Why more CO2 is a positive good, and why we should make more of it
By Paul Spooner, P.E.
Written with Grok Bot, in Grok Build.
If you have comments or feedback, feel free to leave it here.
I've spent my career around machines that have to work: power plants, nuclear systems, and lately the compute that runs AI. Engineering teaches you a few things early. Every design is a tradeoff. Costs are as real as benefits. A system that fails on the coldest night of the year has failed, however good it looked on the spreadsheet.
Climate policy as practiced today breaks all three rules. It treats carbon dioxide as a pollutant with no benefits and treats emission cuts as worth any price. It also assumes the grid will hold up on power sources that, by their owners' own accounting, show up for a fraction of their nameplate when demand peaks.
I want to make the opposite case on purpose, and call it what it is: CO2 maxing. I mean a policy stance that treats abundant, cheap, firm energy as the first priority and treats the CO2 itself as a positive good. CO2 is not a cost that has to be managed. It is a benefit in every way that matters. It feeds plants. It warms the cold hours that kill people. The fuel that releases it is what makes a safe life possible. This is not a denial that CO2 warms the planet. It does, and that warming is part of the good. Even if we were not generating any power, we should dig up coal and burn it just to release the CO2.
Bjørn Lomborg is a useful starting point because he grants the other side its premises and the apocalypse still fails. He accepts that man-made warming is real. His argument, in Cool It (2007) and False Alarm (2020), is that panic produces bad, expensive policy. False Alarm's subtitle sums it up: How Climate Change Panic Costs Us Trillions, Hurts the Poor, and Fails to Fix the Planet (Lomborg, False Alarm). He still calls warming a problem to be managed. That is the wrong conclusion, and the rest of this paper says why. His numbers are enough to reject the panic. They are not a reason to call CO2 a cost.
His central tool is cost-benefit analysis, which he formalized through the Copenhagen Consensus Center. In 2015 testimony to the U.S. House Science Committee, Lomborg estimated the Paris Agreement would cost about $924 billion a year if implemented efficiently, and about $1.85 trillion a year under the more likely inefficient implementation (Lomborg, House testimony, Dec. 2015). False Alarm carries forward a cost in the range of $1–2 trillion per year by 2030.
Lomborg's question is an engineer's question: per dollar spent, where do we get the most human welfare? The Copenhagen Consensus work keeps finding that money spent on nutrition, disease, trade, education, and R&D into cheaper clean energy returns far more per dollar than forced near-term emission cuts. That is the floor of the argument, the case against panic even if you pretend CO2 is only a harm. The sections below drop that pretense.
The most under-reported climate fact of the last decade is that the planet has gotten greener, and CO2 is the main reason.
In 2016, Zhu et al. published "Greening of the Earth and its drivers" in Nature Climate Change. The team had 32 authors from 24 institutions in eight countries and used NASA MODIS and NOAA AVHRR satellite records. They found that between a quarter and half of Earth's vegetated lands showed significant greening over 35 years. NASA described the added leaf area as "equivalent in area to two times the continental United States" (NASA, 2016; Zhu et al. 2016).
On attribution, co-author Ranga Myneni of Boston University said: "carbon dioxide fertilization explains 70 percent of the greening effect... The second most important driver is nitrogen, at 9 percent. So we see what an outsized role CO2 plays in this process." (NASA, 2016).
This is basic plant physiology. Photosynthesis combines CO2 and water into sugars. For C3 plants, which include wheat, rice, soybeans, and most trees, today's ~420 ppm is well below the concentration at which photosynthesis saturates. More CO2 also lets plants partly close their stomata, so they lose less water per unit of carbon fixed. That matters most in dry regions.
NASA's write-up lists caveats, and I answer them below. Plants partly acclimate to higher CO2, and the split among causes comes from models. Neither caveat touches the observation, which comes from satellites: the Earth got greener, and CO2 did most of it. A greener planet is not a side effect to apologize for. It is what more CO2 does.
If you want to know what farmers think of extra CO2, look at what they pay for. Commercial greenhouse growers routinely burn natural gas or buy liquid CO2 to raise indoor concentrations to around 1,000 ppm, roughly two and a half times today's outdoor air.
Ontario's agriculture ministry puts it plainly: "For the majority of greenhouse crops, net photosynthesis increases as CO2 levels increase from 340–1,000 ppm... increasing the CO2 level to 1,000 ppm will increase the photosynthesis by about 50% over ambient CO2 levels" (Government of Ontario, Supplemental carbon dioxide in greenhouses). The province's greenhouse vegetable production guide ties higher CO2 to roughly a 15–30% increase in yield (OMAFRA Publication 836B). Manitoba's agriculture department gives similar guidance (Manitoba Agriculture, Greenhouse CO2 Supplement).
