Neal Stephenson’s Seveneves – A Low-Spoiler “Science” Review

Title image courtesy NASA.

I’ve attempted to make this review as spoiler-light and science-oriented as possible. In some sense, discussing the degree to which spoilers will effect one’s enjoyment of a story can itself be a meta-spoiler about the overall structure of a story, so I will simply say that I read the novel knowing very little about it except Neal Stephenson’s own essentially spoiler-free blurb, and I enjoyed the story immensely. If you want to enjoy the story on the same footing as I did (modulo this intentionally obtuse digression on the nature of spoilers), then stop reading this review and go read Seveneves (the first 26 pages are online here for free). If you want to be spoiled by events in the first 80 or so pages (out of 800 or so), continue reading.

“The moon blew up without warning and for no apparent reason.”

Pg. 1, Seveneves

Stephenson’s best hook since Snow Crash masterfully sets the tone for the remainder of the novel. There are many different ways in which a reader can imagine the moon blowing up—imagery of the death star annihilating Alderaan comes to mind as the most culturally ubiquitous. But Stephenson quickly establishes the hard-SciFi slant of the novel. There are no fireworks, no loud booms—the opening chapters read like historical fiction for what would actually happen if the moon was fractured into (mostly) gravitationally bound chunks by some unknown force. Chiefly, people would freak out a little bit.

An ensemble cast of characters and their reactions span the opening chapters. A character obviously modeled after Neil deGrasse Tyson, Jerome Xavier Harris (or Doob for short), is rapidly recruited to run popular science damage control—essentially, doing exactly what I would imagine Neil deGrasse Tyson doing if such an event were to happen: media appearances and public lectures explaining that the still-gravitationally-bound moon posed no danger.

In some sense, this is naively true. A common physics brain teaser is to ask what would happen if our sun were replaced by a black hole of equal mass. Barring the joke answer of “everything would probably freeze to death”, nothing much would happen. The black-sun would still be in the same place, with the same mass, thus the orbit of the earth would remain unchanged. As it’s established in the early pages of Seveneves, whatever unknown force which fractured the moon didn’t send it on a direct crash-course with the earth—it just imparted enough energy to fracture it into a handful of pieces—pieces which remain in a wildly chaotic orbit around each other, but still somewhat confined to the general vicinity of the moon.

Orbital mechanics problems are notoriously tricky to solve. The so-called three-body-problem was one of the earliest problems tackled once the tools of calculus were developed by Isaac Newton in the late 1600s (a prominent character in Stephenson’s own Baroque Cycle). The problem is almost entirely explained by its title: how will three (or more) celestial bodies evolve in the night sky if you know their starting positions and velocities? Mathematicians eventually proved that there is no general analytic solution to the three-body-problem, barring a handful of special cases. The difficulty arises in the sensitivity of the problem to its initial conditions. If you don’t know exactly where the objects are and exactly how fast they’re going, your estimate can be wildly off.

Because numerical integration is really the only way to actually solve orbital mechanics problems, it’s clearly important that the chosen integration scheme be error free. Tools like Runge-Kutta and Euler integration are good, but they have technical shortfalls—they don’t manifestly conserve quantities that we know should be conserved. This sort of error is more than enough to yield completely nonsense results when trying to predict where things might be going in the night sky.

So, in one of my favorite scenes, when Doob realizes that the wildly chaotic orbits of the moonchunks might not be so harmless, and could even result in an exponentially increasing number of colliding and breaking fragments of moonstuff, I imagine that he sprinted to the nearest python terminal and coded up a symplectic integrator—an algorithm which is particularly good at orbital mechanics simulations. Whatever simulations canon-Doob did run, they soon indicated that the earth had about 2 years before the remaining moonchunks would rapidly break apart, and bombard the earth in moon-asteroids for approximately 10,000 years. This kills the planet. And everything on it.

I thought this would be nice to visualize, so I coded up a symplectic integrator and tried to see if I could reproduce what Doob calls the “White Sky” (the rapid dissolution of the moon) and the “Hard Rain” (the bombardment of the earth with asteroids). I tried to conserve mass and volume of moonchunks when they break apart and use a nearly elastic scattering model. With extremely generous assumptions, and the setting of arbitrary constants to 1 where convenient, I get something like this:

This simulation was very nearly contrived to produce an explosion of chunks in the beginning, and should be taken with a tremendous grain of salt, but hopefully it will allow you to at least suspend disbelief for the doomsday premise of the novel. The actual number of chunks and the rate at which they appear is, unsurprisingly, fairly sensitive to the initial conditions of the simulation.  If nothing else, this simulation illustrates Kessler Syndrome fairly clearly—the rapid creation of dangerous debris from uncontrolled, chaotic orbits in space.

