Hacker News Reader

Comment by majia | Hacker News Reader

> The answer was no: in more 1000 trials with randomly assigned planet sizes put through a virtual Kepler’s detection scheme, a pattern of similarly-sized planets in the same systems never emerged. This computational experiment did not reproduce what we observe in the Kepler planetary systems. Thus, the regular sizes of planets is a real astrophysical pattern.

Any explanation to this? I'd assume small planets are just harder to detect. Large planets could be gas giants that react differently to our observation technique. Also they tend to form further away from the star, leading to a longer orbital period and fewer observations. Could these contribute to an observation bias and does the simulation include these factors?

View on HN

> does the simulation include these factors?

Yes, that's what the "put through a virtual Kepler’s detection scheme" part refers to. So she:

1. Generated a bunch of random planet sizes.

2. Simulated what the detector would have reported for each.

3. Compared those virtual detections against the real detection data.

If the two results had matched, it would imply real solar systems have random planet sizes and the bias they see is from the detector. Since the two results are different, it implies the actual solar systems are not randomly sized with the apparent uniformity coming from detector bias.

(There could, of course, be a third factor where planets are randomly sized but the measured bias comes from something else.)

Another possibility is intergalactic bar billiards[1]

[1] https://sites.google.com/site/h2g2theguide/Index/i/540914

2. Simulated what the detector would have reported for each.

I think this would be really hard. For example, you need to simulate how the detector responds to different planet compositions, different signal to noise ratios at different distances from the star. A sphere in vacuum model may be a gross simplification.

Well, we already deduce exoplanets' masses, orbits, diameters, and so on from the sensor data that basically amounts to a tiny change in the light curve of the star. The astronomers are probably pretty well aware of the limitations of their models and potential confounders.

How would the composition of a planet affect how much light it blocked of its star when passing in front? And signal to noise ratios are presumably quite easy to simulate. So not really sure what's so hard about it.

Composition determines density. A rocky planet would block less light than a gas giant planet of the same mass, because it would be denser and therefore smaller.

But does density have to do with anything Kepler measures?

Since it only measures how much light decreases, I assumed it's only measuring planet size, and that we're totally ignorant about planet mass and density?

I'm not sure the author is taking into account observation time and the orbit diameter of planets.

In other words, it could just be that planets in similar orbits have similar sizes, which would be roughly consistent with our own Solar System.

Jupiter alone takes almost 12 Earth years to get around the Sun. It seems to me that an observer on a remote planet on the same plane as our own Solar System, observing us, would likely only see the inner 4 planets (if anything at all) during a 4 year observation window similar to Kepler.

Yeah I don't like the finality of the statement the `answer was no`. Couldn't the simulation (virtual Kepler’s detection scheme) simply be incorrect?

Anything is possible. You can never prove anything. In science/real life true means very likely and false means very unlikely. So yes the simulation could be wrong, but I guess with everything we know and the error bars the answer was clearly no.

The answer is "no" because the author needs to come to a conclusion for the paper that they are writing. It doesn't matter if they are correct, it matters that they produce a paper, so that's why they came up with this conclusion.

I wish researchers weren't biased to produce papers, it doesn't help us with honestly understanding our universe.

Following up with "Thus, the regular sizes of planets is a real astrophysical pattern" seems particularly premature.

Actually, it appears that while gas giants form at a distance from their star due to requirements around accretion and temperature, they generally tend to end up closer to their parent star than around Sol, although this again could be selection bias.

The Nice model provides a potential explanation for our setup, which essentially requires Jupiter to have formed far out, moved inwards under drag, bombardment, and transfer of angular momentum to planetesimals, and then moved outwards to resonate with Saturn after it formed.

There’s also the Grand Tack model which sees Jupiter careening from 3.5 to 1.5 to 5.2 AU.

As an added bonus, Grand Tack also explains tiny mercury and smallish Mars.

There’s also the question of age - how old are these systems relative to ours? Are we seeing them earlier, or later in their evolution? We don’t have good enough models to yet know.

We need more data, and we also need to understand Uranus and Neptune, which are currently a bit of a question mark.

https://en.wikipedia.org/wiki/Nice_model

https://en.wikipedia.org/wiki/Grand_tack_hypothesis

Judge planets by their size do you?

Reply on news.ycombinator.com