Tuesday, July 27, 2010

Has Kepler Found hundreds of Earth-like planets?

Update (2pm PDT): As pointed out by Daniel Fischer in the comments, the news articles my post is responding to are poorly-reported rehashes of information that came out in the middle of June.  The numbers in the graph that have led to the news articles seem to have been published publicly in this scientific article; the graph in the Sasselov talk was labeled differently but the data seem to be the same.  I hereby retract all implied and explicit criticism of Dr. Sasselov in the original article and apologize to him and the Kepler team for not doing my research on that side of the issue.  I have crossed those statements out, but leave them in as penance. I stand by my scientific statements.

In the past few hours, many news outlets have posted stories claiming that NASA's Kepler Mission has discovered hundreds of Earth-like planets. Here's a reasonable version of the news posted at SpaceRef.com.

These stories are not based on any (new) official announcement, but rather a talk given by one of the Kepler mission scientists at a meeting in Oxford a few weeks ago.  During the talk, mission scientist Dimitar Sasselov showed a slide giving the number of candidate planets discovered so far by Kepler as a function of their radius, and the Earth-sized planet bar is near 150.

According to Kepler's Twitter feed, the team is formulating an official response to this leak.  I'm not on the team, so I don't know what they will say, but let me make a few guesses as to what you can and cannot take away from this news leak:
  1. These are planet candidates, not confirmed planets.  There are many known astronomical objects that can mimic the signals Kepler is looking for.  It takes lots of follow-up work to rule out these false alarms.  Much of the Kepler team is working hard to rule out these false alarms, and a lot of ground-based telescope time this summer is going into these studies.  I'd be willing to wager that the team has a decent idea on the fraction of their candidates that are false alarms, and it's possible that fraction was figured in to Sasselov's graph.  However, since a proven Earth-sized planet would be a huge score for the Kepler team, and we haven't seen any reviewed articles announcing such a discovery.  To me, that says the team feels there is still more work to be done to prove any given case.
  2. Even if they are real, all we know about the planets is their diameter and how far away they are from their parent star.  Kepler works by looking for a drop in the light from a star caused by a planet orbiting between us and the star.   At first, all we know is what fraction of light is blocked by the planet.  Some simple observations of the parent star (many of which were done before the mission, and some of which are done once a candidate planet is flagged) give us the radius and mass of the parent star.  Geometry then allows us to calculate the radius of the planet, and laws derived by the scientists Johannes Kepler and Isaac Newton then allow us to determine how far away from the star the planet is.  But that's all we can be reasonably certain about at the present, at least for Earth-sized planets.  We can guess at the mass of the planet if we assume it is made of rock, and we can estimate how hot its surface is (though this fails dismally for Venus in our own Solar System), but that's it.  We have no way of knowing if the planet has water, continents, or life.  The planet could be hellish like Venus, barren like Mars, or very hospitable like the Earth.
  3. Any real planets discovered so far are so close to their parent star, they are likely very, very, very inhospitable places.  In order for Kepler to have a strong planet candidate, the mission scientists need to see it go in front of the star at least three times.  Once tells you that something might be happening, twice gives you a guess at the length of the planet's "year", and three times confirms that length of the orbit.  Earth takes one year to go around the sun, and Kepler has only been working for a little over a year.  So a true Earth-like planet will only have made one or, in a few lucky cases, two passes in front of its parent star.  Any Earth-sized planets that have been discovered must be closer to their parent star than the Earth, and probably closer than the planets Mercury and Venus.  That means they will likely be hotter than the Earth, and not capable of supporting life. 
    Now, if the parent star is smaller than the sun, the planet could be closer and yet still have a climate appropriate for life.  But that information is not yet public, so I can't guess at what fraction of these planets may have temperatures similar to the Earth.
  4. The numbers of hundreds of planets are probably right.  This is just an educated guess on my part, but astronomers don't put numbers on slides if they suspect the numbers are wildly wrong.
  5.  I'm glad I'm not Dimitar Sasselov.  Whether or not he had permission from the team to include these slides in his talk, you never want your name attached to an unintentional media frenzy.  And if he didn't have permission, teams like this almost always have members sign contracts listing possible penalties for spilling the beans.
So, in short, the only thing I'd feel comfortable taking away from this furor is that Kepler is finding lots of potential Earth-sized planets.  But these are not confirmed planets, any real planets are likely closer to their parent star than Mercury and Venus are to our sun, and we have absolutely no clue if any of these planets are capable of supporting life, let alone if life actually exists on any of them.  Again, any real planets found by Kepler are Earth-sized planets, and NOT NECESSARILY Earth-like planets. (As I tell my students, underlined, italicized, capitalized and bold print means I absolutely mean what I say, and I think it's important, and it will probably be on a test.)


  1. Was this really an unauthorized leak? It was a TED talk, not some small conference.

  2. Good rebuttal of the false stories - but the main point is missing: This was - except for that one histogram - all old news! Back in June when the two papers on the 300+ non-secret candidates came out it was already reported (e.g. online by Scientific American and Sky & Telescope) that the 400 withheld candidates were either circling around bright stars (which could be only a minority) or had diameters smaller than 1.5 Earth diameters (which would have to be the majority, i.e. hundreds of candidates). Some of the bodies were even said to have just one Earth diameter but - as you rightly point out - their masses aren't known a bit. The overreaction to the Sasselov "leak" which wasn't one by numerous science 'journalists' casts a glaring light on the state of this profession. Sigh ...

  3. Excellent point, Daniel. Looking into it further, the numbers on Sasselov's graph that are causing the furor seem to match the numbers in a graph in the paper by the Kepler team that coincided with the June 15 data release (See Figure 2 in http://arxiv.org/abs/1006.2799 ). So it appears that Sasselov didn't do anything by accident other than unintentionally start a media frenzy many days after the fact.

    I'll second your point on the state of science journalism and whip myself 15 times with a wet noodle for not checking that preprint prior to publishing. It's what I get for trying to toss off a post in 15 minutes on a busy day.