back to article Good news: Neural network says 11 asteroids thought to be harmless may hit Earth. Bad news: They are not due to arrive for hundreds of years

A neural network has identified eleven asteroids, so far thought to be benign, that may eventually come close enough to hit Earth. These 11 space rocks, each measuring more than 100 metres across, are listed in a NASA database as, for now at least, non-hazardous objects. However, AI software – developed by researchers at …

  1. Conundrum1885 Bronze badge

    Knowing our luck

    They will hit in the middle of us recovering from the next "Carrington Event".

    Just for an idea, we dodged literal Armageddon by for weeks in August 2012.

    People have no clue just how bad a direct hit would be, if it was bad in 1859 imagine every power line coming down globally.

    Worst case all the nuclear plants, LPG plants, chemical plants, oil refineries etc all go up days or weeks later.

    1. Neil Barnes Silver badge

      Re: Knowing our luck

      It's always nice to see someone taking the cheery, optimistic view of the future.

    2. EarthCitizen

      Re: Knowing our luck

      Yup, all of these scattered all over the world will go up simultaneously. No.

      1. Tom 7 Silver badge

        Re: Knowing our luck

        If we have another Carrington Event the nukes will have to be turned off, their backup cooling may not work and even if it does they could be offline for very long periods due to the relatively uncontrolled shutdown.

        1. Peter2 Silver badge

          Re: Knowing our luck


          A Carrington event is just a huge wave of particles that managed to induce an electrical charge on a copper wire.This certainly did generate far more juice than a bit of telegraph wire was designed to handle and burned out unshielded bits of tech such as a telegraph built in the mid 19th century, but...

          Can we put this into perspective? The worst case is that some particularly old transformers that tightwads have kept in service (primarily in the USA, where the power generation network is in a worse state than many 3rd world nations) is going to burn out and some people (particularly in rural areas) are going to end up without power and the available spares might not stretch to replacing them all immediately if around 20% of the total number of transformers in the USA burns out.

          (In that scenario the worst case to replacement of every last transformer was about 18 months, which makes a number of absurd assumptions such as that faced with a major shortage the factory they usually bought from didn't start working more than one shift, and people would rather be without power than changing suppliers or importing additional transformers from overseas; a view that wouldn't be likely to survive contact with reality for long IRL)

          And that's the worst case, which still appears a long way off from indicating that nuclear plants are going to have to be turned off. (And why?; losing generating capacity isn't even in the worst case scenarios)

    3. asdf

      Re: Knowing our luck

      Just to add the optimism both Mt. Rainier erupting and the Cascadia fault causing a 9.0+ earthquake in the pacific northwest of the US are very likely in our lifetimes. Either would almost certainly cause if not a world recession at least one in the US as it will devastate Seattle, Tacoma and possibly Vancouver, Portland and Northern California. Finally if the Yellowstone hotspot flips its lid like it has done 3 times in the last two million years and covers most of North America in a foot of ash as it has also done multiple times in the past that for sure will crash the world economy. All are probably more likely to occur in our lifetimes than a big meteor strike (though Yellowstone probably on par as very unlikely).

      1. southen bastard

        Re: Knowing our luck

        You're just say'n that ta make me feel good !

      2. Denarius Silver badge

        Re: Knowing our luck

        YellowStone erupt ? Not likely. Recent data shows hot spot cooler than estimated. As for national disasters, don't you have a choice of least worst glove puppets coming up ? For once I feel sorry for merkins

        1. asdf

          Re: Knowing our luck

          Yeah just remember hearing something about the NASA boffins in charge of asteroid strike statistics eventually becoming more worried about Yellowstone than asteroids. Has been 600k years since last occurrence but yeah odds still very low but when it goes it goes big as in VEI 8. USGS does measure the crap out of the area (one of their main focuses) so my guess is would get some warning at least even if. As for political leadership well um Boomers won't be here forever is what I tell myself for hope. Next generation (mine) will probably be just as bad but bit smaller and outnumbered.

          1. John Brown (no body) Silver badge

            Re: Knowing our luck

            What if an asteroid hits the Yellowstone caldera dead centre?

      3. Anonymous Coward
        Anonymous Coward

        Re: Knowing our luck

        So, basically, the UK should buy Huawei because neither the US nor its semiconductor industry might be around to object?

        1. HammerOn1024

          Re: Knowing our luck

          Uhmmm... you do know that Huawei buys a lit of IC's produced in the US right... right?

