Odds are good you won't be hit by a satellite this weekend

A retired climate research satellite will plummet to Earth on Friday. There is a 1-in-3,200 chance of it hitting a person. BUT! Don't worry too much about that, says Scientific American reporter John Matson. A 1-in-3200 chance of a piece of the satellite hitting somebody, is not the same as a 1-in-3200 chance of it hitting you, specifically. He calculates the risk of that as 1-in-22 trillion.


  1. I’m glad someone is finally pointing this out.  Every news article I’ve read about this satellite has been incorrectly citing the odds.

  2. NASA is really pushing that “1 in 22 trillion” number, as if that somehow means it’s somehow not a real danger. 1 in 3200 is a pretty high chance of hitting someone, if that’s the real number. If you’re on the receiving end of a hot piece of space debris, my guess is that it doesn’t really matter how many other people didn’t get hit.

    If NASA wants to play with statistics, let’s have some fun; there are probably a couple hundred research scientists and program staff who are part of the U of Colorado program that dreamed up this satellite, 18,800 employees of NASA and 126,000 employees of Lockheed Martin, all of whom failed to remember the basic physics of “what goes up (probably) must come down” and didn’t create a satellite that would completely burn up or have any way to push it beyond earth’s gravity well. Let’s call it an even 145,000 people who (in some sense) bear some responsibility for this Angry Bird of Death. That means that come friday, there’s a 45 to 1 chance that there will be someone responsible for the impact, or in other words, a 100% chance that 45 people are going to feel really shitty. That’s the way math works, right?

  3. “He calculates the the risk of that at 1 in 22 trillion” sounds way more impressive than what he did: “3200 * 7 billion = 22 trillion”.

  4. Best if it aims at Australia – they can’t even catch a rugby ball, let alone a satellite.

    (disclosure:  I’m allowed say that, personal history.  and I hope it doesn’t nail someone in Oz)

  5. well, 1 in 3200 of someone, but there is some chance (say 1 in 100) of it hitting a group (say, a bus), so the chance of being hit is not 3200x(world population below 60′ latitude), it is more likely.

  6. wasn’t this the pilot for “Dead Like Me”?

    seriously though, when I’m old enough for death to be more or less imminent anyway, this is _exactly_ how I want to go.

  7. When Skylab came down I ran an office pool for the biggest chunk to fall — sold off 10-degree slices of longitude on a polar projection map for $1 each. I didn’t win.

  8. Since people aren’t distributed uniformly across the surface of the planet, I don’t think it’s fair to simply  divide the chances of it hitting someone by the number of humans and say that’s the chance it’ll hit _you_.  You can only multiply probabilities like that when the events are completely independent.

  9. Allegedly the one in 3200 calculation assumed everyone was standing up (less surface area to hit) and also assumed the debris wouldn’t hit anything that could cause collateral damage (like a bus or gas tank).

  10. Here’s a citation for my previous comment (from the Safety Critical mailing list):


    Essentially the figure 1:3200 is derived from using a detailed model to
    work out how many bits are going to hit the earth, the size of the
    “debris casualty area” of each bit, and the population density in the
    strike zone. There appears to be an (unfounded) assumption that all of
    the individuals are standing up :-). Since they use data from events to
    validate and refine the model on an ongoing basis, it’s one of the more
    empirically-valid quantitative risk assessments that I’ve heard of.

    Pity about anyone lieing down though, or in the proximity of anything
    explosive or toxic that is hit, or …A good example of how a few undocumented/invalid assumptions can change
    a carefully quantified risk estimate into a wild guestimate.
    (Please note that this is not intended as a criticism of the analysis –
    I think it is indicative of the state of both art and practice)

  11. Here is a back-of-the-bus-transfer calculation (with lots of simplifying assumptions– but do we really know if the official calculation is more sophisticated?)

    Land area (without Antarctica): 135 million square km
    H. sapiens:      7 billion
    —> 19,300 square metres of land per person.

    NASA is predicting 26 substantial satellite chunks.  So if a person is a one square metre target, there is a 1 in 750 chance that someone, somewhere is hit– about four times more likely than the official prediction.

    Ergo:  the official NASA target size of a human being is about 1/4 square metre (or about 2.5 square feet).

      1. Ahh…  thanks Steve.  A fudge too far… two heads are so much better than one.   Add a factor of 1/4 for land area/total area.

        1 person / 19,000 sq m of land   *  1 sq m land / 4 sq m potential impact area  *  26 pieces

        = 1 in 2900 chance of someone being hit.  

        I think that’s a pretty neat Fermi approximation.  Who needs all that fancy stuff about ‘debris casualty areas’ and ‘population densities in the strike zone’?  ;)

  12. I saw a news report about this the other day. They stated that the risk of it hitting you is 1 in 3200, but then confounded their error by running this whole segment about how much more likely you are to be hit by this satellite than be struck by lightning, win the lottery, etc. It left left me screaming at the TV, waving my arms and shouting “you fools, don’t you realise there are seven billion people on the planet? If the odds of each of them being hit were 1 in 3200, that would mean NASA expects the satellite to hit two million people! It’s really, really obvious that the 1 in 3200 figure is the probability of it hitting *a person*, not any particular person. Couldn’t you just sit and think about it for one moment before going to the trouble of making all those animated infographics to illustrate something that’s obviously completely false to anyone with half a brain cell?” But the TV couldn’t hear me so I posted it here instead.

Comments are closed.