Guide to the Cosmos

Making the Wonders of our Universe Accessible to Everyone


     Newsletter: Exoplanets: Headlines, Hype & Hope


Public and scientific media recently headlined spectacular claims of discovery of billions of other Earths, some “just around the corner.” But is this hype, hope, or reality?


The spectacular claims include:

  • Astronomers found 647 Earth-like planets
  • 25 Billion habitable planets exist in our galaxy
  • Water was found on five exoplanets
  • Earth 2.0 may be only 12 light-years away


Some context is important. In May, the $550 million Kepler Space Telescope, which was marginally able to detect Earth-like planets in the habitable zones of Sun-like stars, failed before gathering definitive data. Scientists now want a more capable, and more expensive, new satellite. Is funding more likely if habitable planets are extremely rare or “just around the corner?” In Big Science as elsewhere, sizzle sells.


To understand what’s real, I read the original scientific publications in the Proceedings of the National Academy of Science (PNAS), the Monthly Notices of the Royal Astronomical Society, and the Astrophysical Journal. I also analyzed data on Kepler’s website and NASA/Caltech’s archive of exoplanets “officially confirmed” as of Dec. 1, 2013.


My conclusions are:

  • Planets may be as numerous as stars.
  • Recent claims are more spectacular than the evidence.
  • What some call Earth-like is far from habitable.
  • Earth is remarkably rare, but probably not unique.
  • We are nowhere near finding a true Earth 2.0.


While Scientific American;s website said Kepler found 647 possible Earths, Kepler’s website actually lists only 167 confirmed planets of all types. Only 13 are within 30% of Earth’s size and all those have surface temperatures above 260°F. Anyone who calls that Earth-like should be exiled there.


The Hubble Space Telescope did find water vapor traces on five exoplanets. All these were “hot Jupiters” roasting in 400 to 2500 times Earth’s sunlight.


The key paper was by Erik Petigura et. al. in PNAS. Given Kepler’s meager data, this is fine research, but certainly open to scientific challenge. They state: “Detections of Earth-sized planets in [Earth-sized] orbits are expected to be rare in this study…Indeed, we did not detect any such planet…” (My emphasis) Without seeing any, how can they claim billions exist? Never underestimate clever people with powerful computers and agendas.


In brief, they define Earth-like far too broadly, make bold swags, and extrapolate liberally. (Swag is a technical term for Scientific Wild A** Guess.)


Extrapolation is the art of estimating the unknown from the known. For example, imagine having survival data for Americans, but only for those under 50 years old. The dark blue line in the graph below shows the fraction of Americans still alive at each age up to 50. To estimate our life expectancy with only this limited data, we could assume the trend from 15 to 49 continues indefinitely. The red band extrapolates that trend, and shows that 77% of us will still be alive at age 200. How wonderful! The nearest bicentennial man must be “just around the corner.” Alas, reality is the light blue line.



Clearly, extrapolating can yield false conclusions. It can work if one understands the underlying physics and uses an appropriate mathematical model. We do not, however, have an adequate model of planet formation and abundance — indeed our ignorance is why we launched Kepler in the first place.


Petigura separates detected exoplanets into “bins”, each spanning a range of planet radius, orbital period, and intensity of host star’s light hitting the planet. For each bin, Petigura computes the odds of detecting a planet with that bin’s parameters. With modern computing power, binning is a crude approach; one should compute odds for each detected planet instead of averaging over very broad ranges.


For the “Earth” bin, the computed detection odds are 1 in 585. For each detected exoplanet in that bin, Petigura claims 585 actually exist, a huge extrapolation that may well be far off. The detection odds depend critically on the radii of the star and the exoplanet’s orbit, neither of which is actually measured. Kepler doesn’t take “pictures” of stars or exoplanets; it only measures dips in starlight intensity, which astronomers interpret as “transits”, exoplanets passing in front of a star and blocking a tiny portion of its light. The most important quantities, planet size and orbit, are estimated.


Petigura detected 8 exoplanets in Earth’s bin and claimed 4673 others exist but were missed, and thus “22% ± 8% of Sun-like stars” have planets like Earth.


Petigura says Kepler’s data is “challenged by correlated, non-stationary, non-Gaussian noise.” While scientists understand and can model simple noise, one must wonder if Petigura can properly model this more complex, ill-defined noise. They also say 28% of their initial exoplanet candidates were actually eclipsing binary stars. It’s normally easy to distinguish stars from planets. This 28% error rate shows how challenging and uncertain their analysis is.


Petigura defines the “Earth” bin as planets whose radii are 100% to 200% of Earth’s and whose starlight intensities are 100% to 400% of Earth’s. Many fear the dire consequences of a 1% rise (3°C) in Earth’s temperature — imagine 300% more sunlight.


Reducing the upper limit on starlight from 400% to 200% drops the number of Earth-like exoplanets from 8 to 2, and their abundance from 22% to 5.5%. Reducing that limit to 186% leaves only 1 Earth-like exoplanet and the extrapolated abundance drops below 3%. It’s disturbing that the final result depends so critically on a questionable definition.


Petigura’s Earth bin is far too broad. The NASA/Caltech archive, the field’s gold standard, lists 935 confirmed exoplanets from all sources including Kepler. Of the 296 exoplanets with measured masses and radii, 19 are within 100% to 200% of Earth’s radius. Only 2 of those are less massive than Neptune, an uninhabitable gas giant. Thus 90% of exoplanets in Petigura’s radius range are not Earth-like based on real unbiased, unextrapolated data.


It’s harder to set a firm starlight intensity limit. Here are some choices. Of 935 confirmed exoplanets, 365 have known starlight intensity. Comparing with Earth: only 1 is within ±10%, 5 are within ±20%, and 16 are within a factor of 2.


Furthermore, all the above only addresses two planetary characteristics: size and surface temperature. In my books, I list 14 remarkably special characteristics of Earth. For each characteristic currently quantifiable with real data, Earth is better than 90% to 99.5% of all known planets. If the odds of Earth 2.0 were 10% raised to the 14th power, there would be zero other habitable planets in our galaxy.


Without definitive data, all we really know is that the number of habitable planets in our galaxy is between zero and 25 billion.


‘Tis the season to rejoice that we’re on the only known habitable planet.



Very best wishes for the Holidays and a Happy New Year,



December 17th, 2013