This doomsday premise neatly synthesizes two of Stephenson’s long running interests—space travel and long term sustainability. He’d previously touched on both of these in his last SciFi novel, Anathem, but here, he’s far more technically grounded. This is the Stephenson thinking long deep thoughts about rockets—this is the Stephenson pondering how to stave off the harshest elements for thousands of years. These are ideas based in reality, and as the novel details the reconfiguration of the world economy into heavy-lift-rocket-factory, you’re stirred with a bittersweet hope—look at what we could do when faced with an existential threat—look at what we could achieve.

Because, ultimately, humanity’s only real hope of survival in the novel (and, arguably, in reality, were such a catastrophe to actually occur) is retrofitting a mildly beefed up fictional version of the International Space Station (henceforth, Izzy) into a civilization-ark. Thus, Izzy becomes the pivot of the second conflict of the story (after, you know, not dying in space): humankind versus bureaucracy.

In the early days of Google, when Google’s total employee count barely exceeded the double digits and when Larry Page was first CEO, he famously fired all of their managers. This didn’t last very long, essentially everyone thought it was a bad idea (including the engineers being managed), and the managers eventually got their jobs back. However, by Larry Page’s explanation, and on paper, it had a delicious sort of Dilbert-logic to it: he didn’t think engineers should have to be supervised by people without technical knowledge. This is a fairly common idea, if not outright complaint, that bureaucratic waste is the root of all of life’s ills.

Everyone that ever makes it to the International Space Station is outrageously competent and technically proficient. This is especially true of Stephenson’s astronaut characters. They are scientists and engineers, roboticists and technicians, but they are not bureaucrats—not politicians. And the balancing act required to both run a spacestation—usually occupied by highly competent technically minded people—with a large enough population to ensure a minimum viable population–that is, the minimum number of people necessary so that genetic defects from inbreeding don’t kill everyone—in a way that satisfies all of the mostly doomed earth—makes for a compelling sociological conflict.

That actually bears repeating as I don’t really consider it much of a spoiler: there is no technology available to humanity that could prevent the shattered chunks of moon from destroying everything on the surface of the earth. This is not the sort of novel where aliens arrive in the middle act and blast moonchunks out of the sky with Sufficiently Advanced Technology, nor where anti-gravity moonteleporting fields are invented by a lone genius at the eleventh hour. The vast majority of humanity is (likely) completely doomed.

That said, there’s a lazy way of approaching this sort of bureaucratic conflict, and there’s an interesting way. Suffice to say, Stephenson does not Fire All the Managers, and his characterization of the Bureaucrat in Chief, President of the United States, Julia Bliss Flaherty (JBF), is fantastic.

As for the main conflict, i.e. not dying in space, the book is technically exhaustive: fuel constraints—the Tsiolkovsky rocket equationthe necessity of artificial gravity for bone growth—the difficulty of growing plants—the structural integrity of an ever-growing, sprawling space complex—the balance of that same integrity with a need for maneuverability in the threat of bolide impact—the omnipresent threat of cosmic radiation—the physics of nuclear reactors—Stephenson peppers the story with enough detail that you believe whatever fraction of humanity makes it to space has a chance, albeit slim.

At its core, Seveneves is a Stephenson Novel’s Stephenson Novel: sometimes page-long asides into the intricacies of space technology and scientific vernacular interleave character conversations; there are lethal Russians and cryptographic battles of wits; it harkens to nearly every novel he’s written, and honestly reminded me the most of Cryptonomicon and The Big U, the latter of which he can’t seem to decide whether or not to include on his book back-covers. Some of the eponymous Seven Eves are more thoroughly characterized than others, but they’re all individually memorable. It’s a story about complete, incomprehensible tragedy (literally incomprehensible: imagine a puppy getting kicked. Now imagine a hundred puppies being kicked. Now imagine 7 billion puppies being kicked—our brains aren’t really sensitive to large numbers!). It’s a story about humanity.


Rating: S(6,6) out of Graham’s Number

Elevator pitch: Will make you want to be an astronaut.

