Hacker News Reader

Comment by nabla9 | Hacker News Reader

The background for the article is 'tilt erosion time' and the large number of tidally locked planets in the habitable zones of numerous low-mass stars. Many new discovered planets are tidally locked: Kepler-438b, Kepler-442b, K2-3d, K2-155d. It was previously assumed that tidally locked planets would not be habitable. New climate modeling seems to change this view.

No snowball on habitable tidally locked planets https://arxiv.org/abs/1705.08904

Stabilizing Cloud Feedback Dramatically Expands the Habitable Zone of Tidally Locked Planets https://arxiv.org/abs/1307.0515

Tidal obliquity evolution of potentially habitable planet s https://arxiv.org/pdf/1101.2156.pdf

View on HN

"New climate modeling seems to change this view."

And this modeling was confirmed to be accurate... how, exactly?

This strikes me as a classic case of goosing the model until it says what you want it to say, and then getting in the news. We have no reason to believe these models. We have no reason to believe prior models, either. We have no reason to have any confidence in our ability to model a tidally locked planet, of unknown and arbitrary chemistry, around an unknown star, for unknown periods of time, with unknown orbital characteristics across geological time, with zero data points. We particularly have no reason to believe in any estimates of how long such a system could last; it doesn't seem to do life much good if there's a place for it to exist for a couple of million years, followed by freezing. The simulation is almost certainly unstable here; very small changes in the stability of this configuration will have massive changes to the result over even a few million years.

There's also more to life than just a comfortable temperature. Life is possible on Earth in part because we have a moon and plate tectonics keeping this stirred up, so some of our vital nutrients don't just sequester themselves somewhere chemically convenient after a couple of million years and then hang out there for the rest of the planet's existence. Do these types of planets have a solution for that? We have no idea. (One advantage of the Earthlike planets is that we do in fact have an idea... we know it's at least possible once.)

>We have no reason to believe these models ....

What is your level of expertise or knowledge of planetary astrophysics and climate modeling?

Sufficient to know our data point count is zero, and the model can not possibly have been verified against anything.

This does not require extensive study.

Furthermore, as a bona fide, credentialed expert in computer science, I observe with the full power of my credentials that other fields frequently abuse modeling to get into the news, and that as I said, this shows all the hallmarks of being one of those. There is a profound, mathematical way in which models simply spit back out at you what you put in. This profound mathematical understanding seems to be broadly lacking, and it makes people grant wildly excessive credence to unverified models. As just a single for instance, I expect you have no idea how many models they ran that produced lifeless planets until they finally found one that yielded a result conducive to life. Given the almost-certainty that these systems are deeply, deeply unstable, I expect it is almost certainly the case that they had a large number of runs where the system simply ran away in one direction or another.

Edit: I challenge the downmodders to produce a single data point about an eyeball planet in a habitable zone and demonstrate a model that correctly represents it. "We have to have data before we can verify a model" is not some sort of wild anti-scientific statement; the belief that a model can be trusted without validation is the wildly anti-scientific position!

If some of you are mistaking this for a position on the climate debate, note that we do have data for Earth's climate. Not as much as we might like (could always use more!), but it's certainly much greater than zero. You can literally get more than zero data just by walking outside and observing the temperature right where you are. This point has nothing to do with the climate debate on Earth.

Perhaps relevant is that we have also been very bad at modelling the recent and short term future climate of our own damn planet!

Things are finally settling down to the point where models match independent data, e.g. the recent research into sea level changes as measured by satellite.

So now, finally, we can start making informed policy choices based on cost benefit analysis. (Actually, no, who am I kidding - Greens worldwide will keep hating on industry and capitalism, and everyone else will keep pretending AGW isn't real. But at least we could start forming rational policies if we wanted.)

Back on topic, to assume that we can make any kind of predictions about the climate of life bearing exoplanets seems like the height of arrogance given the state of the art in climate modelling.

Mainstream models disagree by less than 2 K (±0.3%) over the next 100 years. That's pretty darn accurate! 2 K makes a difference to us, because we built so many cities within 100 m of sea level (1% of an ocean depth)

Predicting whether life might exist on a planet can probably tolerate ±10% temperature accuracy. There's no reason to believe that smart geophysicists can't achieve that level of accuracy.

Yes but a) they weren't originally that accurate, and b) have only just been validated against independent data (satellite measurements of sea levels).

My point, perhaps snarkily made, was that we are only now getting the hang of this, after decades of work.

Unless I'm missing something, we're not going to be approaching anything like 10% error bars for planets for which we have far less data.

