Catch up on stories from the past week (and beyond) at the Slashdot story archive

 



Forgot your password?
typodupeerror
Check out the new SourceForge HTML5 internet speed test! No Flash necessary and runs on all devices. ×
Space

More Than Half of Kepler's Giant Exoplanets Were False Positives 88

StartsWithABang writes: By surveying an area of the sky containing over 150,000 stars visible to it, the Kepler satellite monitored each one over a multi-year period looking for periodic changes in brightness. Thousands of planetary candidates emerged via the transit method, where periodic dips of 3% or less were noted with regularity. However, a follow-up study has come out on the giant exoplanets, finding that over 50% of them aren't giant planets after all, but wound up being eclipsing binary stars. Perhaps our lone star Solar System is the oddity, after all.
This discussion has been archived. No new comments can be posted.

More Than Half of Kepler's Giant Exoplanets Were False Positives

Comments Filter:
  • by macraig ( 621737 ) <mark...a...craig@@@gmail...com> on Thursday December 03, 2015 @02:01PM (#51051069)

    We all live in a lone star state, even those of us who don't live in Texas.

  • When you search for positive responses you get a lot of false positives. You should search what systems have no planets before searching the ones that could have one.
    • Re:Confirmation bias (Score:5, Informative)

      by gstoddart ( 321705 ) on Thursday December 03, 2015 @02:20PM (#51051229) Homepage

      They can search for things because the candidates show a difference which suggests you should look closer.

      How the hell can you search for nothing to then use that to help you find something?

      The false positives can be pared down with closer looks. There is no way in hell you can look at all of the stars, determine they don't have planets, and then use that to find the stars which might have planets.

      The ONLY way forward on that is by finding anything which might be a positive, false or otherwise, and then exclude things which were false.

      But you sure as hell can't look at every star, rule them out as having planets, and then use that list to find the ones which do have planets -- that's completely backwards.

      I'm pretty sure if there was a better way to be looking, they'd be doing it.

    • by Z00L00K ( 682162 ) on Thursday December 03, 2015 @02:21PM (#51051251) Homepage

      I wouldn't really call it false positives since there's a sliding scale between giant planet and star.

      It would have been a lot worse if it was revealed that there was no object at all.

      • I'd say it's pretty binary:
        fusion: star
        no fusion: planet

      • "I wouldn't really call it false positives since there's a sliding scale between giant planet and star." No there isn't. There's a very binary distinction that occurs when there's sufficient mass to cause nuclear fusion.
        • by DRJlaw ( 946416 )

          (1) No fusion versus (2) deuterium fusion versus (3) normal hydrogen fusion is not a binary distinction. At least in the eyes of the astronomical community.

          That's not even considering that we infer fusion from the observed temperature of the body, which is definitely a sliding scale, in combination with an assumed mass and age. Once you get down to brown dwarf scale, you cannot draw a binary conclusion.

          So, yes, there is a "sliding scale" in practical astronomy.

  • by tarpitcod ( 822436 ) on Thursday December 03, 2015 @02:08PM (#51051121)

    Even if 99% of the 2000+ exo-planets are not exo-planets, that's still 20 detected. Which isn't half-bad considering how long we've been seriously (space based telescopes etc) looking for them.

    The conclusions in the article are weird to me. They are saying 52% of the exoplanets may not be exoplanets for this Kepler system example. Even if that holds, given the 2000+ exoplanets, if 48% are still probably exoplanets, that's 960 of them.

    I'm assuming that Wikipedia's exoplanet count is sort-of right, and that it hasn't been already halved because we think 50%ish of them are probably other things.

    • by Anonymous Coward

      I agree. This is a general pattern I have seen in science writing and journalism that many writers do not understand the simple idea of percentage increase or decrease (which is probably why they did not decide to become mathematicians). This also applies to many approaches to trying to solve global warming or reduce energy consumption. It comes down to two misunderstandings: absolute vs relative scaling factors and something equivalent to Amdahl's Law. For scaling factors, if you have a large number a

    • It is a big number, but there's always the possibility that there are other glitches which could account for the remainder. Sunspots (on the distant star, not our Sun), flares on the distant star, variable stars, intervening objects in the hundreds or thousands of light years between us and the other stars, instrument glitches, I don't know what else, and of course I don't know which of these potential glitches have already been accounted for.
  • a bit exaggerated (Score:5, Insightful)

    by NostalgiaForInfinity ( 4001831 ) on Thursday December 03, 2015 @02:10PM (#51051133)

    They are "false positives" in the sense that the stellar companion may be a small star or a brown dwarf instead of a "planet". But the distinction between a "big hot gas giant" and a "brown dwarf" is fairly academic, in particular if you're concerned with things like habitability.