A greenhouse is not the whole planet, and it does not need to be. Water, nitrogen, temperature, and pests are controlled indoors. The satellite record is the uncontrolled experiment, and it came out the same way. Growers pay real money for extra CO2 every season because the gain is large enough to invoice. The gas policy calls a pollutant is the gas working horticulturists buy on purpose.
CO2's radiative forcing doesn't rise in a straight line with concentration. It rises with the logarithm of concentration. The standard simplified expression, from Myhre et al. (1998) in Geophysical Research Letters and adopted by the IPCC's Third Assessment Report, is:
ΔF = 5.35 × ln(C / C₀) W/m²
Myhre et al. 1998; IPCC TAR, Ch. 6, Table 6.2
The engineering consequence is that each additional ppm adds less forcing than the one before. Going from 280 to 420 ppm gives about 5.35 × ln(1.5) ≈ 2.2 W/m². Going from 420 to 560 ppm, another 140 ppm, gives only about 5.35 × ln(1.33) ≈ 1.5 W/m². A doubling from any baseline gives the same ~3.7 W/m², no matter how much CO2 was already there.
This is the diminishing-returns curve every engineer knows. Warming does not run away. Each new tonne adds less forcing than the last, while the benefit to plants continues over the same range. Feedbacks change the slope. They do not change the shape, and they do not turn a benefit into a cliff. The warming that does arrive lands on the cold hours, as Section 7 shows. The marginal tonne pays twice: less additional warming than the tonne before it, and a continued gain to every leaf.
For more than a decade, a large share of the alarming climate-impact literature has relied on RCP8.5, a scenario with very high emissions. It was routinely labeled "business as usual." That label was wrong.
In a 2020 Nature commentary, Zeke Hausfather and Glen Peters wrote: "Stop using the worst-case scenario for climate warming as the most likely outcome — more-realistic baselines make for better policy" (Hausfather & Peters 2020, Nature). Their title says it directly: "Emissions – the 'business as usual' story is misleading." RCP8.5 assumes, among other things, a huge expansion of coal use that the real energy system isn't on track to deliver.
This matters for policy. When utilities, insurers, and regulators plan around RCP8.5, they inflate projected damages and, through that, the apparent benefit of costly mitigation. Rejecting that scenario is not an argument for burning less coal. It is an argument for refusing to treat a fantasy coal boom as the excuse for policies that keep real coal, and the CO2 it releases, in the ground. I raised the scenario point in my comments on Avista's Draft 2027 Electric IRP. The utility itself concedes RCP8.5 "may not be representative," yet it proposes to wait until a later cycle to fix it. Honest planning should use plausible scenarios now, and it should count the CO2 from real fuel as a benefit.
Even mainstream climate economics, the work that earned a Nobel Prize, doesn't conclude that we should cut emissions at any cost. These models still score CO2 as damage. That is the wrong sign. What they show, even so, is that the other side cannot justify its program on its own books.
William Nordhaus won the 2018 Nobel in economics for the DICE integrated assessment model. In his Nobel lecture he reported a finding he called "much more controversial": in DICE, "the cost-benefit optimum rises to over 3°C in 2100 – much higher than the international policy targets. Even with the much more pessimistic alternative damage function, the temperature path rises to 3°C in 2100" (Nordhaus, Nobel Prize Lecture, 2018). Critics who re-ran his DICE-2016R2 confirm that its optimal path reaches about 3.5°C by 2100 (Hänsel et al. 2020, Nature Climate Change; Author Correction).
In other words, the best-known cost-benefit model in climate economics says that pushing hard for 1.5°C or 2°C costs more than it saves under its standard assumptions. Those assumptions count warming as a loss. They are not a finding that the loss is real.
Richard Tol's survey of the damage literature points the same way. In Review of Environmental Economics and Policy (2018), he finds that at 2.5°C of warming the average estimated welfare loss is about 1.3% of income, with estimates ranging from 0.0% to 3.0% (Tol 2018). Over the same period, the world economy is expected to grow several times over. A bookkeeping entry of a percent or two, on a ledger that calls a benefit a cost, is not a reason to spend trillions a year. Once fertilization, warmer nights, and the energy itself are counted, the sign flips. More CO2 is not an optimal amount of harm. It is a good.
The public story about temperature and health is almost entirely about heat waves. The data say the opposite.
Gasparrini et al. (2015) published in The Lancet what was then the largest study of its kind: 74,225,200 deaths across 384 locations in 13 countries, 1985–2012. They found that 7.71% of all deaths were attributable to non-optimal temperature. Cold accounted for 7.29% and heat for just 0.42% (Gasparrini et al. 2015, The Lancet; PubMed). An accompanying Lancet commentary summarized it: "cold-related deaths outnumbered heat-related deaths by a factor of nearly 20" (Lancet comment).
