We Might Be Alone in the Universe

[Res Ipsa]

Please don't make specific claims about a study without quoting from it. What leads you to conclude that?

A star's rotation rate can be used to estimate a star's age. How do you calculate the age of non-periodic stars? Does the second study (of non-periodic stars) take age into account?

So? Did the authors of the paper take into account the stars ages? If so, how did they do it?

[Res Ipsa]

Let me ask, you really have to answer this one.

Are the authors concluding that the Sun is quieter than most Sun-like stars (with similar rotation periods) because the Sun happens to be a non-periodic star? Do you realize there are two studies in that paper?

Please.

I did. Patience, young man.

[Physics Guy]

Perhaps it would be logical to expect one ancient, advanced civilisation, out of many, to have created a galactic empire. Suppose for the sake of argument, then, that this is in fact a reasonable assumption.

If all it takes to destroy an argument on an enormous subject about which we know nothing, is for one assumption that seems reasonable to us in our ignorance to be totally wrong as a matter of fact, then it isn't a convincing argument. There are so many things about intelligent alien life that we don't know, and which could all be decisively important, that it would take a weird fluke for all of our ignorant guesses to be right about them. Trying to build a strong conclusion of any kind, in our current state of knowledge, is trying to build a heavy structure out of uncooked spaghetti. None of the components is strong enough to bear the weight it will need to bear.

[Physics Guy]

Okay, I just read through this article by Reinhold et al. about the sun being quieter than otherwise similar stars. The paper was published in Science Reports in 2020 (which is not as impressive as being published in Science itself). The version found by Res Ipsa, from the ArXiv e-print archive, is the same paper; for the past thirty years or so almost all physics and astronomy papers in the world have been posted, publicly and legally, on ArXiv. I don't really know how this practice overcame objections about copyright and peer review, but somehow it just did and by now it's the unquestioned norm.

The paper never suggests that the sun is a rare kind of star. It offers two alternative interpretations of its data, both of which assume explicitly that the sun is not a rare kind of star. The two alternative explanations that the paper offers are these.

1) Stars generally have a kind of magnetic menopause in middle age. When their gradually slowing rotation rate slips below a critical value, in relation to their size and temperature and composition, something changes in the complex magneto-hydrodynamics of stellar convection, and the result is a steadier stellar magnetic field. The sun may be in this magnetic menopause, or have gone through it already, while more of the stars of which the rotation rate could be measured have not yet reached this point.

2) Perhaps all stars that are roughly like the sun have occasional quiet periods that last ten thousand years or more, when they happen. Our sun happens to be in such a lull now.

Both of these interpretations assume that the sun is a common kind of star for its size, composition, and age—none of which is very rare.

The 2020 Science paper attracted at least some discussion; someone published a Comment about it. The original paper's authors then replied to this Comment, here. As you can read, the Comment argues that the magnetic menopause idea is more favoured than the original authors represented it to be, while the Reply still upholds both interpretations 1) and 2) as consistent with the data. Neither the original paper, nor the Comment, nor the Reply ever try to make out that the sun is an especially unusual kind of star.

One point about which I wonder, which neither the paper nor the Reply seem to make clear, is that it can be easier to measure the rotation rate of magnetically more active stars. That's because what mainly makes magnetically active stars show more variation in brightness is that sunspots are concentrations of magnetic field that make cooler, darker patches on a star's surface. Measuring how the star's brightness changes, and assuming that this is due to big sunspot clusters coming in and out of our view as the star rotates, is one way to measure how fast a star rotates. There are other ways to measure this, too, though, so presumably the authors relied on these other methods and did not make it through peer review with a conclusion based partly on a trivial sample bias.

According to what these authors themselves say several times, there is no reason to think that our sun is an unusual kind of star. It may now be in a somewhat unusual episode, which happens occasionally to most stars, or it may actually be in a post-menopausal state which most stars will enter at some point, and which it has entered at a typical time for its size, composition, and initial rotation rate.

None of these papers about stellar magnetism say anything about intelligent life. Conceivably their results do have some bearing on the life question: perhaps intelligent life is more likely to evolve around magnetically quiet stars. In that case it might be no coincidence that we have developed during a long lull in our sun, or as our sun entered menopause. Since most other stars will have lulls at some point, or go through a magnetic menopause, there would be no implication that intelligent life must be rare.

