One of my many ongoing projects is working on how we in the near-Earth object (NEO) observations community* can improve the clarity of our communications about NEO impact risks, hazards, and threats.
A key part of improving these communications is coming up with better ways of characterizing the uncertainty that is in inherent in mathematical/statistical predictions. It’s a challenging task. What we’re talking about here is predicting the future orbital movements of asteroids that may come close to Earth – that is, NEOs.
Now, consider that “the future” is uncertain, and thus “predictions” about the future are uncertain. Also consider that “near” and “close” are relative terms.
By NASA’s definition, a near-Earth object is an asteroid (or comet) whose orbit periodically brings it within approximately 195 million kilometers (121 million miles) of the Sun – that’s within about 50 million kilometers, or 30 million miles, of Earth’s orbit. A potentially hazardous asteroid (PHA) is an asteroid whose orbit is predicted to bring it within 0.05 Astronomical Units (just under 8 million kilometers, or 5 million miles) of Earth’s orbit; and of a size large enough to reach Earth’s surface – that is, greater than around 30 to 50 meters.
In technical terms, a “close approach” is a predicted event in which a NEO passes within the orbit of Earth’s Moon. Some passes of larger NEOs close to the Earth-Moon system but not between the two bodies are also called close approaches. In lay terms, a “close approach,” while not a risk, hazard, or threat to Earth, is an event that scientists want to keep an eye on.
Are we clear so far?
(Just FYI: in searching Google Images for a visual depiction of uncertainty, I came across a picture of a road sign reading “doubt and fear just ahead,” another road sign reading “lost confused unsure unclear perplexed disoriented bewildered,” and a quote from Voltaire: “Uncertainty is an uncomfortable position. But certainty is an absurd one.”)
In working on the challenge of improving communications about uncertainty, I’ve just read a book by Dylan Evans, Risk Intelligence: How to Live with Uncertainty (Free Press, NY, 2012), that’s provided some good food for thought.
What Evans calls “risk intelligence” is “the ability to estimate probabilities accurately.” (You can take his simple “risk intelligence quotient,” or RQ, test online.)
Evans by no means advocates the abandonment of probabilistic risk assessment. He argues that we need to better understand, and develop better ways of explaining, probability. “Probabilities are an expression of our ignorance,” he observes; “by quantifying uncertainty, we are already conceding that we don’t ‘know’ the relevant facts with 100 percent certainty and admitting that we have to work on the basis of educated guesses.”
Numerical characterizations of risk are more accurate than verbal descriptions, he says. However, “replacing verbal labels with numbers is not enough by itself to make things any better.” While verbal labels are imprecise, and numbers may be less so, even numerical labels can mask uncertainty. “It is not enough to supplement verbal labels with numerical translations; the labels should be dispensed with altogether, because people tend to ignore the numerical translations and interpret the labels in their own idiosyncratic ways.” Imprecise statements about uncertainty can lead to miscommunication between analysts and policymakers [I’d say “experts and non-experts”], he notes, and for that matter, among analysts themselves.
This last point gets to one of my favorite “problems” in science and risk communication – the all-too-common use of subjective terms such as “serious,” “severe,” “likely,” “unlikely,” “significant,” “important,” and so on. These kinds of words are subject to interpretation. Does “likely” mean 50 percent probable, 90 percent probable, 99 percent probable? What would render the prediction of a possible – emphasis on possible – asteroid impact with Earth 20 years from now “significant”? It depends on who you ask.
By the way, according to the asteroid impact risk table maintained by the Jet Propulsion Laboratory’s Center for NEO Studies (CNEOS) and updated daily, no known NEOs pose any risk of impacting Earth over the next 100 years. Also by the way, CNEOS has reported that an upcoming asteroid flyby originally characterized as a possible close approach is no longer predicted to be close. Based on early observations, asteroid trackers predicted that an asteroid named 2013 TX68 would pass by Earth on March 5, at a distance as far out as 9 million miles (14 million kilometers) or as close as 11,000 miles (17,000 kilometers). Further observations enabled trackers to eliminate some of the uncertainty surrounding their calculations and refine their prediction. They now report that 2013 TX68 will pass by Earth on March 8 at a distance of about 3 million miles (5 million kilometers). “There is still a chance that it could pass closer, but certainly no closer than 15,000 miles (24,000 kilometers),” according to CNEOS. (So uncertainty has been reduced, but not eliminated….)
Now back to Professor Evans for some final thoughts: “The very tools that can help enhance risk intelligence – the mathematics of probability theory and the collection of statistical data” – can lead to overconfidence, “a sense that we can master Lady Luck [let’s say “chance” or “randomness”] through science. But as the financial crisis of 2007-2008 reminded us, the models we build to manage risk are always fallible…. If such thoughts make you panic, then you haven’t come to terms with the irreducibility of chance.”
* My work is funded in part by NASA’s Planetary Defense Coordination Office, which encompasses the agency’s NEO Observations Program.