempt to restore a process through engineered structures, blurring any simple line between restoration and construction.
Conservation now often means caring for species inside human-made systems
A fish’s whole world can be a pool. The Devils Hole pupfish lives in a small, isolated pool in Nevada, with a total natural habitat smaller than any other vertebrate’s known range. Pumping groundwater for development lowered the pool and exposed the shallow shelf where the fish spawn. After the population fell to 35 in 2013, officials rushed a $4.5 million replica habitat into operation. The artificial pool copied the original’s light, temperature, and food web, but a tiny beetle began eating pupfish larvae.
Dependence has a name. A conservation-reliant species is one that needs ongoing human help to persist. Devils Hole pupfish receive supplemental food, and their refuge population requires careful daily management. The label captures an uncomfortable reality: people can damage a habitat, then become responsible for keeping its inhabitants alive. Kolbert’s point is not that this care is futile, but that preserving a species may mean sustaining an altered, fragile relationship rather than restoring an untouched past.
The pupfish challenges the idea that conservation always means returning nature to a pristine state. When a species has a tiny range and its habitat has been altered, active care may be the only available ethical response. This resembles intensive care in medicine: successful treatment can keep a patient alive without restoring the original conditions of health. Yet dependence has costs, including staff time, public funding, and vulnerability to technical failure. The case makes conservation more concrete by asking not only which species survive, but what continuing obligations their survival creates.
Use coral engineering as triage, not permission to keep heating reefs
Some corals survive the stress. Rising ocean temperatures trigger bleaching, in which corals lose the algae living inside their cells and can starve if the heat persists. After severe bleaching events, marine biologist Ruth Gates proposed assisted evolution: identifying more resilient corals, breeding them, and testing whether their offspring could withstand hotter conditions. Researchers also explored whether hardier algal partners or microbes might help. The experiments use controlled stress to select organisms able to tolerate projected future conditions.
A rescue effort cannot remake a reef quickly. The Great Barrier Reef contains thousands of reefs and extraordinary biodiversity. After the 2016 heat event, more than 90 percent of it was affected, and roughly half its corals died. Breeding and reseeding might preserve some function and buy time, but interventions at reef scale are immense. Kolbert sets the lab work against the approval of a major coal mine near the reef, exposing the contradiction between engineering survival and continuing emissions.
Assisted evolution is a form of triage: it tries to preserve living capacity under conditions that humans have already made more hostile. The approach connects conservation biology with selective breeding, but reef ecosystems are far more entangled than a single crop or livestock strain. A coral’s survival depends on relationships with symbionts, microbes, and neighboring species, so resilience in a tank may not translate neatly to an open reef. That uncertainty argues for careful trials, not dismissal. It also keeps the priority clear: adaptation can buy time, while emissions reduction addresses the pressure causing the emergency.
Gene drives can turn one genetic edit into an ecological release
CRISPR makes editing easier. CRISPR is a set of gene-editing techniques adapted from bacteria that can target and alter specific DNA sequences. A gene drive goes further: it biases inheritance so an engineered trait spreads through a population more often than ordinary genetics would predict. Researchers considered using one to suppress invasive mice, perhaps by making offspring overwhelmingly male. Models suggested that 100 such mice could eliminate a population of 50,000 on an island within a few years.
The same reach creates danger. Australian scientists used CRISPR to produce cane toads whose toxin was disabled, though the animals were not intended for release. Such work might help train predators to avoid toxic toads. But a self-spreading gene drive could cross boundaries or behave unpredictably in the wild. Researchers have proposed brakes and other containment strategies, yet the book presents these as prospective safeguards, not guarantees. A tool designed to overcome nature’s usual checks could make an ecological mistake unusually hard to reverse.
Gene drives shift genetic engineering from changing an organism to influencing the inheritance patterns of a population. That makes the technology a governance problem as much as a laboratory problem: effects may cross property lines, political borders, and generations. The strongest case for use is the harm caused by invasive species and existing poisons, which can kill non-target animals. The strongest caution is that a self-propagating intervention may be difficult to recall. Decisions therefore need independent review, ecological testing, and meaningful consent from affected communities, not just confidence in technical safeguards.
Remove accumulated carbon without treating capture as a license to emit
Carbon behaves like a filling tub. Emissions can fall while atmospheric carbon dioxide keeps rising, because much of the gas already released remains in the climate system for a long time. During the 2020 lockdown, global emissions dropped an estimated 17 percent, yet atmospheric CO2 reached a record 417.1 parts per million that May. Cutting emissions is essential, but it does not remove what has already accumulated. Negative emissions means removing more CO2 from the atmosphere than is added.