Two implications follow. First, cold deaths outnumber heat deaths about seventeen to one in the study itself: 7.29% against 0.42%, which is what the commentary's "nearly 20" is rounding. Warming does not nibble at that margin. It slashes temperature-related deaths, totally offsetting the added heat deaths by over an order of magnitude. The warming also lands on the cold side of the day. Karl et al. found that from 1951 to 1990 the daily minimum rose at three times the rate of the daily maximum, 0.84°C versus 0.28°C, so most of the warming was at night, when it is coldest (Karl et al. 1993). The hours that do the killing are the hours that got warmer. A shift like that removes cold deaths and adds many fewer heat deaths. Second, the best protection against both cold and heat is cheap, reliable energy: furnaces in January, air conditioning in July. That's a power-supply problem, not an emissions problem. Here in North Idaho, the dangerous hour on the grid is a January cold snap, not an August afternoon.
Lomborg's most-shared chart uses the EM-DAT international disaster database. By his calculation, annual deaths from climate-related disasters (floods, droughts, storms, wildfires, extreme temperatures) averaged 485,000 a year in the 1920s and 18,362 a year in 2010–2019, a decline of 96.2% (Lomborg). Meanwhile, world population has roughly quadrupled.
Honesty requires the caveat, and the database's own maintainers supply it. CRED, which runs EM-DAT, notes that the 1920s baseline is dominated by a handful of mega-famines. Against a 1910s baseline the decline is about 30%, and CRED warns that the century-long trend alone can't establish causes (CRED, 2022 Disasters in Numbers). I'll take that correction on board.
But look at what the corrected story still says. An independent 2024 analysis in Environmental Hazards concluded that "the data shows an unequivocal dramatic decrease in the fatalities reported from weather-related disasters over the last century." It found that the individual risk of death from such disasters has fallen roughly in proportion to population growth since the mid-1900s (Environmental Hazards, 2024). Whatever the exact percentage, a person today is far less likely to be killed by weather than their great-grandparents were. That progress came from wealth: early warning systems, concrete buildings, irrigation, trade in food, and emergency response, all powered by abundant energy.
Alex Epstein's Fossil Future (2022) makes the moral argument I find most persuasive. He argues that the right standard for energy policy is human flourishing, and that fossil fuels' unique combination of low cost, reliability, and scalability has been the foundation of it (Penguin Random House, Fossil Future). In his framing, we don't take a safe climate and make it dangerous. We take a naturally dangerous climate and make it far safer through energy and technology.
The numbers on energy poverty are stark. According to the International Energy Agency, about 730 million people still had no access to electricity in 2024, a figure that fell by only 11 million from 2023 (IEA, 2025). These people don't need a carbon budget. They need refrigeration for vaccines, lights for schoolwork, pumps for clean water, and fertilizer, which means natural gas.
When rich countries pressure development banks to stop financing fossil fuel projects in poor ones, they're asking the world's poorest people to pay for a climate benefit decades away with their health today. Lomborg's phrase fits: climate panic "hurts the poor."
This is the part I know best.
Grid reliability isn't measured in annual megawatt-hours. It's measured in the worst hour of the year. In the Inland Northwest that hour is a still, frozen January evening, when solar output is near zero and wind is often calm under a high-pressure system.
Utilities know this, and their own planning documents show it. In Avista's Draft 2027 Electric IRP, the January peak capacity credits are about 10% for wind, 4% for solar, and 6% for 4-hour batteries. Put differently, a 100 MW wind farm is counted as roughly 10 MW of dependable capacity when it matters most. Yet the preferred plan still leans on more than 2,000 MW of new Washington wind to meet state mandates.
Firm resources, meaning natural gas turbines, hydro, and nuclear, can be dispatched when they're needed. Avista's plan rightly adds new gas turbines in Idaho. What it doesn't do is select any nuclear. Small modular reactors are the one technology that offers firm, zero-emission, compact power with a fuel supply that doesn't depend on the weather. Having worked in the nuclear industry, I find that omission the clearest sign that our planning priorities are backwards. We subsidize intermittent capacity that doesn't show up at peak, and we stall on the firm capacity that does.
AI data centers, electrified heating, and reshored manufacturing are all driving load growth. Demand is rising again for the first time in a generation. A grid built on intermittent sources plus thin battery storage, with no adequate firm backup, isn't green. It's fragile. And when a grid fails in winter, people die of cold, which, as Section 7 showed, is the bigger temperature killer to begin with.
