Moonshine and Lunacy

I got an email from a reader yesterday asking for help in understanding a video that she’d seen, in which a citizen-scientist performs an experiment with a very surprising result: moonlight makes things colder! How could this be? To find out, I took a dive into the well-intentioned but deeply problematic world of Youtube science.

The first thing that I discovered on this journey was that this experiment is a popular one, particularly among “Flat Earth” theorists. (I wish I were kidding about that, but the idea of a flat earth seems to have experienced a resurgence in popularity this year.) A quick search for “Moonlight cold” turned up multiple videos on Youtube, most of which show some variation of the following experiment being performed:

An object is placed outside under a full moon, partly in the shade and partly in direct moonlight. The experimenter gives the thing some time to reach equilibrium, then takes a temperature measurement with an infrared “laser” thermometer, which reveals the moonlit side to be lower in temperature, sometimes by as much as a few degrees!

Obviously, something’s afoot here, but it’s not so simple as a hoax—nearly everyone making these videos is very much convinced that they’re seeing a real effect. So how does this phenomenon arise? While you probably don’t need convincing that it’s not “heat-sucking moon beams” at work, it’s absolutely worth exploring how these earnest attempts at science can go so wrong.

A handful of these videos are shot in near-complete darkness, as even a full moon can’t provide enough light for the average cell phone camera, so practically the only discernable features in the footage are the thermometer’s laser aiming dot and backlit temperature display. Still, I admire the creators of these videos for their effort to reduce the noise in their data and provide a cleaner signal; noise is one of the biggest sources of error in any experimental setup. However, these home experiments fall victim to a far more significant source of error: poor instrumentation.

There are two factors common to virtually all these videos: a point-and-shoot infrared thermometer, and a yokel behind the camera who has no idea how the device he’s holding works. See, while it’s tempting to assume that your “laser thermometer” is telling you the temperature at the illuminated spot, that’s only half the story—the laser actually doesn’t do anything besides help you aim! The actual measurement apparatus uses a lens to focus infrared radiation from your target onto a digital sensor. What this means is that, rather than taking the temperature at a point, you’re taking the temperature of a circle centered at that point. How big the circle is depends on how far away you are from your target, along with the internal optics of the thermometer.

A standard IR thermometer has a D:S (distance to spot) ratio of roughly 10:1, meaning that making a measurement from ten feet away will give you the average temperature of a spot one foot in diameter. In the video below (which has twenty-something-thousand views, as of this writing), the experimenter seems to be standing at least ten feet away from his target, trying to check the temperature on either side of a wallet. Once you understand how the IR thermometer works, it’s plain to see where his anomalous result comes from: the spot he’s measuring on the moonlit side includes the ground beyond the edge of his pool deck, which is doubtless cooler than the stone that makes up the patio.

You’ll want to skip to about 8:50, unless you’ve got time to kill.

The potential for this kind of error is why it’s important to repeat your measurements in different ways; if he had moved the wallet to the other side of his table’s shadow, he might have noticed that the temperature difference vanishes!

This brings me to my last point. The beast that aggregates these errors into a katamari of colossal wrongness (and the real reason so many people are apparently convinced that the moon emits some kind of anti-photon) is one of modern science’s greatest enemies: reporting bias.

If the guy who made that video had noticed, halfway through filming, that he wasn’t getting the results he expected, do you think we’d have seen it? Statistically, for every person who finds that moonlight makes things colder, there’s at least one who conducted the same experiment and found that it has no effect, or even the opposite, making things warmer! (Which technically it should, but it’s likely a micro-kelvin kind of difference.) The difference between the two cases is that the null hypothesis is obvious, while the alternative is literally incredible, so only the latter gets uploaded and viewed and shared. When you consider that the uploader gets paid by advertisers based on the number of views they get, you realize that this system incentivizes bad science, not just socially but financially.

Unfortunately, this problem isn’t exclusive to Youtubers howling at the moon—it permeates the real world in serious ways. If a flipped coin land on “heads” ten times in a row, it’s very reasonable to suspect that it’s a weighted coin, but it becomes a different story entirely if you find out that someone spent all afternoon flipping in order to get that ten-heads streak. In the same way, unreported repeat trials can make a statistical fluke seem like good data, which can be a huge issue if we’re talking about something like the efficacy of a nutritional supplement.

As someone who loves unconventional theories, I can say from personal experience that there’s a deeply forlorn feeling you get when an idea that you thought was revolutionary turns out to be bunk; it’s like waking from a flying dream to discover that it’s Tuesday and you’re late for work. People fight that waking feeling, usually subconsciously, because our beliefs are guided by emotions rather than logic an embarrassing amount of the time, and because everyone wants to be the new Galileo. But science—GOOD science—is the struggle against that instinct to cling to the dream as it dissolves around you, because as Randall Munroe said, “You don’t use science to show that you’re right. You use science to become right.”

The null hypothesis, that the moon just reflects light from the sun, will never “go viral”. It’s not profitable, or terribly exciting, and it doesn’t have the symmetrical allure of its contrived alternatives. But precisely for those reasons, it’s essential that it be defended, that we demand extraordinary evidence from those who make extraordinary claims, and with zeal to match theirs. It’s a scientist’s duty to shake awake those dreamers who’ve taken in a touch too much moonshine, because there’s a real world out here that needs exploring, and it’s full of more wonder than anyone could dream up.

You may also read these articles