The proposed scale is daunting. Direct air capture can bind CO2 and store it underground, while other proposals use forests, crushed rock, or bioenergy with carbon capture. At the time Kolbert reported, Climeworks charged $1,000 per ton for permanent storage in Iceland. Most modeled routes compatible with the Paris temperature goals relied on carbon removal. But land, energy, cost, and fairness all constrain deployment, especially when lower-emitting countries would bear the effects of choices made elsewhere.
Carbon removal resembles waste management: preventing new waste matters, but accumulated waste still requires handling. That framing can reduce moral grandstanding and focus attention on infrastructure, costs, and measurable storage. It also carries a risk: if capture is treated as a future escape hatch, it can weaken pressure to cut emissions now. The equity issue is central because cumulative emissions are unevenly distributed, while land-intensive removal can compete with food production and local livelihoods. Sound policy must pair reductions with removal, and assign responsibility according to both historical contribution and practical capacity.
Solar geoengineering has an off switch capable of abrupt warming
Volcanoes demonstrate the cooling effect. Mount Tambora’s 1815 eruption sent particles into the atmosphere, dimming sunlight and contributing to the cold, failed harvests of 1816. Solar geoengineering would deliberately reflect sunlight by placing particles in the stratosphere. Unlike carbon removal, it could begin lowering temperatures quickly, but it would mask warming rather than remove the greenhouse gases causing it.
Stopping could make matters worse. Stratospheric particles eventually fall back to Earth, so a program would need constant replenishment. If it stopped while emissions remained high, masked warming could return rapidly, a risk called termination shock. Other concerns include altered rainfall, ozone damage, whiter skies, and conflict over who controls a planetary intervention. Even a relatively cheap technology could be launched by a small number of capable nations, leaving those most affected with little say.
Solar geoengineering presents a classic governance mismatch: the intervention could affect everyone, while the decision to begin it might be made by a few actors. Its potential speed is valuable in a severe emergency, but speed also compresses the time available for deliberation and consent. The idea of a planetary thermostat is misleading because Earth has no single temperature setting, and regional rainfall or ecological responses may differ. Research can clarify risks without committing the world to deployment. Any serious discussion must keep emissions cuts central and define who can authorize, monitor, and stop the intervention.
Do not mistake today's calm for climate's natural setting
Ice cores reveal abrupt change. Greenland ice preserves layers of ancient snow, including bubbles that hold samples of past air. Analyses uncovered 25 Dansgaard-Oeschger events, abrupt climate swings during the last ice age. Greenland temperatures sometimes rose by as much as 8°C in about 50 years, then fell again. By contrast, the climate of the last 10,000 years was unusually stable. That relative calm supported agriculture and the development of cities and states.
Stability is not a guarantee. Kolbert describes recent warming and accelerating Greenland ice loss alongside this deep history. The comparison does not make human-caused warming natural or harmless. It shows that the climate system can shift dramatically, and that societies built around coastlines, roads, and fixed infrastructure are poorly prepared to move. With enough ice on Greenland to raise sea levels by about 20 feet, a slow process can still become a lasting constraint on where people live.
Paleoclimate evidence changes the frame from a stable planet occasionally disturbed by humans to a dynamic system whose recent stability was historically unusual. That perspective makes abrupt change more imaginable, but it should not blur the difference between natural past events and current human forcing. The ice record shows that climate can shift quickly; present measurements identify human activity as a major driver of today’s change. Societies also face a distinctive vulnerability: modern cities and infrastructure are difficult to relocate. Planning for climate risk therefore requires both physical science and decisions about which places can be defended, adapted, or left behind.
Analysis
Kolbert’s book is organized as a journey through three increasingly vast arenas of intervention: rivers and deltas, endangered species and genomes, then the atmosphere and ice. The sequence steadily enlarges the scale of the question. A canal can redirect fish and sewage; gene editing can potentially reshape a population; solar geoengineering could alter the climate shared by every nation. Across these cases, the recurring pattern is not simply that technology fails. It is that each intervention enters a system already altered by earlier decisions, then produces new dependencies, beneficiaries, and hazards.
The book’s strongest contribution is its refusal of two easy positions. A pristine nature is no longer available as a universal policy target, yet technological ingenuity does not amount to reliable mastery. Pupfish care and coral breeding can be morally serious responses to damage, while also revealing how much ongoing labor a repair may require. The same ambiguity appears in carbon removal: it may be needed to address accumulated emissions, but reliance on it can postpone the reductions that limit further accumulation.