Favorite moment: Somehow manages to top “Hiro Protagonist” with a new Best Character Name.

Seveneves is for sale on May 19th, 2015

Leave a Reply


  1. f knauss

    Can you share the code for your model? So, you know, we can play with it and see how fragile the initial conditions are?

    • Daniel Freeman

      I’ve been meaning to polish up my code because it’s definitely not in a releasable state. There’s also still some energy leakage due to how I’m treating collisions. I might have some free time in a month or so…

  2. Pingback: Neal Stephenson’s SEVENEVES: Political Theory from the Future | The Daly Planet

  3. Mike

    In my humble opinion, Lottery Discounts is the best Neal Stephenson character name.

  4. Fred

    Good review. I found the book sometimes jaw-dropping and sometimes so pedantic that I just got bored…it took me months to finish it.

    Can I buy a clue as to which name tops Hiro Protagonist – Aida? Doob? Dinah? Julia? All seemed interesting but I must have missed the joke…

    • ursus

      I’m pretty sure its Sonar Taxlaw.

      • Daniel Freeman

        This is what I had in mind 😀

  5. Ben

    The way Stephenson describes it, I didn’t quite get what could change the pieces’ velocities such that so many of them would end up falling into the earth.

    I was picturing 7 chunks still extremely close together and all basically maintaining the same velocity. When the first 2 collide hard and fast enough to be a “watchable” event where people can see 1 of them break in half, I was very confused.

    While I still don’t quite get it, this helps!

    • Anonymous

      I think the idea was that as they collide, they spit off bits and pieces, which would then reenter the system or enter the earths atmosphere. In the case of the former, they would add some energy and spit out more chunks, and we know what happens to the latter.

      Not sure if that’s remotely correct or even possible, but that’s what I thought when reading it.

  6. Paul Botts

    Regarding the 7 big moonchunks, I had two questions about potential Earth responses once it was understood that the big pieces would in about 2 years results in the “hard rain”. Both of these questions presume the ability on the part of Earth’s spacefaring nations to hit an individual moonchunk with an Atlas or similar rocket that is carrying a nuclear warhead.

    (1) What if instead of all 7 moonchunks disintegrating at the same pace they could be “picked off”, i.e. blown apart, one at a time? Nuke a moonchunk, deal with the planetary damage resulting from its individual level of “hard rain”, wait three months, rinse and repeat. It’s gonna be bad, lots of people will die, the Earth’s ecosystem will be seriously damaged. But maybe by spreading out a bit the damage ends up being survivable?

    (2) In the novel the scientists realize that one of the moonchunks, the one that was the moon’s core, is mostly iron. Maybe that one unlike the others could be _moved_ by a nuclear weapon rather than disintegrated by it? And if so perhaps a nuke or two placed in the right spot on the underside of that moonchunk could move it out to a higher orbit and hence remove it from the orbital plane of the other moonchunks? Thereby taking that moonchunk’s material out of the escalating collisions and so reducing somewhat the eventual “hard rain”?

    • Daniel Freeman

      Without doing more than a back-of-the-envelope calculation, I think the problem with (1) is that nukes don’t come even nearly close enough to having enough energy to break apart chunks of planet (and moons are essentially planet-sized for these purposes):

      All nukes that have ever been detonated have yielded about 2 million TeraJoules of energy. Let’s be optimistic, and say that earth could scrabble together 1,000 times this, or 2 billion TeraJoules of nukes ( The moon weighs about 7 * 10^22 kilograms. So, let’s charitably call a moonchunk 1 * 10^21 kilograms.

      That’s enough to change the total velocity of the moonchunk by about half a meter per second, max (if it’s already in motion, you’ll have an even smaller effect because kinetic energy scales quadratically with velocity.) (*+10^9+teraJoules%29+%2F+%285+*+10^21+kilograms%29%29).

      If you were *extremely clever*, you *might* be able to do something with that half meter-per-second.

      • Paul Botts

        Oho, yea that’s several orders of magnitude more mass than I was picturing. Thanks for clearing that up.

      • Anonymous

        I just started reading Seveneves and had a similar issue. I’m a nuclear engineer (from Berkeley) by training, so I decided to do some basic calculations.