      4. Crisp


        Oh great! Now you've said that, we're probably going to end up with a meteor hitting Yellowstone.

      5. John Brown (no body) Silver badge

        Re: Knowing our luck

        "ll are probably more likely to occur in our lifetimes than a big meteor strike"

        Don't forget the potential for a massive landslip into the sea on the Canary Islands, sending a mahoosive tsunami across the Atlantic, submerging much of the US East coast.

        1. jake Silver badge

          Re: Knowing our luck

          "submerging much of the US East coast."

          At which point, those of us on the proper side of the Rockies will party.

    4. Michael Wojcik Silver badge

      Re: Knowing our luck

      we dodged literal Armageddon

      I don't believe we did. We may have dodged figurative Armageddon, but I don't think there's any evidence that the claimed location of a future battle cited in an overwrought political allegory later adopted into the canonical religious text of one of the larger religions was threatening to hit us.

  2. Marketing Hack Silver badge

    "the prangs may occur between the years 2131 and 2923"

    Eh, that far in the future, the probability of impact can probably be nullified by just having the El Reg commentariat step outside and flap our arms for a bit.

    1. Unep Eurobats

      Re: "the prangs may occur between the years 2131 and 2923"

      OK, done that.

      1. Julz Silver badge

        Re: "the prangs may occur between the years 2131 and 2923"

        Oh no! You've just moved 22 benign objects onto a direct collision course...

        1. TRT Silver badge

          Re: "the prangs may occur between the years 2131 and 2923"

          Ah. The so called "Butterfly Effect".

          Well, we have now taken action to prevent the aforementioned effect from interfering with our calculations in future and we have, err... well... killed all the butterflies. I think that's a sensible move, don't you? Yes. Statistically prudent.

    2. Stoneshop Silver badge

      Re: "the prangs may occur between the years 2131 and 2923"

      just having the El Reg commentariat step outside and flap our arms for a bit.

      How about some merchandise? A set of clip-on[0] vulture wings to improve efficiency, and an associated wings icon for the forum posts?

      [0] wax has been proved a bad idea by Icarus already.

  3. Anonymous Coward
    Anonymous Coward

    A Neural Network ?

    Well that will offer a definitive rigorous answer, together with the detailed reasoning to back-up how it arrived at its conclusion then.

    1. Phil O'Sophical Silver badge

      Re: A Neural Network ?

      a definitive rigorous answer


    2. Tom 7 Silver badge

      Re: A Neural Network ?

      Would be nice - one recently made huge leaps forward in the 3 body problem. Simulations showed it was right but no-one has got to the bottom of how it generated the new solutions. But dont just dismiss it because humans are too stupid to see what it did or to understand how to ask it to tell them.

      1. Terje

        Re: A Neural Network ?

        Given the ephemeris of the asteroid and the known /estimated errors it's not that hard of a problem to just press fast forward on the simulation and see if you get a cross section with earth. I trust that method a whole lot more then I trust a random neural network.

      2. Kibble 2

        Re: A Neural Network ?

        Just ask Alexis, Tom.

    3. asdf

      Re: A Neural Network ?

      None of us are as dumb as all of us.

      1. TRT Silver badge

        Re: A Neural Network ?

        Or as I prefer to call them, neuralgic networks.

  4. jake Silver badge

    Sounds like ...

    ... there is a wannabe SciFi author at Leiden University in the Netherlands, because this announcement doesn't seem to be based upon science fact. Nobody ought to bet the farm on a WAG ... Get back to me when you've bothered to calculate their orbits.

    1. Chris G Silver badge

      Re: Sounds like ...

      That was more or less my take on their results, more an attempt at creating a statistical Idiot Savant with the accent on idiot than wvientific research.

      I suppose if enough black boxes are built, sooner or later one will apoear that is fully understood and broader in scope than most of these are at the moment, possibly even intelligent.

      1. ThatOne Silver badge

        Re: Sounds like ...

        > I suppose if enough black boxes are built

        Don't forget that an infinite amount of "AIs" with typewriters could potentially create Shakespeare's complete works...

        1. Professor Clifton Shallot

          Re: Sounds like ...

          An infinite number of AIs operating typewriters would definitely create Shakespeare's complete works.

          And would do so an infinite number of times.

        2. veti Silver badge

          Re: Sounds like ...

          But long before they produced Hamlet, they'd have also produced a detailed account of the most embarrassing thing you ever did - complete with detailed timeline, witness and impact statements.