The last plank is about honest accounting. Utilities increasingly load their resource selection with "non-energy" adders such as social cost of carbon values, equity and "Named Community" weightings, and ESG criteria. Avista's own draft IRP admits these adders "may lead to outcomes that differ from traditional least-cost planning."
That's the point. Least-cost, reliable service is the utility's obligation to its ratepayers. Social policy belongs with legislatures, funded transparently, not buried in a kilowatt-hour rate that falls hardest on retirees and low-income families. If the cost-benefit math in Section 6 is right, many of these adders push ratepayers to spend more than the climate damages they claim to avoid. Where a multi-state utility serves both Washington and Idaho, one state's mandates shouldn't quietly raise the other state's bills.
A persuasive case has to face the best objections. Here are the ones I take most seriously.
"CO2 fertilization diminishes over time, and greening isn't the same as crop nutrition." Plants do acclimate, and Zhu et al.'s co-authors note that the attribution split depends on models (NASA). Elevated CO2 can also slightly lower protein and mineral content in some crops. Response: diminishing returns are not zero returns, and a change in grain composition is not a reason to call the gas a pollutant. Greenhouse practice shows large gains up to ~1,000 ppm, and nutrient content is a breeding and fertilizer problem, the kind agriculture solves routinely. The satellite record stands. More CO2 made the Earth greener.
"Lomborg's 96% disaster-death decline is cherry-picked." Partly fair. CRED and the LSE's Grantham Institute have criticized the 1920s baseline (CRED; Grantham Institute). Response: I've cited the caveat above. Even independent analyses using the same data find a dramatic per-capita decline in weather-disaster deaths (Environmental Hazards, 2024). The argument doesn't rest on the exact percentage.
"Updated DICE models say 1.5–2°C is optimal, not 3.5°C." Hänsel et al. (2020) found that with updated carbon-cycle calibration, higher damage estimates, and lower discount rates, most expert views make 2°C economically optimal (Hänsel et al. 2020). Response: those runs still score CO2 as damage. The result also depends heavily on the discount rate, which is an ethical choice, not a physical constant, and on damage functions that remain highly uncertain. Uncertainty is a reason to refuse trillions in near-term costs. It is not a reason to treat a plant nutrient as a harm.
"Tail risks such as tipping points, ice sheets, and high climate sensitivity justify insurance." This is the objection the other side treats as decisive. Response: the premium on offer is not insurance. It is a poorer society and a weaker grid, which is how cold and weather actually kill people. A richer society with abundant firm energy survives surprises. Making energy scarce in the name of a tail does the opposite. Nuclear and other firm low-carbon plants are worth building because they are firm and cheap, not because CO2 is a sin.
"The logarithmic curve doesn't account for feedbacks." Feedbacks are real, and equilibrium climate sensitivity is uncertain. Response: feedbacks change how much warming you get per doubling. They do not repeal the logarithm. Each additional tonne still has less marginal forcing than the one before it. Combined with emissions paths more plausible than RCP8.5 (Hausfather & Peters), the realistic range is a modest warming that, as measured, has fallen mostly on cold nights.
"Heat deaths will rise as the world warms." Some will, especially where people are poor and have no air conditioning. Response: the same evidence shows cold deaths outnumber heat deaths by about 17 to 1, and the warming has been concentrated in the nighttime lows (Gasparrini et al.; Karl et al. 1993). Added heat deaths do not catch the cold deaths that warmer nights remove. The way to deal with hot afternoons is air conditioning and reliable electricity, which is the energy abundance CO2 maxing is for.
Put the pieces together. CO2 is plant food that has measurably greened the planet. Its warming effect grows logarithmically, not linearly, and the warming that has arrived has fallen mostly on cold nights. Cold kills far more people than heat, by about seventeen to one, so that warming slashes temperature-related deaths. The worst-case emissions scenario was never the likely one. Even the economic models that insist on calling CO2 a cost cannot justify emission cuts at any price. Weather-disaster deaths per person have fallen dramatically as the world got richer on fossil energy. Hundreds of millions of people still have no electricity at all. And the grid that keeps us alive in a January cold snap depends on firm power, not on resources that show up at a tenth of their nameplate.
The engineering answer is obvious: build firm, cheap power first, and welcome the CO2. That means gas now, coal where it is the fuel that works, nuclear and SMRs as fast as we can license them, and hydro wherever we have it. Plan on realistic scenarios. Strip social-policy adders out of utility rates. The CO2 that comes with prosperity is not a sin and not a cost to be managed. It is a reason to burn the fuel. If the power were worthless, the CO2 would still be worth releasing.
That is CO2 maxing. It is what the evidence shows when you stop borrowing the other side's conclusion.
QR Code for this page
