NASA uses computer model to find exoplanets

Please consider donating to Behind the Black, by giving either a one-time contribution or a regular subscription, as outlined in the tip jar to the right or below. Your support will allow me to continue covering science and culture as I have for the past twenty years, independent and free from any outside influence.

Garbage in, garbage out: Using statistical computer modeling only, NASA today announced that they are certain that almost a third of Kepler’s candidate exoplanets are really exoplanets.

Analysis was performed on the Kepler space telescope’s July 2015 planet candidate catalog, which identified 4,302 potential planets. For 1,284 of the candidates, the probability of being a planet is greater than 99 percent – the minimum required to earn the status of “planet.” An additional 1,327 candidates are more likely than not to be actual planets, but they do not meet the 99 percent threshold and will require additional study. The remaining 707 are more likely to be some other astrophysical phenomena. This analysis also validated 984 candidates previously verified by other techniques.

This is actually a stupid announcement. They haven’t learned a damn thing from this statistical analysis, but are merely saying that because Kepler found a lot of candidates, a lot of those candidates must be real planets. Worse, NASA is also implying here that confirming some of these candidate exoplanets by hard observations is now really unnecessary, since they can do it statistically.

This smacks of the corruption that has ruined much of climate research, allowing a computer model to replace actual observations. Big mistake. But I also suspect this announcement occurred for the same reasons: NASA wishes to justify its work and its funding, and thus decided to make a big deal about this very minor statistical analysis in order to puff up the discoveries of Kepler, even though there is no reason to do so.

I expect a lot of mainstream news organizations to write big puff pieces extolling this announcement in the coming days, which will once again prove that almost no one in journalism today has the slightest ability to apply their own independent analysis to the press releases they receive.



  • Tom Billings

    It started with “social science”, moved to psychology and psychiatry, proceeded to envelop medicine and then biochemistry, and then into what *should* be a “hard science”, whose name change to “climate science” in the late 1970s-mid 1980s should have warned us. Now it is proceeding to overtake every scientific field that must get their political masters’ attention by public fluff announcements. It destroys the recognition of the necessity of replication in science, and enthrones “peer review” in both funding and publication.

    Until the monopsony of government in funding science is decisively broken, this willl proceed to degrade the ability to trust results. In this case, we will have to wait till space launch becomes cheap enough, and space manufacturing common enough, that private sources can fund space telescopes to replicate this work. It may even be true!

  • Max

    Has anyone calculated the odds that a planet would pass between us and their sun? That planet would be large enough to be detected and yet have such a small orbital time that it can be replicated enough to be verified?
    I would think this would be an extremely rare occurrence. I go with Ochman’s razor, a more simpler explanation. Large sunspots for example would change the brightness of a star.
    I’m not saying the data is incorrect rather agreeing with Tom. Orbital telescopes. Observations needs to be verified. Computer manipulation does not reassure me.

  • LocalFluff

    The alien life story is certainly over done in today’s astronomy. It might backfire if nothing is found. It is however not fair to compare astronomical models with climate science models. The climate is a complex mystery while astrophysics is so extremely efficient that it is somewhat of a philosophical problem of why it is so. Transiting exoplanets is a game of statistics. Some exoplanets are observed directly, but with transits one can only talk about averages and not about any particular planet. Continuously adjusting these statistical averages is not a problem for science. This particular uncertainty of science is deliberate and completely under control.

  • Nick P

    Could it be that NASA is saying that their computer models correlate so well with other means of detection, that those computer models can replace the other techniques and free the equipment for other fields of investigation?

  • pzatchok

    Have you not learned by now?

    Stop doubting the computer and its models.

  • Laurie

    We should all place out full confidence in Big Circuit.

  • D K Rögnvald Williams

    Next, they will statistically reclassify a percentage as “dwarf planets.” Isn’t this politically incorrect?

  • Wayne

    Related Topic–

    Dr. Pedro Domingos
    “…on Machine Learning and the Master Algorithm.”

Leave a Reply

Your email address will not be published. Required fields are marked *