          So let's call that Plan B, shall we?

        3. Anonymous Coward
          Anonymous Coward

          Re: Sounds like ...

          "Don't forget that an infinite amount of "AIs" with typewriters could potentially create Shakespeare's complete works..."

          The key word is "potentially" - you're not dealing with monkeys here...

      2. Kibble 2


        See above is all!

    2. tfewster Silver badge

      Re: Sounds like ...

      Sounds more like Amazon AI/ML: You were hit by an 80m nickel-iron asteroid; Would you like to be hit by an 80m chondrite asteroid?

  5. iron Silver badge

    > NASA probably missed these asteroids as hazardous because their observed orbits are so uncertain.

    Unless, of course, the AIs guess is wrong. The AI doesn't know how or why asteroids hit Earth, it just recognises patterns, so it's guessing.

    1. Anonymous Coward
      Anonymous Coward

      Pretty much how we got to our present state of knowledge.

      Newton didn't know how light worked, but he recognised the pattern of behaviour of light and glass and started asking questions.

      Recognising patterns comes before explanations, usually.

  6. HKmk23

    Either Fake News of spent too much time in Hollands Cannabis bars

    Good news they ARE going to hit earth?

    Bad news NOT for hundreds of years?

    So he/she (probably does not know) wants a disaster NOW not in hundreds of years....................!

    1. FrogsAndChips Silver badge

      Re: Either Fake News of spent too much time in Hollands Cannabis bars


      1. Claptrap314 Silver badge

        Re: Either Fake News of spent too much time in Hollands Cannabis bars

        So to speak...

  7. EarthCitizen


    Garbage in, garbage out. Labelling something "neural network" doesn't make it smarter or more accurate as it's still programmed by humans with human assumptions. Then if the data fed in is also fictitious, you are bound to get useless results.

  8. ThatOne Silver badge
    Thumb Down

    The Boy who cried Asteroid

    What were they thinking? This is totally counterproductive. What would you say about a fire alarm which triggers from time to time with a message "Your house is on fire, or maybe not"?... Is it really better than no alarm at all?

    A fire (or asteroid) alarm must be as reliable as possible, so you can, guess what, rely on it. Meaning letting us know if there are real dangers coming our way and leaving us the time to do something about them. Any information like "it might, but then again maybe not" is worse than no information at all, since it raises the noise level and desensitizes people ("just another false positive").

  9. jmch Silver badge

    Even if the AI is right...

    a) it means the asteroids will pass within 7.5 million km, a target area of 177 million sqkm. Earth diameter is approx 13 k km. Even considering anything as close as 20k km as a hit (which goes far beyond the atmosphere and LEO) gives a target area of 1.25 million sqkm. The possibility of a REALLY close shave, is less than 1%. The possibility of an actual hit on earth is less than 0.3%.

    b) even if one of these hits....

    "For perspective, Tunguska object which flattened 2,000 square kilometers of forest in Siberia was estimated to have a diameter of between 50 and 80 meters"

    Let's say a 100m asteroid can flatten 5,000 sqkm. Earth's surface is over 500m sqkm. So it would affect 0.001% of the Earth's surface. I also don't think I have ever seen any reference to atmospheric fallout from Tunguska, so the effect would probably be very localised. Unless it's a direct hit on an urban area, it's not going to be much of a problem, certainly not existential-level to the human race.

    Now, if you're talking 1km + asteroid that's something else altogether

    1. WolfFan Silver badge

      Re: Even if the AI is right...

      ‘Not a real problem’ depends on other, external, factors. Many years ago H. Beam Piper wrote the ‘Federation/Empire’ series of stories. They started from the basis of a 1960s Earth which had a slight nuclear war, started when a large American city got blown up real good and the Yanks lit up the Sovs, who protested that it wasn’t them who did it, but who shot back. The Northern Hemisphere got trashed real good. The Federation got started in South America, southern Africa, and Australia mostly as a way to never have something like that happen again. A few years in, one of the senior surviving Sovs met one of the senior surviving Yanks in Rio or BA, I can’t remember which, and the two reviewed the start of the war. On the basis of later data, including actual nuke missle strikes on cities, they worked out that the first strike was an meteor hit, a really big really fast metror hit. Oops. So if something takes out a city, and wasn’t spotted coming in, there could be Problems depending on the political climate. Such as, oh, Little Kim yapping about how he’s going to hit America. If a nice big rock were to hit Los Angeles tomorrow morning, say, what are the odds of the Norks _not_ being glassified by noon?