    • Re:a bit exaggerated (Score:4, Interesting)

      by AK Marc ( 707885 ) on Thursday December 03, 2015 @02:38PM (#51051415)
      If Jupiter were a brown dwarf, would that have prevented life from forming on Earth? If not, then the change in results is irrelevant to the search for Earth-like planets. In fact, it could help, as I've seen it postulated that Jupiter increased the chances of life on Earth by clearing the system of threats.
      • We simply don't know how important Jupiter was for life on Earth. People argue that it shielded Earth from collisions, but for that to be effective, Jupiter had to be outside of Earth's orbit. Most of these planets/brown dwarfs would be inside the orbit of any planet in the habitable zone. Gravitationally, they are likely pretty much indistinguishable from just a central sun. If they have an effect, it may be in causing or shielding solar flares.

  • We still have too little data to guess whether our planetary system is special. Transits and Doppler wobbles are being detected in only a small fraction of the stars we observe. One reason is time: it takes an a few orbits to establish a pattern. So it's only natural that most of the systems we've found have been compact. They're they low-hanging fruit. It will take a bit longer to get a good statistical understanding of the proportion of less compact systems.

  • by enjar ( 249223 ) on Thursday December 03, 2015 @02:16PM (#51051201) Homepage

    So someone comes up with a list of things that might be planets, then someone does further analysis and finds out that some aren't. Even (gasp) 52% of them! Science must therefore be useless.

    No, that's how science works -- you do an experiment, examine the results, then refine your experiment. Or someone else does. Repeat ad nauseum.

    • by gstoddart ( 321705 ) on Thursday December 03, 2015 @02:27PM (#51051311) Homepage

      More importantly, they found something using the technique. The technique worked.

      That they then looked closer and said "wow, not a planet, but another star" doesn't mean anything other than we're getting remarkably good at identifying candidates and then figuring out what they actually are.

      I'd say it's a great success, because they're actually finding things to look closer at. If some of those turn out to be not planets, but still actual things, then the technique is working just fine.

      • Science, where failure equals success. Failure is part of the process.

      • by enjar ( 249223 )

        Yes, indeed -- the fact that you can find the candidates, then do further analysis of the candidates and determine if they are planets or stars is quite amazing. They "only" surveyed 150K stars out of the 100 billion (as the lower bound -- could be 400 billion) stars in the galaxy, so this is indeed only scratching the surface of what can be surveyed.

        • They "only" surveyed 150K stars out of the 100 billion (as the lower bound -- could be 400 billion) stars in the galaxy

          This is the most important point. The summary's claiming that since half were false positives, we're likely unique. Except we've now gone from 2,000+ planets to around 1,000 planets out of 150K stars. If we extrapolate up to the 100 billion stars in our galaxy, that would be over 660 MILLION stars with planets in our galaxy. Some of those won't have life, but 1 in 660 million seems like

    • by Passman ( 6129 )

      So someone comes up with a list of things that might be planets, then someone does further analysis and finds out that some aren't. Even (gasp) 52% of them! Science must therefore be useless.

      Just for clarification from TFA, they did not disqualify over 1,000 candidates. What they found was: 67 Binary Stars and 3 Brown Dwarfs out of the 129 candidates they actually looked at.

      That seems like an awfully small sample size to me, but hey I'm not a scientist.

      • Just for clarification from TFA, they did not disqualify over 1,000 candidates. What they found was: 67 Binary Stars and 3 Brown Dwarfs out of the 129 candidates they actually looked at.

        That seems like an awfully small sample size to me, but hey I'm not a scientist.

        Actually, if that represents a random selection from their initial pool of candidates - that is, if they didn't do any initial pre-sorting to enrich their selection for stars over planets - then that's a reasonable sample size. As long as their sample was random, it's actually the absolute number of stars in their sample that matters. The standard deviation in their estimate of the number of non-planets goes as roughly the square root of the number of non-planets in the sample. We'll say the square root

  • So what? We only need one exoplanet capable of supporting life. And we will not even need that for decades, more likely centuries, possibly millenia.

    Our first co-location effort is far more likely to be a world (or moon) in our solar system where humans live in a contained environment while terraforming the rest of the world. The transit time and complexity in getting seeding life forms there is literally astronomically less than trying to do that on a world in another solar system.

