A Question of Phosphine
Remember all the excitement last month about the claimed detection of phlogiston phosphine in the atmosphere of Venus?
Well, a wet blanket appears to have been thrown over it by a new paper on the arXiv by Snellen et al. Here is the abstract:
The conclusions are very clear, but the paper hasn’t been refereed yet. Let’s see what the authors of the original work have to say. For myself, I think a proper (i.e. Bayesian) analysis of the data is called for…
I should also mention in this context another paper on the arXiv from a few days ago, which uses a null detection to place an upper limit on the phosphine abundance.
Note that one of the authors of this second paper is Jane Greaves, who was on the original discovery paper.
Follow @telescoper
October 21, 2020 at 11:59 am
*12th order* polynomial background subtraction?!
October 21, 2020 at 5:56 pm
Quite. With 12 parameters you can fit everything but predict nothing. At least we can be confident that we are not in imminent danger of invasion by Venusians now.
William of Ockham must be turning in his (unmarked) grave.
October 22, 2020 at 8:19 am
With four parameters I can fit an elephant, and with five I can make him wiggle his trunk.
(Attributed to von Neumann by Enrico Fermi, as quoted by Freeman Dyson in “A meeting with Enrico Fermi” in Nature 427 (22 January 2004) p. 297)
https://en.wikiquote.org/wiki/John_von_Neumann
October 21, 2020 at 12:13 pm
Hang on. This was published in Nature…with a 12th-order polynomial background subtraction that has no physical underpinning at all other than spuriously “improving” the SNR? Did none of the referees suggest that this might be a somewhat optimistic — and other words beginning with “op…” — approach to take to data analysis?
Fig. 2 of the first paper to which you refer, Peter, brings this out very nicely, but it’s hardly a staggering revelation when it comes to data treatment. That’s lazy refereeing, at best.
October 21, 2020 at 12:23 pm
As you say, it was published in Nature…
October 21, 2020 at 1:42 pm
Indeed. It would be interesting to do the statistics: if it is in Nature, and it is sensational, is it probably wrong?
October 21, 2020 at 2:52 pm
Ouch!
October 21, 2020 at 9:01 pm
It’s not quite as bad as this: https://muircheartblog.wpcomstaging.com/2018/10/24/how-not-to-do-spectral-analysis-101/
…but it gives it a damn good run for its money.
October 22, 2020 at 8:14 am
“Indeed. It would be interesting to do the statistics: if it is in Nature, and it is sensational, is it probably wrong?”
One I remember was convincing observational evidence that we live in an Einstein-de Sitter universe. Shortly after the paper appeared, at least three other papers appeared, all rebutting it, but for different reasons. In other words, it had at least three major shortcomings. None of these rebuttals were published in Nature, and I doubt that Nature would have published them had they been submitted there. (My only real rejected paper was a short rebuttal, submitted to Nature, to a sensationalist claim in Nature which is now known to be false. I now realize that something like that would probably never be published in Nature.)
October 21, 2020 at 1:56 pm
A good example as to why, at least for accepted papers, the names of the referees should be published along with the paper.
October 21, 2020 at 8:59 pm
I’m uncomfortable with “forcing” referees to reveal their identities, Phillip. For one, it will cut down dramatically on the number of scientists willing to referee. But on the other hand, I agree that if it were the norm to put our name to our comments, the peer review system might be a little more robust.
October 22, 2020 at 7:51 am
Note that I suggest this only for accepted papers, so no referee need fear retribution for having rejected a paper. But why should the referee object after acceptance? The only reason, it seems to me, is the fear that he let through a paper which was not good enough, but that is the whole point of refereeing, so those potential referees who would object are probably not doing a good job anyway. Or am I missing something?
(I’m ignoring the possibility that the referee approves a good paper and fears retribution from those who just don’t like the result of said paper.)
The above applies to why a referee might not like this idea. In the larger picture, it would make it clear who does how much refereeing, without making the reasons clear why some do more than others, so that might be a reason to keep it secret.
October 21, 2020 at 1:43 pm
Interestingly, a former colleague is on the original paper, and another former colleague on the first of the two papers above.
October 21, 2020 at 10:27 pm
Nature Astronomy and Nature are not the same thing.
October 21, 2020 at 11:29 pm
Astronomy and Nature Astronomy aren’t the same thing either…
October 22, 2020 at 7:45 am
I once went to a talk announced as “The Nature of High-Velocity Clouds”. Shortly before the talk, the corresponding paper had been accepted, so that title was changed to “High-Velocity Clouds in Nature“. 🙂
October 22, 2020 at 8:40 am
I have found the refereeing in Nature Astronomy as good as any of the major astronomy journals. I think you are wrong in attacking a journal because of what you think of a relative. As for the result, perhaps give the original authors a chance to respond?
October 22, 2020 at 2:49 pm
I very much dislike the how the Nature Publishing Group tries to force its way into more and more fields by establishing Nature this and that journals. The journals are, of course, very expensive and because of the Nature glamour at least deans and such think they’re the journals everyone should aim for.
A few years ago there was a lot of talk about boycotting Elsevier journals, but I thing the Nature Publishing Group is just as bad if not worse.
October 22, 2020 at 3:12 pm
Easy solution: dock points off of applicants who have published in such journals.
October 22, 2020 at 9:49 am
The only reason, it seems to me, is the fear that he let through a paper which was not good enough..
I agree with your arguments in the main, Philipp, but I’ll just note that referees are not exclusively male…
October 22, 2020 at 10:07 am
Isaac Asimov (I’ve read most of his books) had a book called The Intelligent Man’s Guide to Science and was asked if the title wasn’t rather sexist. Without missing a beat, he quipped “No; I’m the intelligent man”. 🙂 The title was actually later changed to Asimov’s Guide To Science. (To be sure, the reason was that the second edition was The New Intelligent Man’s Guide to Science, and The New New Intelligent Man’s Guide to Science wasn’t a good title for the third edition. Since I don’t think that he would have quipped “I’m the new intelligent man”, the original title probably was sexist, but at least it provided a good setup for the comeback.)
Seriously, I take your point, but I note that increasingly “she” is being used to refer to unnamed individuals whereas in the past it was “he”. Not much better, and two wrongs don’t make a right.
Must practice the singular “they” (which goes back at least to Shakespeare’s time).
October 23, 2020 at 3:47 am
[…] A Question of Phosphine links to Re-analysis of the 267-GHz ALMA observations of Venus: No statistically significant detection of phosphine. This team re-analyzed the same data used in the first Phosphine gas paper, but came to the opposite conclusion. So it comes down to a very complicated question of statistical analysis. […]
October 28, 2020 at 12:57 pm
[…] continue the ongoing saga of Phosphine in the atmosphere of Venus there’s a very strongly worded paper on the arXiv […]
February 8, 2021 at 12:09 pm
[…] detection of Phosphine in the atmosphere of Venus? If you do you will remember that the claim has been contested. Now there is another paper on this matter claiming that what was interpreted as a spectral feature […]