This framing also makes political questions inseparable from ecological ones. The Isle de Jean Charles community bears consequences of interventions it did not choose, while climate technologies could affect regions far from the people authorizing them. Risk is therefore not only the chance of technical failure. It includes who decides, who benefits, who pays, and whether a policy can be stopped or reversed.
A limitation of the book’s vivid case-study method is that spectacular failures can loom larger than quieter interventions that work. The stories are diagnostic rather than a comparative risk calculation. That is useful, provided readers do not mistake cautionary examples for proof that every large intervention is doomed. A stronger decision rule emerges from the cases: compare intervention with inaction, demand evidence at the relevant scale, disclose uncertainty, and build accountability and exit conditions into the project. The book’s title image captures the result: the future is neither untouched nature nor a fully controlled machine, but a world made by humans that still exceeds human command.
Review Summary
Under a White Sky explores humanity's attempts to solve environmental problems caused by previous interventions. Kolbert examines efforts to control invasive species, save endangered animals, and mitigate climate change through geoengineering. While some reviewers found the book informative and thought-provoking, others felt it lacked cohesion or depth. Many praised Kolbert's writing style and ability to make complex scientific concepts accessible. The book raises important questions about the ethics and consequences of human interference in nature, leaving readers to contemplate our role in shaping the planet's future.
People Also Read
Glossary
Anthropocene
Human-shaped geological eraA proposed geological epoch in which human activity has become a major force shaping Earth’s climate, landscapes, rivers, and living systems. The book uses it to describe a world where no clear boundary remains between human influence and nature, and where attempts to manage environmental damage often build on earlier human alterations.
CHANS
Coupled human-natural systemAn abbreviation for a coupled human and natural system. It describes environments where people, infrastructure, and ecological processes continually shape one another. Kolbert applies the term to the Mississippi delta, whose behavior cannot be understood as either a purely natural river system or a purely engineered landscape.
conservation-reliant
Species needing ongoing interventionA term for species whose survival depends on continued human assistance. That help can include captive breeding, supplemental feeding, habitat management, or protection from predators and competitors. The Devils Hole pupfish is an example: its habitat has been altered, and both the wild population and its refuge population require active care.
assisted evolution
Human-guided adaptationA conservation approach that deliberately selects or breeds organisms with traits likely to help them withstand environmental change. In the book, coral researchers test heat-stressed corals, cross individuals from different reef regions, and explore more resilient algal partners. The aim is to improve survival under warming conditions, not to recreate an unchanged past.
gene drive
Biased genetic inheritanceA genetic system that causes a particular trait to be inherited more often than standard Mendelian inheritance would predict. A CRISPR-based drive can copy its own instructions into an organism’s chromosomes, potentially spreading a chosen trait through a population. The book discusses drives proposed for suppressing invasive rodents and highlights the risk of uncontrolled spread.
negative emissions
Net removal of atmospheric carbonThe removal of more carbon dioxide from the atmosphere than is added over a given period. Methods discussed include direct air capture and geological storage, enhanced weathering, forests, and bioenergy with carbon capture. Climate models rely on negative emissions to meet ambitious temperature limits, though the methods face major questions of scale, cost, land use, and fairness.
solar geoengineering
Reflecting sunlight to cool EarthA proposed intervention that would reflect a portion of incoming sunlight to reduce some effects of global warming. One approach would release reflective particles into the stratosphere, where they could scatter sunlight. It would not remove greenhouse gases, would require ongoing replenishment, and could affect rainfall, ozone, and international relations.
Dansgaard-Oeschger events
Abrupt ice-age climate swingsRapid temperature shifts recorded in Greenland ice cores during the last ice age. The record contains 25 such events, some involving substantial warming over a few decades followed by cooling. They demonstrate that the climate system can change abruptly, while the relatively stable climate of the last 10,000 years was unusual in the longer record.
hydrologic separation
Disconnecting linked river basinsThe physical separation of two drainage systems so that water and aquatic species cannot move between them. In the book, it refers to proposals to break the connection between the Mississippi River basin and the Great Lakes basin created by Chicago’s canals. The Army Corps estimated that restoring separation could take 25 years and cost up to $18 billion.
Download PDF
Download EPUB
.epub digital book format is ideal for reading ebooks on phones, tablets, and e-readers.