Read on Twitter
Ok... the P9-is-not-real-because-of-observational-bias story has made another comeback, this time as a re-analysis. Interesting stuff, but inconclusive. While no longer staying up at night to look for the damn thing has undeniable appeal, we press on. Why? thread below
Let's start with the basics: orbital clustering among long-term stable, distant KBOs is one of the key lines of evidence for P9's existence. The orbits *look* clustered, but are they really? Or is this all due to selection bias? There are two ways to answer this question.
The first is observability analysis, where you look at the whole dataset together and estimate a false-alarm probability based upon the distribution of a large number of independent detections. We did this a couple of years ago, and the stats determine the clustering is real.
The other (better-if-possible) way is to simulate the biases of individual surveys. This has also been done for the OSSOS & DES surveys, where the biases are well-characterized. In such a case, however, the already-small sample size reduces to only KBOs found by these surveys.
The OSSOS/DES analyses are no small task, and the groups did careful modeling of their data. Unfortunately, they have been unable to sufficiently overcome their own biases to demonstrate the absence or presence of orbital alignment.
The new Napier et al. paper does a thorough job of combining the surveys and doing a joint re-analysis. The answer here is basically the same: you can fit the OSSOS/DES/ST subset of the data with a clustered or a uniform distribution.
Importantly, these conclusions are different from saying that there is no clustering.
In part, the null conclusion is a result of DES being strongly biased to discover objects IN the cluster (DES has added several perihelion-aligned KBOs). It’s rather counter-intuitive: by adding peri-aligned KBOs, the Napier et al. confidence in a uniform distribution increases.
All in all, imagining that this paper proves that the underlying orbital distribution is uniform -- however tempting -- is just not correct.