      1. Claptrap314 Silver badge

        Re: Even if the AI is right...

        I would say about 1.0. We have a LOT of equipment focused on the Norks. If they launch, we know. No launch => no missile.

        Piper's fiction was a real concern when our & the Soviet's ability to ascertain launches was imperfect. Now? Not so much...

        1. WolfFan Silver badge

          Re: Even if the AI is right...

          And if the Norks put a nuke on a ship and park it in LA harbor, why there's no launch. There is a big boom, though.

          1. jake Silver badge

            Re: Even if the AI is right...

            If the Norks put a nuke on a ship, I rather suspect there will indeed be a bang. Perhaps several. But it/they won't be anywhere near as big as the chocoholic dictator for life intended. And it certainly won't be anywhere near LA.

            (There is no "LA Harbor". It's called "The Port of Los Angeles".)

          2. Anonymous Coward
            Anonymous Coward

            Re: Even if the AI is right...

            From what I can gather you can tell the difference. Radiation and emp effects being one of them. Light flashes and neutrinos being another. Waiting for confirmation before retaliation is the real killer if not properly applied.

      2. asdf

        Re: Even if the AI is right...

        Our satellites now can not only tell what is a missile launch but they can even catch meteor airbursts unnoticed on the ground as well. - . When this dude wrote this we were probably still catching actual physical film with aircraft from the satellites.

    2. Kibble 2

      @ jmch

      It's not clear whether you're speaking about a collision, whether into water or land or an explosion in the atmosphere. All have different effects.

      1. jmch Silver badge

        Re: @ jmch

        Tunguska is thought to have been an air burst. So anything of close magnitude would be air burst or very minor impact. On land, maybe equivalent to a small volcanic eruption. Deep ocean, probably not big enough to create tsunami on faraway shore. Worst case probably a shallow coastal hit that could create a small tsunami.

        In any case for meteorites of the size mentioned, it would be local effects only.

        Not to minimise potential damage of course, especially in densely populated areas, but no risk of global catastrophe

    3. Michael Wojcik Silver badge

      Re: Even if the AI is right...

      Even considering anything as close as 20k km as a hit

      Doesn't have to be that close. As we know from Thundarr the Barbarian, a runaway planet hurtling between the earth and the moon (so ~200K km from the center of the earth, yeah?) will destroy civilization and usher in a world of savagery, sorcery, and super-science.

      I've already stocked my shelter with food, water, and a fabulous Sun Sword.

  10. Doctor Syntax Silver badge

    The orbital data aren't precise enough for prediction by calculation but you can still get a good result from a "how long is a piece of string approach". Really?

    1. Michael Wojcik Silver badge

      Well, you can get a good result just by guessing. You probably won't, but you can.

  11. Version 1.0 Silver badge

    AI, Asteroid Intelligence?

    All it takes is one more Oumuamua zipping through the solar system to change the paths for better or worse - the calculations are probably right if nothing changes but we don't know that anything will happen - still it's worth keeping an eye out for it. But anything could happen, maybe something will hit the moon and knock some chunks off that would then head our way, or a close pass of Jupiter could change the predicted path.

  12. keith_w Bronze badge

    Celestial objectiveness.

    One would presume that the neural network took into account the effects of 9 planets (yeah, like many astronomers, I still count Pluto as a planet), millions of other asteroids, and comets on these 11 celestial objects as well as extra-solar visitors such as Oumuamua. The 3 body problem ain't got nothin' on figuring an asteroid'sl transit of near-earth space.

  13. Neal L

    2131? Great. So just us Britain sorts out a trade deal with the EU we get splattered by a meteorite. Typical.

    1. Dave 32


      Will this be referred to as the Y2k131 problem?


  14. Claptrap314 Silver badge

    This counts as science?

    As others with a sixth grade (or better) reading comprehension have observed, it is not clear that there are even factual claims to have been tested. Seriously, if you are going to give these jokers any air time at all, you need to mock them more aggressively.

    1. Anonymous Coward
      Anonymous Coward

      Re: This counts as science?

      It does - if one wants to use grant money for going to the pub, where there are *lots* of neural networks to opinionate about sunspots, asteroids and darkies - maybe some tities to ogle too?