And once again, even if it were true that our sun were somehow very unusual, there would still not be any valid logical implication about intelligent life requiring a star of that unusual kind, just because we have one. The "it would be a fluke for us to be here, if we could be anywhere" argument may sound good at first, but it does not actually hold any water, when you examine it closely. It's just a subtle logical trap like the Monty Hall problem.

Intelligent life may well indeed be extremely rare; or it may not. Trying to draw any strong conclusion from our limited information is bound to be a bogus effort to spin straw into gold.

[Res Ipsa]

Thanks, PG. I was curious about why Kepler was unable to detect rotational periods for such a high percentage of stars. It turns out that one important reason is that "quiet" stars don't exhibit the surface features as frequently as "noisy" stars. The same characteristic -- changes in brightness -- is used to determine both the period of rotation and the "noisiness" of the star. That's why it's important that the authors simulated how Kepler would have seen our sun. If the sun was so quiet that Kepler couldn't even determine a rotation period, then any star that is actually sun-like would be found in the 87% percent of the sample for which a rotation period also could not be found.

There are, of course, other reasons why Kepler (actually, the tools used to interpret the Kepler data) could not determine rotation periods, such as the orientation of the star. I was able to find a paper from 2021 that asked the question: given a star with the sun's variability, how likely is it that Kepler would be able to detect an orbital period for it? http://www2.mps.mpg.de/dokumente/publik ... i/j546.pdf

The answer was surprisingly low:

In this study we identified biases in the period determination of stars with solar-like variability. The detection rates among these stars are lower than for stars of other spectral types. In particular, only 2.9% of them would be detectable using the thresholds set in MMA14. This is mainly caused by the small variability amplitudes of the rotational tracers and their relatively short lifetimes compared to the rotation period.

The very low detection rate explains the large discrepancy between the number of measured rotation periods (MMA14), and those predicted by Galactic evolution models (van Saders et al. 2019). The predicted number of stars with detectable periods (78), and that for which rotation periods have actually been measured (73), is remarkably similar. Figure 4 shows that many more rotation periods of solar-like stars may be measured when lowering the thresholds in the automated period surveys. However, this will also add a number of false periods, depending on how the thresholds are set.

Our study revealed that the rotation periods of most solar- like stars will go undetected using standard frequency analysis tools. Thus, we emphasize the importance of alternative methods for period detection such as the GPS method (Amazo-Gómez et al. 2020; Shapiro et al. 2020) or new approaches based on Gaussian process regression (Foreman- Mackey et al. 2017; Angus et al. 2018; Kosiarek & Crossfield 2020).

The Amazo-Gomez paper discusses the same problem:

Available methods for retrieving rotation periods from pho- tometric time series, e.g., autocorrelation analysis or Lomb- Scargle periodograms, appeared to be very successful for de- termining periods of active stars with periodic patterns of vari- ability. The transiting planet-hunting missions such as COROT, Kepler, and TESS, [/quoteBordé et al. 2003; Borucki et al. 2010; Ricker et al. 2015) opened unprecedented possibilities for ac- quiring accurate high-precision photometric time-series of stars different from the Sun. The new data from these missions en- abled studies of stellar magnetic activity on a completely new level. In particular, it has become possible to measure rotation periods for tens of thousands of stars (Walkowicz & Basri 2013; Reinhold et al. 2013; García et al. 2014; McQuillan et al. 2014). At the same time, the pattern of brightness variations in slow ro- tators such as the Sun is often quasi-periodic and even irregular. The irregularities are mainly caused by the short (in comparison to stellar-rotation period) lifetimes of magnetic features, such as starspots/sunspots, and a large degree of randomness in the time and position of their emergence on the stellar surface. This ren- ders the determination of rotation periods for low activity stars very difficult. For example, van Saders et al. (2018) showed that rotation periods of about 80% of stars in the Kepler field of view with near-solar effective temperature remain undetected. Conse- quently, the stars with known rotation periods represent only the tip of the iceberg of Sun-like stars. This can lead to biases in conclusions drawn based on the available surveys of stellar ro- tation periods. The relatively low efficiency of standard methods in detecting periods of stars with non-regular light curves might affect studies aimed at comparisons of solar and stellar variabil- ity. For example, solar variability appears to be normal when compared to main-sequence Kepler stars with near-solar effec- tive temperatures (Basri et al. 2013; Reinhold et al. 2020). At the same time, when comparisons are limited to main-sequence stars with near-solar effective temperature and with known near- solar rotation periods, the solar variability appears to be anoma- lously low (Reinhold et al. 2020). One possible explanation of such a paradox is the inability of standard methods to reliably detect rotation periods of stars with light curves similar to that of the Sun (see also discussion in Witzke et al. 2020). Along the same line, Reinhold et al. (2019) proposed that biases in deter- mining rotation periods might contribute to the explanation of a dearth of intermediate rotation periods observed in Kepler stars (McQuillan et al. 2013; Reinhold et al. 2013; McQuillan et al. 2014; Davenport 2017).

