Is there a planetary system around tau Ceti?

If you heard about the (proposed) tau Ceti planetary system late last year, then chances are you would believe that there are five planets orbiting this bright star with 100% certainty (note the “starchild” video embedded in this story), and that at least one of them was habitable. The truth – as always – is a little less clear.

I’m a member of the team that published the research, but I was really involved with the observations rather than this particular analysis. The result comes from a new technique developed using Bayesian statistics, a technique that we’re still working to validate. This means that any detection could only ever be at “candidate” status.

So why would we put out a press release on a non-conclusive result? The answer has two parts: one was to keep control of the story; the other, to make sure a competing team did not steal any potential “glory”. (Sadly, in science it is often better to be first than right.)

These days many astronomical papers get uploaded to a web server called arXiv after they have been accepted for publication, but before they have actually been published. There are some science journalists who trawl through the papers uploaded to sites like these looking for interesting stories. We wanted to avoid this, and the chance of the story going in an uncomfortable direction.

We did our best to emphasise the preliminary nature of the result in the press release. The headline was: “New neighbours? Closest single star like our Sun may have habitable planet“. Note the word “may”: this tended to get lost in the reporting of the story, especially in tabloids and non-science websites.

One tricky aspect of this story is that in any team of researchers, some will find the result more compelling than others. As such, some of the quotes in the release sound more convinced of the result than others. The message became somewhat mixed, so the reporting often picked up the more conclusive sounding comments.

I admit to being one of those who find this result less compelling than others I’ve seen, and although I was not quoted in the text, I was listed as a media contact in Australia. When asked for an interview by the ABC’s PM program, I happily gave one, and included the caveats about the technique and the uncertainty. The ABC journalist Jennifer Macey, to her credit, reported them, although the headline is far too definitive.

In Sky & Telescope, an astronomer from one of the competing teams expressed his doubts about the result, commenting that the result was “only a first guess”. I think that calling it a “guess” is a little harsh, but I would say that is only a first attempt at analysing and interpreting what is a large and tremendously complicated set of observations.

The tau Ceti result is actually reminiscent of the “planet” announced orbiting alpha Centauri B earlier in the year, in that it requires independent confirmation to be fully accepted. The press release in that case also made the result sound far more conclusive than it really is. Looking at the paper, the data is very far from conclusive (and I’ve said so elsewhere). Without going into too much detail here: all stars are complex beasts; the team claims to understand them; I don’t believe they do.

In the end, this has taught me one thing: you can’t really control how your science gets reported. You can do your best, but media outlets need to make money, and to do that they need the most exciting news. “We might have found five planets orbiting a naked eye star” can never compete with “Brightest Sun-like star has habitable planet!” Perhaps we should have stayed a little more “on message” in the press release, I’m not sure.

And all this not long after I wrote about the manipulation of headlines to garner attention for scientific results…