Potentially major ALMA array discovery? Options

[vikingmars]

Well, WISE has ruled out brown dwarfs within the solar system. Here's a SciAm article about the whole thing.

Yes, thanks a lot Emily for this much interesting article about WISE.
But they say also inside about their WISE methodology that "(...) As a result, the W3 and W4 bands may reach a solar companion at 1–2 MJup out to greater distances than W2, although only ∼20% of the sky was mapped twice in those bands at their full sensitivity, which is necessary for a survey based on parallactic motion."

[Herobrine]

Well, WISE has ruled out brown dwarfs within the solar system. Here's a SciAm article about the whole thing.

Luhman only suggests that the constraints indicate that there "probably" isn't a brown dwarf in the outer solar system. I wouldn't interpret that as ruling it out. In the Scientific American article you linked, when Billings says it is ruled out, he seems to be repeating Clara Moskowitz's questionable interpretation of Luhman's findings, that they "ruled out" anything. In truth, Luhman simply didn't find evidence of a brown dwarf companion from his search, which is very different from determining that there can't be one.

I'm no expert in radio astronomy; I do think that Liseau et al. are trying really hard to see what they want to see, though. What they're looking at seems like it could be just about anything, anywhere, two different things, or even nothing.

[JRehling]

This story reached the Washington Post:

https://www.washingtonpost.com/news/morning...omepage%2Fstory

Unfortunately, it seems to be a case of bad science reaching the public, rather than good.

I'm heartened to see two professional astronomers calling out (among other things) the same thing that I did: What they detected are "blips" – deviations from the background noise that can readily happen as a matter of chance. It's all a matter of SNR. Spurious signals with SNR of 3.6 can and must happen quite often in the pixels of a sky survey.

An enormously useful principle I learned from Kepler analysis: Three observations of a possible dynamic object moving in one-dimensional parameter space are vastly more compelling than two. Two blips in different locations might be a moving object – but they might be two blips. You have an unconstrained degree of freedom in the delta of time and – in this case – spatial location. If you had three detections, then that degree of freedom goes away. Then, it becomes really convincing if you see the object has the same rate of motions between detections 2 and 3 as it did between detections 1 and 2. (Or, for Kepler, the same time interval between detections 2 and 3 as between 1 and 2.) This enormously reduces the probability of random noise posing as detections of a real object. (If an object moving across a visual field covered a significant portion of an elliptical orbit – as if we were seeing on object in the Alpha Centauri system as opposed to in the solar system – then the deltas would not the same, and then you don't actually have a good constraint with only three observations, because an infinite number of elliptical orbits are possible. You'd need four or more observations.)

This case never should have been web-published (in this form) and it won't pass peer review. If they had two observations with an SNR of 20 or three with an SNR of 7, that would be compelling. Two with an SNR<4 is simply nothing. It's like finding a footprint in London and saying you've solved the Jack the Ripper case.

[alan]

From the SciAm article:

“Normally I prefer to only upload accepted papers,” says Wouter Vlemmings, an astronomer at Chalmers University of Technology in Sweden and co-author on both studies. “This time, however, we had exhausted our ideas. … With the arXiv upload we specifically wanted to reach the community that could tell us if we overlooked something, in which case we fully intend to withdraw the papers….

Reminds me of the FTL neutrinos from last year.

[mcaplinger]

The signal-to-noise ratio for the two "detections" are about 3.6 and 2.1...

Where does it say that? Table 1 shows S/N of 21.4 for one observation and 4.3 for the other.

[JRehling]

Where does it say that? Table 1 shows S/N of 21.4 for one observation and 4.3 for the other.

My mistake; I was reading the delta flux.

S/N of 21.4 is indeed something rare. 4.3 isn't. So we have one observation that might be something real… But there are other explanations possible. Kepler had anomalies due to inconsistent performance of the electronics. So an S/N that was seemingly real would be reported when it was a miscalculation: The noise was actually much higher in certain situations than typical situations, so real S/N was much lower and therefore much more likely to happen by chance than originally calculated.