https://arxiv.org/pdf/2002.03455.pdf

Kepler was amazing in terms of the number of exoplanets it was able to find. But it was just our bad luck that it was exceptionally bad at identifying the rotational periods of solar-like stars (quiet type G stars of the same age, temperature, etc.). To treat the periodic star subsample as representative is to commit the base rate fallacy. What DT is trying to claim about the Reinhold et al paper fails to take into account the extremely high rate of false negatives in the data.

ETA: corrected citation link

[Physics Guy]

So is it true, then, that the "periodic" stars were mainly or partly pre-selected for being magnetically active, because high magnetic activity was what let their rotation period be measured? I was going to have to track back through more references to find this out, and I didn't want to do that. I was also expecting that the authors would have discussed a big confounding factor like that explicitly, if it was really present. There is at least one other way of measuring rotation (Doppler broadening of spectral lines, because the two sides of the star are moving towards and away from us as the star rotates). For all I know, Doppler broadening is too small an effect to be useful except for big stars or high rotation rates, but I was figuring that it must have been used, because otherwise the authors would have mentioned the correlation issue between having a measurable period and being magnetically active.

Perhaps they didn't discuss that issue because they, their peer reviewers, and their intended audience all knew all about it. For the kind of broad audience that Science Reports tries to address, that kind of thing should have been mentioned, but sometimes things like this slip through.

It would be more understandable for it to slip through if the two populations of stars were never supposed to be an important feature of the article, because all the article was really trying to say was that there exists a bunch of stars that are a lot like the sun except for being much more magnetically variable—and not that this bunch of magnetically noisy stars represents a big majority among all stars that are otherwise like the sun. The point would then just be that if you look for magnetically noisy stars that are otherwise similar to the sun, then you can find a fair number. This would raise the question of why the sun doesn't currently seem to be like those ones, but it wouldn't imply that they are predominant and so for the sun to be different is weird. The title of the paper, "The sun is less active than other solar-like stars" would then mean only "The sun is less active than some other solar-like stars," not, "The sun is less active than most other solar-like stars".

One of the problems in interpreting technical literature is that technical papers often say things that don't seem, to lay people, to be worth saying. We don't normally expect people to bother saying such trivial things. To specialists, though, even small details can be quite important enough to write a paper about them—or read a paper about them. Combine that with a bit of unfamiliar technical jargon that doesn't really mean quite what it sounds like, and it becomes easy for a lay person to misinterpret a paper as saying something more dramatic than it really does.

A paper in a journal for chess fanatics, for instance, might be about this wonderful new tactic that is incredibly powerful. Someone who is not so into chess might come away with the impression that this new tactic revolutionises the entire game, when the truth is only that it gives you good odds of gaining a one-pawn advantage in circumstances that come up in 0.1% of all games. To chess freaks, that could really be big news; but if you start talking to one of them about how the game will never be the same now that this new tactic is known, they'll have no idea how you could possibly have gotten that notion. To them it's the fact that the story was only about 0.1% of all games that was too trivial to mention.

[Res Ipsa]

Yes. If our total population is the stars included in the sample, given a star whose rotational period can be measured, it is highly probable that it's variability will be higher than the sun's. Given that 87% of the sample was composed of stars whose rotation period could not be determined, it's clear that the subsample of period stars is not in any way representative of the entire sample.

It is puzzling that the authors did not discuss the detection problem. One explanation could be that Kepler was designed to find exoplanets; it was not designed for stellar research. The authors acknowledge that early in the paper, but note that the mission also provides helpful information on stars. Most of the earlier literature I read concerning Kepler was all about the exoplanet data -- not stellar data. It may be that this was one of the first papers to examine the Kepler data with a focus on the stars rather than the exoplanets. The authors simply may not have recognized the detection problem because there was no published literature on it. The 2021 paper was the first one that I found that discussed the detection issue as it relates to stars, as opposed to planets. (The detection problems with long period planets orbiting G and K type stars was discovered years earlier).