I mean, you’re not wrong, despite the hivemind’s disapproval. Math and CS are our fields, and that’s what these other fields are using and abusing, and many of us have significant cross-disciplinary expertise and experience that gives us good intuition about bullshit. The lack of skepticism is why the replication crisis is such an issue... your first reaction to any publication should be doubt.

A bunch of programmers talking crap on astronomers because "Math and CS are our fields"... Oh geez.

Somewhere in an alternative universe, there must be an astronomers' forum talking crap on developers because Heartbleed and Spectre show they all misunderstood Goedel's theorem, and believe me, I totally know what I'm talking about, I use computer every day.

Fucking hell, astronomers do much more complex mathematics than 95% of computer scientists. The fucking hubris in this thread is astounding. Us "computer scientists" can't even design secure, working systems in our own fields, what authority do we have to lecture other fields?

"don't roll your own crypto" is a frequent reminder in programming threads that most people don't understand cryptographic mathematics and we shouldn't play with toys we aren't old enough to use yet.

But apparently the same is not valid for complex fields like astrophysics, spacephysics or astronomy.

>> Furthermore, as a bona fide, credentialed expert in computer science

That's alright. I'm a credentialed expert in computer science myself and I'm also always right about everything, including entire scientific fields I know next to nothing about.

If you understand how computers work, you understand how everything that uses computers works!

My favorite example of this phenomenon remains Soylent. Their first formula neglected to include trivial, unnecessary "nutrients" like iron.

Somebody - not me - told his mother about Soylent.

She said, "That's like SlimFast."

Since it was marketed for tech guys, it was a tech discovery.

>This profound mathematical understanding seems to be broadly lacking, and it makes people grant wildly excessive credence to unverified models.

3d General circulation models and basic energy balance models are verified against range of temperatures, pressures, and atmospheric conditions in the Earth, Mars, Jupiter and Venus. The same model used for exoplanets is used to model paleo-Mars, paleo-Venus, and Titan. If the model is somehow completely wrong outside known limits, so is the parent model that is used in the Earth Climate change studies. The models work just fine at this level of required accuracy. You put parameters that describe the Earth, Mars or Venus into the and you get good description of atmosphere and climate in Venus and Mars that matches observations.

We don't have to model specific planets to get interesting information. In generally the interest is to model different categories of the planets and discover how their environments vary when we vary the parameters. It's possible to say something generic about tidally locked planets and their climate. This is what these simulations do.

GISS modelE GCM https://www.giss.nasa.gov/tools/modelE/

ROCKE-3D version of GISS modelE GCM https://www.giss.nasa.gov/projects/astrobio/

Thank you for providing these links. Some people, especially on HN, seem to be surprised how far GCM modeling for exoplanets has gone.

Besides the examples you mention, research groups are now doing data assimilation for Mars (https://www2.physics.ox.ac.uk/research/geophysical-fluid-dyn...; https://agupubs.onlinelibrary.wiley.com/doi/full/10.1002/201...).

Other groups are testing and improving models for Jupiter and Saturn, as you mention, based on various remote sensing observations. (For Saturn, see: https://theoryofclimate.sciencesconf.org/conference/theoryof...).

All of these cases have allowed modelers to observationally test their models and model physics against non-Earth conditions. This work has been going on for a decade now, but it somehow has not reached the technically-aware audience.

The case of tidally-locked planets is another step beyond the above solar-system planets. Thank you for the references you supplied, in your original comment, on GCMs for this case.

I think Jerf's point is that for systems that complex, it is more sane to expect that we will be off by a lot. Remember that weather casting is frequently wrong in your own city.

I enjoy modeling, and I think it's useful, interesting as well as respectable work. But a little humility would in science is always a good things.

I don't think this is a good take. Firstly, we're discussing climate, not weather, which is easier to predict than chaotic swirling fluids. Secondly, if we were basing a practical decision off this information, then we should apply some kind of higher threshold, but this research is essentially a discussion piece and it seems pretty plausible that this might be the case on some worlds.

Regardless, it's more useful to point to particular areas of uncertainty than to point to a general miasma of uncertainty around climate modeling as though nothing can be learned. It doesn't really move any conversations forward.

Weather forecasting (short term details) may be wrong, but climate modeling (long term trends) is generally correct.

Just because we can't predict when an individual hurricane will occur doesn't mean that we can't predict "hurricane season".

Climate modelling on Earth generally refers to the scale of tens or hundreds of years, not to billions of years. Our climate model for the next few billion years is essentially "Eventually the sun will turn into a red giant". And we live here.

edit: Apparently it's not even certain whether the Sun will engulf the Earth or not!