    The only exceptions I

    • exoplanet capable of supporting life

      Define Capable. Mars, technically might be capable. But that doesn't mean it actually exists there. And until we actually go there, dig up something that fits our description of life, it is cannot ever be concluded that there is life, only probability.And then, we'll still likely be wrong, miss forms of life that do not conform to our definition.

  • When Kepler "found" the giant exoplanets, what was the certainty associated with the findings? Was it 100%?
    .

    I doubt if it was even close to 100%. So it should be expected that follow-up surveys would help to start sorting out the false positives. That's the way science works.

    In other words, this is news?

  • A better success rate that your fucking articles, you woad-smeared fixie-riding bald beardy bastard.

  • by mbone ( 558574 ) on Thursday December 03, 2015 @02:30PM (#51051335)

    Kepler has always required an independent check before moving a candidate exoplanet (with only Kepler data) to a confirmed exoplanet. That's why there are always a lot more Kepler candidates than confirmed exoplanets. From Wikipedia [wikipedia.org]:

    As of January 2015, Kepler and its follow-up observations had found 1,013 confirmed exoplanets in about 440 stellar systems, along with a further 3,199 unconfirmed planet candidates

  • so 50% is correct! (Score:4, Informative)

    by Ubi_NL ( 313657 ) <joris@ideeeGAUSSl.nl minus math_god> on Thursday December 03, 2015 @02:48PM (#51051485) Journal

    In my field of science, having a lead-finding system that returns a 50% hit-rate after confirmation is better then I've ever encountered. I'd say that this is an extremely impressive result.

    • by gstoddart ( 321705 ) on Thursday December 03, 2015 @02:54PM (#51051543) Homepage

      And, really think about it ... it's actually got a 100% detection rate of finding something, some fraction of which seems to be stars instead of planets.

      If I understand this correctly, none of these are "nope, there was nothing to see here", more that we're realizing that some of the candidates are stars instead of planets.

      That's not a miss in my books.

    • And when you think that we're detecting planets that are dozens to hundreds of light-years away... The mere fact that we can detect any at all is amazing and speaks to how far we've advanced. A 50% confirmation rate is phenomenal.

    • The problem is with the "science" journalism, of course. The sensational accounts reporting "billions of planets found!!!" set up the equally-sensational "half of planets false!"

    • I don't think the hit-rate after confirmation is known. This study ruled out only one of the potential confounding issues; there may be more. In the end, unless you can observe a planet's disk with a telescope, all you know for sure is that something causes periodic dimming on N stars, and you've ruled out a certain number of potential causes of dimming for those N stars.
  • Perhaps our lone star Solar System is the oddity, after all.

    What if it's impossible for life to form in a binary star system? That would change the probabilities of alien life quite a lot, wouldn't it?

    • by enjar ( 249223 )

      Somewhere in a binary star system, there's someone saying it's impossible for life to evolve in a star system with a single sun.

      "How can they live without continual light from a star? They would need to have a lifespan of one planetary revolution, and that's nonsense. No way intelligent life could come from an environment like that."

      • Unless all the planets these hypothetical observers can see happen to be tidally locked, and they assume that's normal, then the relevant period is one planetary rotation, not one revolution. In our solar system, rotation periods range from ten hours or so (Jupiter and Saturn) to one revolution (Mercury, about 60 days--tidally locked), to more than one revolution (Venus). Planetary revolutions range from about 60 days (Mercury) to 165 Earth years (Neptune, with Pluto even longer). Having a lifespan of
    • What if it's impossible for life to form in a binary star system?

      Dunno for sure about binary stars, but it's pretty clear that a ternary star system [amazon.com] will have serious problems.

  • Given the size of the cosmos (think: big. Really big), I don't think we're that unique.
  • by BenJeremy ( 181303 ) on Thursday December 03, 2015 @03:20PM (#51051769)

    It's terribly incorrect, to say, as the summary says, "Perhaps our lone star Solar System is the oddity, after all" - to the point of blunt stupidity.

    Kepler's transit method will find some exoplanets, or at least the signs that something is worth taking a closer look at, but it also relies on a system's elliptical plane being aligned just so such a large exoplanet can cross the path of the star - in other words, we have to be able to see that plan edge-on. This reduces the likelihood of practically using this method to actually find something around 100-to-1, if it even exists.

    • I'm glad someone else had a problem with the end of the summary.

      Was that last statement intended to irritate all of the logical thinkers on this site, i.e. nearly everyone?

  • Does this mean there are lots more binary stars than we realized?

In every hierarchy the cream rises until it sours. -- Dr. Laurence J. Peter

Working...