In general, you can't find what you aren't looking for, and the paper's authors weren't looking at detection issues. Given the 2021 paper, there is a third explanation of the results found in the 2020 paper -- the correlation between stars whose period can be detected and "active" stars. From my own lay perspective, I'd put my money on the third -- it requires no theorizing at all about the nature of solar-like stars.

I don't think that makes the 2020 paper a "bad" paper. As is true of many, many published papers, they may become superseded by later research. I think it's fair to assume that future missions to look for earth-like planets will take into consideration Kepler's limitations.

[Gadianton]

The doomsday argument seems the weirdest of all. Assuming that every process is always most likely to be about mid-way through its total duration would imply that the moment when every process should be expected to have the shortest time left in it is the moment when it has just begun, and conversely that the moment when any process should be expected to have the longest duration remaining is the moment just before the process actually ends. A rule which is supposed to be the most logical estimate should surely not be so badly wrong, so consistently.

Hunting down notes on Carter Catastrophe, I accidentally found the answer to this. It turns out that you're right and also wrong. According to Laplace's rule, if I notice this big yellow thing rising in the sky one day providing me light and warmth, it's 1+1 / 2 + 1 chance it will rise again -- next day is 2 + 1 / 2 + 2. On day one, just like you said, the estimate is 2/3 the sun will rise again, which sucks. And by the time it's about to run out, the odds are very low it will fail. But wrong consistently?

I might be missing something, but either way, I have to say you nailed the generalized form of the question.

I think I threw that sci-fi book out, but it turns out I'm not special and someone else read the book. Assuming their example is from the book, then the argument: you draw a white ball from a bag, you draw a second white ball, then you draw a red ball. If all you know is it's the only red ball in the bag, then you reason the bag doesn't have very many balls. Now: You being born is the red ball; total human population ever must not be galactic in size based on the number of white balls drawn before you.

Shifting gears to Kipping, he says if you pick a green ball you live, but had you not picked a green ball then you die. I am here, therefore I picked a green ball. He says our survival biases us to think there are other green balls, but the truth is nothing in statistics can tell us how many green balls there are.

He might be wrong. There is a bag of green and white balls, green balls are planets that generate life; we drew a green ball first try. Our first observation may be guaranteed to be green, but the same is often true in inductive reasoning, as often there is no prior.

If I believe a physical process generated intelligent life on earth just like it generated a sun that rises, then Laplace's rule applies just as well here. The first and only ball I pick from the bag is green, and so it turns out that the probability of other intelligent life is 2/3s. Translation: we might be alone, but it's more likely we aren't, based on our observation of ourselves as examples of intelligent life.

A rebuttal might be that Kipping is trying to say that we have to add in a prior to correct the survivor bias on that first observation. But faced with the problem of induction, not knowing what we don't know, it's no different than any other purist induction problems like the sunrise problem where we're supposed to pretend we don't know anything about the sun nor have ever seen the sun rise. Unless there is no process, we have to start somewhere in the process under a hypothetical, which means with the positive observation of the sun rising or of the white duck. We can't start with absent white duck because we don't know what a white duck is before encountering it. First observation is always positive, and always 2/3s, apparently.

Looks like I'm singing Kumbaya with Res Ipsa and DT is going to have to join us.

[Physics Guy]

If "we probably won't last too many times longer than we already have" follows from "we know nothing about how much longer we'll last, we only know how long we've had so far", then that can only be because "we probably won't last too many times longer than we already have" is being used as a synonym for "we know nothing about how much longer we'll last". I'd rather make that admission in the plainer language.

[Doctor CamNC4Me]

This morning I watched a clip of Robin Hanson on the Lex Fridman podcast talking about, well, what you’re talking about on this thread. The conversation bounced around between the probabilities with regard to our own existence (Hanson lands on the idea that we’re incredibly early with regard to the age of the universe) and grabby aliens. Here’s the 17 minute clip that, in my opinion, is worth watching:

https://youtu.be/4WAoajXZN94

https://en.m.wikipedia.org/wiki/Robin_Hanson

Imma probably do a deep dive on the grabby alien thing because the basic idea is if aliens exist and were expanding slowly across the cosmos we’d already see them, but if they’re expanding rapidly we won’t detect them until one day we see giant spheres in the sky (I believe Hanson leans toward the latter notion).

So. Cutting to the chase, I recommend DT check out The Grabby Aliens theory. Perhaps this’ll be a good start:

https://youtu.be/llS4thbZkd0

- Doc