> I challenge the downmodders to produce a single data point about an eyeball planet in a habitable zone and demonstrate a model that correctly represents it.

At the risk of misunderstanding “eyeball planet”, you’re living on the most obvious example.

Elsewhere in Sol, Mars is in the “optimistic” habitable zone (I think that means “it would be if only it had been massive enough to retain an atmosphere”).

Earth climate models you already believe in so I won’t bother to name.

Can anyone else tell me the accuracy of The Ames Mars General Circulation Mode?

I hadn't heard it either, but the term does have a specific meaning: https://en.wikipedia.org/wiki/Eyeball_planet

> At the risk of misunderstanding “eyeball planet”, you’re living on the most obvious example.

You should read the articles on this website please.

To be fair even the title says "may" in it. The context is that eyeball planets we're not actively considered, and climate considerations are a weaker rejection criterion for habitability. There is also a secondary result which is that systemwide habitability zones may be extended in the case of tidally locked planet by generating a geographically fixed planet-scale habitability zone. These are interesting results, even if they don't take into account the full range of factors contributing to habitability, it's just a message saying don't count them out on this one factor that naively might be considered exclusionary due to unfamiliarity.

Btw tidal locking can cause tectonic activity, as in the case of the Jovian moons.

Ah yes, the “you’re not qualified to judge made up models” attack. Got heem.

Every challenge he puts forth requires at best high school level critical thinking skills.

No every challenge he puts forth demonstrates he has no idea about how modeling is used. He also clearly has no idea nor interest in the current state of astrophysics and astrochemistry if he's declaring that we definitely know nothing about planets we observe.

By his own "logic", we can't even be sure they're tidally locked (because how did we determine that? Modeling!) Which I mean, is technically true because hey welcome to science where confidence intervals are a thing.

> current state of astrophysics and astrochemistry

Surely the specialist knowledge comes after the general knowledge? If a Nobel prize winning Astrophysicist says "there are exoplanets where you can make a perpetual motion machine" you would not let him get away with that?

The guy further up was just saying that there are general things about modelling (a very broad topic that reaches across all of science) that don't add up.

Ultimately, reasoning about observations are authoritative, not credentials.

> If a Nobel prize winning Astrophysicist says "there are exoplanets where you can make a perpetual motion machine" you would not let him get away with that?

You're demonstrating exactly why general knowledge isn't enough to analyze specialized concepts! Your rhetorical question may seem to the layman to be a witty retort but it just demonstrates ignorance of the topic at hand.

In the case of your perpetual motion machine example, anyone with even slightly specialized knowledge of (astro)physics would know that there are no fundamental phenomena like the conservation of energy that prevent life on an eyeball planet, unlike with a perpetual motion machine. This is the rhetorical equivalent of comparing a Mount Everest expedition to intergalactic teleportation.

> The guy further up was just saying that there are general things about modelling (a very broad topic that reaches across all of science) that don't add up.

No, he was just saying that there are general things about how modelling is used. He didn't say a single concrete thing about the article itself and proceeded to demonstrate vast ignorance wrt this specific subfield of scientific modelling.

No, you have totally and completely missed the point. The point is that being an authority in a particular area is irrelevant when what you're saying is badly supported.

He did say something concrete, which is that we have no observations to compare the model to.

> He did say something concrete, which is that we have no observations to compare the model to.

Which is wrong. That's my point. You wouldn't know that without some specialized knowledge in astrophysics and how the models are developed and verified.

You're completely missing the point: challenge the ideas, not the person.

This is why it's so easy to tell the planet anything.

The people with the highest credibility tells the rest how it is, and we don't question it. Now it's magic eyeball planets, next year it's spinning discs with Santa in the middle.

As long as credentials are impressive, it all checks out doesn't it...

I dont disagree with OP.

There is such a gigantic massive amount of information we dont know.

Any words spoken about how life is created should be met with the same skepticism we have toward religion. We require evidence and the scientific method, a simulation made my students who need to be correct for a PhD isnt evidence.

It's still useful though. We create models, and then when we find more data points on tidally locked planets, we can compare those points to these models. For example, we may be soon able to determine atmospheric composition of both sides of the planet thanks to different spectrograph data available from our perspective. These models can tell us something about the concentration of gasses in those atmospheres at different times. Then those models become testable models.

I don't understand this anger about having untestable models. How would science progress without having theoretical possibilities before testable possibilities?

Indeed, it's though all the false starts that one gathers subject knowledge. All the people who build "untestable models" for their PhDs have at least learned something about atmospheric modelling, what numerical approaches work and what approaches work less well.

Reply on news.ycombinator.com