The next time you’re gazing up at a cloud, wondering whether that perfect heavenly orb looks more like a heart or a bunny rabbit, do the world a favor, and give it one of these:

[protected-iframe id=”bf2782f4c373bd3c8a90f56bbcbb68dc-5104299-80108757″ info=”//giphy.com/embed/tGJNtCjoSodyM” width=”480″ height=”437″ frameborder=”0″ class=”giphy-embed” allowfullscreen=””]

But don’t do it because there’s a big new study showing that clouds might not slow global warming as much as we once thought. Do it because despite this important new finding, clouds remain a big wispy wild card when it comes to making climate predictions.

Accounting for clouds in climate models is a huge pain in the cumulonimbus. Here’s why: They’re constantly changing shapes; they’re made of microscopic particles but can collectively extend for hundreds of miles; individual droplets can interact in fractions of a second, yet climate predictions span hundreds of years; clouds both reflect sunlight and absorb heat; and since they cover about 70 percent of Earth at any given time, they definitely play a role in climate change. But at the same time, climate change influences their formation by messing with global temperatures and precipitation patterns.

Grist thanks its sponsors. Become one.

And all of this means that they’re very difficult to simulate on computers. Global climate models have to break the planet up into big chunks between 60 and 300 miles wide due to limited processing power, so they can’t capture the small time and space scales on which clouds operate. Which leaves us with a lot of open questions — How are clouds influencing the melting of the Antarctic ice sheet? What about Greenland? Does ocean acidification affect cloud formationWill rising temperatures mean fewer clouds? Will fewer clouds mean more warming?

So back to the big new study, which just adds to the pile: Scientists may have misjudged how effective a certain type of cloud is at slowing climate change.

According to the study, which uses data from the Calipso satellite, “mixed-phase” clouds contain less ice than we once thought. So as the planet warms, these clouds have less ice available to turn to water. And this is a problem, because water is good at reflecting sunlight away from earth, and we were kind of banking on this whole ice-to-water transition as a natural way to slow warming.

As a result, the researchers conclude, a doubling of CO2 in the atmosphere compared to pre-industrial levels could lead to a temperature rise of between 5 and 5.3 degrees C, compared to previous predictions of between 2 and 4.6 degrees.

Grist thanks its sponsors. Become one.

Yet, even the study’s lead author, Yale graduate student Ivy Tan, told Climate Central that the temperature rise predicted from their model should be taken “with a grain of salt” and that more models need to corroborate their finding.

Gavin Schmidt, the director of NASA’s Goddard Institute for Space Studies, told The New York Times that “headlines that scream ‘Scientists say sensitivity higher than thought!’ will not be justified” and that “this is one extra ingredient that needs to go into the hopper.”

A hopper full of ice, water, wind, and uncertainty.

It’s going to be a while before we fully understand the role of clouds in climate change. For now, people are calling for advanced computing power to run more fine-grained models that can better account for clouds. And last year, the National Academies of Sciences held a “classified workshop” with members of the intelligence community to discuss how “national collection systems” (i.e. satellites) could help better understand clouds.

That’s right — we’re asking government spies to help us understand clouds. So be wary the next time you read a headline about clouds and climate change. Chances are, things aren’t as simple as they seem. And remember — that isn’t a teddy bear in the sky; it’s a demon made of water vapor and ice crystals, and it’ll screw us over if we let it.