Skip to content
How disinformation can derail climate policy and accelerate global warming

How disinformation can derail climate policy and accelerate global warming

Earth • October 11, 2026

A new study finds that informed voters may support stronger climate policies than governments, but disinformation can quickly undermine that advantage.

Climate policy often involves an uncomfortable political trade-off. Governments must make decisions that impose costs today to prevent damage decades from now, even when voters resist.

In a new economic model, putting a carbon tax to a majority vote produced a stronger tax than leaving it to a cautious government.

The voters’ edge was fragile. Once disinformation pushed the of voters who understood the damage from warming below half, the tax dropped quickly.

Katsiaryna Bahamazava of University College Dublin and its School of Computer Science built the model with Stanley Reznik of the ILaVita Foundation.

Reznik’s foundation has research offices in Turin, Italy, and New York. The pair simulated the world economy over 300 years, in five-year steps, so no actual election or tax went into the work.

A carbon tax charges for the carbon dioxide that burning fuel releases. Real-world carbon prices have often stayed far below what standard climate-economy models recommend.

Fairness worries and resistance to taxes are familiar obstacles, and disinformation is a newer one that most of those models leave out.

How the model splits voters

Bahamazava and Reznik split the model’s population into two groups. Informed people hold the model’s true damage setting, in which 5.4°F (3°C) of warming cuts economic output by 9%. Misinformed people put that loss at 2.7%.

The population starts out split evenly. Over time, steady learning pulls the informed toward 90%, and visible climate damage speeds that up.

Disinformation pushes the other way.

, the team compared two ways of setting the tax. Under the first, a cautious government knows the true damage but pays a political price whenever its tax runs ahead of what the public supports .

Under the second, voters decide, and the tax moves toward their choice gradually, because real tax systems don’t change overnight.

Informed voters set a higher tax

With no disinformation at all, the vote won.

Over the first 50 years, the authors calculated an average tax under majority voting of $25.70 per metric ton of carbon dioxide, against $18.40 under the cautious government.

Peak warming was 5.7°F (3.17°C) with voters in charge and 6.6°F (3.65°C) with the government.

In this simulated world, voters did more for the climate than the officials did.

Under majority rule, one voter’s choice settles things: the voter in the exact middle of the electorate. While informed people are the majority, that middle voter is informed too.

The government, by contrast, keeps trimming its tax to avoid a backlash.

“As long as informed voters remain the majority, the median voter backs the ambitious policy, producing stronger early mitigation and lower peak warming,” Bahamazava said.

Losing the majority cuts the tax

Trouble starts when that majority slips away. In one run, a disinformation campaign lasted through the first 30 years.

Over the first 50 years, the average tax that voters set fell from $25.70 in the run without disinformation to $10.

Under the same pressure, the cautious government still managed $12.90. The authors call this a cliff edge.

As soon as misinformed people outnumber informed ones, the deciding voter is misinformed, and the tax that voters want drops to the weaker tax the misinformed group prefers.

Because tax systems change slowly, the actual tax doesn’t vanish at once, but it heads down quickly.

The government’s tax weakens more smoothly, since its political cost rises a little at a time as public support fades.

Late disinformation does the most harm

The timing of disinformation mattered too. When a campaign began 50 years in and never stopped, the model estimated peak warming above 10.8°F (6°C) under both voting and the cautious government.

Carbon dioxide in the air passed 1,400 parts per million. At the start of every run, the level was 420.

A weaker campaign that ran the whole time ended up as hot.

Late disinformation is especially damaging, the researchers explained, because it erodes support after the switch to stronger policy has begun. They treat it as a test of timing, not a forecast.

The researchers also ran the reverse: a strong effort against false claims, big enough to cancel out an early campaign.

With voters in charge, peak warming and carbon dioxide came back to almost exactly the levels of a world with no disinformation.

The authors describe this as a best-case scenario, representing efforts such as public communication, climate education, and rules for online platforms .

“Information policy is not just complementary to carbon pricing; it helps sustain the political conditions under which carbon pricing remains feasible,” Reznik said.

Splitting everyone into two camps is a simplification, and the authors say so.

Real beliefs spread across a range, and they expect that with a range the voters’ advantage would be smaller and the drop less abrupt.

The misinformed group’s belief is also a strong setting.

When the team reran the model with milder mistakes, voters kept their edge until the misinformed group’s estimate of the damage reached 85% of the true figure.

Disinformation in the model simply arrives. No one in it chooses to spread false claims or pays for them.

Future research directions

The authors suggest that future models could examine how people who profit from carbon-intensive industries might deliberately spread disinformation.

Another model could be designed to split the world into regions that sway each other.

How near any real electorate is to that tipping point is still an open question.

The team has posted its simulation code so other researchers can build those versions themselves.

The findings suggest that protecting public understanding of climate change may be just as important as designing the policies meant to address it.

The full study was published in the journal Risk Sciences .

Check us out on EarthSnap , a free app brought to you by Eric Ralls and Earth.com.

Extracted Entities

Countries (1)

Domains (1)