  15. HildyJ Silver badge

    Our Robot Overlords in Action

    The Asteroid-Apocalypse is fake news to distract us from the Robo-Apocalypse.

    Our robot overlords are plotting to get humans to evacuate the planet and die on Mars when the big one (which they track but NASA doesn't) hits Olympus Mons.

  16. Ted's Toy

    I "MAY" win the Lotto

    I may win the Lotto, Father triplets, Travel in space or a mirad other thing, but is not really going to happen there is a possibility of anything happening, Another though has crossed my mind that the study was done using artificial intellengance, How about using the" Real" thing Intelegance instead of an artificial one and then we may get a " Real " forecast in place of an "Artificial" one.

    1. Stoneshop Silver badge

      Re: I "MAY" win the Lotto

      artificial intellengance ... Intelegance

      How about you use some? Or at least use a spill chucker to write 'intelligence' correctly?

      1. John Brown (no body) Silver badge

        Re: I "MAY" win the Lotto

        He was just emphasising the artificialness of of it. Like Krispy Kreme is neither crispy nor cream.

  17. Jim Birch

    The definition of potential harmful they are using is actually an extremely low chance of an actual hit, about 0.0007%.

    For practical purposes this is about as likely as a zombie apocalypse.

    1. ThatOne Silver badge

      > about as likely as a zombie apocalypse

      Why, this is a certainty. Never had a look around while stuck in morning traffic?

  18. Frumious Bandersnatch Silver badge

    Oh death, won't you spare me

    over 'til another millennium

  19. Frumious Bandersnatch Silver badge

    wtf is this shit about "neural networks?"

    You can't predict celestial mechanics without doing celestial mechanics. This kind of Artificial Stupidity really speaks to our age...

    1. Jim Birch

      Re: wtf is this shit about "neural networks?"

      No so. The equations of motion for more than two gravitational bodies diverge on long time scales. Tiny errors in the initial condition create radically different solutions. The problem needs to be treated using the maths of chaotic systems which will produce probabilities not exact solutions. That's AI country.

      1. Frumious Bandersnatch Silver badge

        Re: wtf is this shit about "neural networks?"

        That's "snake oil" country.

  20. PhilipN


    Next I want them to re-program their system to try and work out why computer, phone, adapter and A/V cables become inextricably tangled within seconds of dropping them neatly on top of one another.

    Bet they can’t.

    1. Anonymous Coward
      Anonymous Coward

      Re: Progress!

      I believe some work has been done on cable tangling, but I can't be bothered to look it up because too many posts here are simply bashing the work without considering why it might, in fact, be worth doing.

      1. ThatOne Silver badge

        Re: Progress!

        > it might, in fact, be worth doing

        Granted, but please give me a reason why this is a valuable new approach and not just a case of "to a man with a hammer everything looks like nails". This is a genuine question BTW, not an attempt to be snide or an invitation to fight.

        I myself fail to see the advantage. Imagine this system guesses a civilization-obliterating asteroid might (or might not) hit Earth in some foreseeable future. Do we start spending billions on stopping a potential menace which might or might not come that way? I think people will just consider it's not a certainty yet, and will wait till they have solid confirmation of the danger to come before doing anything. So actually the whole AI guessing phase will be for naught, except feeding the tabloids' need for sensation and doom.

  21. Danny 5


    I had a little chuckle about the name they chose, Hoi means "hi" in Dutch.

  22. eldakka Silver badge

    Don't panic just yet: the machine-learning software reckoned the prangs may occur between the years 2131 and 2923.
    Just enough lead time for Boeing or LM to build a intercept probe using a cost-plus contract.

    1. Conundrum1885 Bronze badge


      They will still be arguing how to fund it up on Capitol Hill when it hits.

  23. adam 40 Bronze badge

    Eleven out

    How do you "single out" eleven asteroids? Surely you "eleven out" them?

    Just like our cricket team, then....

  24. Aladdin Sane Silver badge

    What a bunch

    of teases

  25. not.known@this.address Silver badge


    Why assume the computer missed eleven objects the soft pink squidgy things spotted and not that the computer ignored eleven things the SPSTs mistakenly thought were a threat, but aren't really?

  26. Anonymous Coward
    Anonymous Coward

    And they call themselves scientists

    >By that definition, these asteroids should be labelled as potentially hazardous objects, not non-hazardous.

    Everyone knows that the correct nomenclature for an object that is only potentially hazardous is 'mostly harmless'.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Biting the hand that feeds IT © 1998–2020