I can't assess ALMA's data pipeline performance, but what we have is one noteworthy observation. Whether that occurred in the instrument or with a real object in space, I can't say.

[ngunn]

Reminds me of the FTL neutrinos from last year.

The difference I see is that whereas those were inherently implausible the ALMA observation is not. The paper makes it clear that there is a significant region of size-distance-temperature space that only ALMA can observe. There's nothing inherently implausible about objects populating that region.

[JRehling]

There's nothing inherently implausible about objects populating that region.

True, it's plausible that a real object would show up like this. What's implausible, as Mike Brown noted, would be that one super-narrow FOV observation would just happen to turn one up near Alpha Centauri. It's either incredibly lucky (hard to believe) or the outer solar system has lots of previously-unknown big objects (impossible).

[Explorer1]

Stranger things have happened: what are the chances of a comet flying ~100,000 kilometers from Mars in a human lifetime? Yet just last year it actually happened, and with instruments able to observe it from both Earth and Mars! There are so many possible low probability events and coincidences, they happen all the time. Though I agree it's still too early to be certain what ALMA found.

[Spock1108]

how many chances Galileo Galilei had to see Neptune, while he was in conjunction with Jupiter? Everything is possible ... then its proximity to Alpha Centauri would have masked WISE ... we can not rule anything!

[JRehling]

Yes, events occur despite the fact that the probability of the event, as calculated after the fact, was very low. But there's a critical difference between saying so first and after the fact.

And science is not new to this idea. A good methodology is to create a Monte Carlo model of possible explanations and see which one is most likely given the observation. There are many explanations that could, in principle, explain this data. But the burden is on the researcher to vet the possibilities carefully, not just mention one or two theoretically possible explanations.

[ZLD]

"the burden is on the researcher to vet the possibilities carefully, not just mention one or two theoretically possible explanations. "

The intent of the paper, as I understood it, was that after compensating for likely errors in the data, they felt like they may have found something they couldn't further explain, and wanted further investigation and input by others. In other words, they were already out of other ideas they were willing to put forth.

[JRehling]

The intent of the paper, as I understood it, was that after compensating for likely errors in the data, they felt like they may have found something they couldn't further explain, and wanted further investigation and input by others.

The title of the paper states: "ALMA discovers the most distant object of the solar system."

That's quite different from saying they had an observation they couldn't explain further.

[TheAnt]

I have my personal perspective on this since my work is not like physics or astronomy, where others can attempt to duplicate my result in the lab - or repeat the observation with their own instrument for verification. So I need hundreds if not several thousands of datapoints/observations to make a claim about anything, so that peer reviewers will not tear my paper to very small pieces - then jump up and down on the shreds with contempt.

Now that observations with ALMA submillimeter range falls more or less in the same exclusive category, since its the only radio observatory of its kind.
They need a little few more repeated observations over quite some time to make any claim to a discovery of anything.
My guess is that they did indeed observe something, at least for the observation of higher strength of signal, but that it was a transient phenomena quite further away in their line of sight and not anything comparatively local.

[fredk]

JRehling has already made most of the points I had upon hearing this story. But I would add a couple of comments.

First, it is astonishing that we didn't see a simple sanity check. Planetary astronomers must have crude, order-of-magnitude estimates of the number density vs size and distance for outer solar system bodies. Of course these will be largely model-driven and likely uncertain by orders of magnitude. But if those number densities implied the odds of finding such an object given ALMA's sensitivity and sky area mapped was between, eg, 1:100 000 and 1:10 000 000, then we could move on to other possibilities such as instrumental artifacts. There are quotes floating around from Mike Brown suggesting this is indeed the case, but I haven't seen any actual numbers.

Even more astonishing is that we only have two"blips" each (some with questionable S/N), ie one (perhaps) velocity each, as many have pointed out. At least for the alpha Cen case, it seems that this matter could have been trivially resolved by simply making another observation next observing season. The proper motion (if real) is such that the object should still be near alpha Cen in 2016. Is there some reason that this can't be done?