        The short answer is that it would take between 31237 and 3123774 nuclear weapons, depending on how powerful we can make them (between 100 times the size of tsar bomba and just the size of tsar bomba) to prevent the catastrophe from happening. In essence, it would take these many nuclear weapons to fragment the six large moon chunks to a size of 50m, which is approximately 2 times the smallest size of a meteorite that can be burned up by the atmosphere. This is assuming that the chunks are of similar density and are spherical; and assuming that the loss of one chunk doesn’t affect how fast/slow all the other chunks are hitting each other. If we created a computer simulation, I am sure that the fewer the chunks the slower the exponentiation, implying that the time to the hard rain would decrease exponentially after we destroyed each moon chunk. Aka it’s a solution to the problem that they didn’t consider, thereby potentially destroying humanity. Tsk.

  7. Alex Rosser

    Some technical quibbles.
    1) Angular momentum. As illustrated in the front end-papers of the hard-back, the axis of the torus is aligned with the direction of motion of Izzy plus asteroid. As the asteroid acts as a shield, the whole shebang rotates around the port-starboard axis once per circuit around the earth, once each 90 minutes. But the gyroscopic effect of the torus would resist this, resulting in considerable stress on Izzy’s structure and on the bearings at the centre of the torus.
    This problem can be avoided by having the axis of the torus parallel to the port-starboard axis, which is also the axis of the circle that Izzy is making around the Earth. So it should look like a trailing bicycle wheel. Another solution is to have a matched pair of contrarotating tori.
    The banana, the meeting room, would unbalance the torus when filled with lots of people. More stress on the bearings, or better still automatic counterweights on the spokes.
    The bolos seemed a good idea. But once spinning they would take as much enery and propellant to stop as it did to get them started. Reeling them together would just increase the spin rate until the occupants were squashed and/or the cable broke. Entry and exit would have to be accompanyed by crawling along the cable to its middle.
    2) Mine cooling. All mines need cooling as well as ventilating. The Diggers would have had a major problem sucking in hot air from outside, then using a heat exchanger to expell it even hotter and thus cool their habitat. Where the inlet and outlet pipes emerged out of the mountain would be continually hit by bolides, and hence need continuous repair. Difficult by remote control. A huge problem that deserved a mention. And where did the energy come from? A nuclear reactor, even if it were a fast breeder, would require many tons of Uranium over five thousand years.

    • Anonymous

      They did add the second torus, though I cant recall if it was rotating counter to the original. I think it was. That was during the section all about how they were making improvements to izzy, so I could see how that all would get lost in the shuffle.

      Perhaps the mine cooling pulled a reverse geothermal, expelling the heat into the earths crust? That would be a much lower temperature that the surface, post white sky, and thus more efficient than pulling in hot air and adding heat to it. Also, in keeping with the adaptations of the diggers and the swimmers (whatever they were called), perhaps the diggers can handle hotter environments.

      Not sure how the energy came about, though. That stumps me.

  8. Anonymous

    Which name do you think is better than Hiro? Because I’ll never forget Hiro, but I don’t remember many of the names from seveneves (or anything I read, for that matter). That doesn’t mean they’re not better, just not memorable to me personally.

    • Anonymous

      My initial guess was the MacQuarie Family, but is that guess deep enough?

  9. Anonymous

    Some thoughts:
    1. You have an energy sink: The energy required to break rock. You could model this by assuming each rock splits in half on each collision. Or more likely, to split into a hundred, a thousand pieces. (The premise that anything could be gentle enough to only produce 7 chunks is one that I’m skeptical about.)

    2. I don’t think that the collisions are going to be anywhere elastic. You can get some numbers on this by looking up the energy requirements for ball mills used to turn metal ores into powder.

    3. If the cloud is still a small patch of sky after a week, then the bulk of the energy will just decrease random velocities of rocks, and the moon will recollapse.

    4. The only situation where you will get significant speed increases relative to the overall center of mass will be when a small object hits a large object. Under optimum conditions the small object’s velocity doubles.

    5. The moon as a whole has about a 2 km/s escape velocity. A dispersed moon will be lower than this. Suppose that it is 1 km/s. What materials can undergo an elastic collision at even a few hundred meters per second?

    6. For the orbits to be even short term stable, you need to add a bunch of angular momentum. The moon is close to non-rotating. The orbital period of rocks that are within a 10000 mile perigee is going to be much shorter than a month. Their apogees are going to be self intersecting. Remember that once ‘launched’ an elliptical orbit will return to the impulse end point. These aren’t elliptical.

    7. Hmm. A better model may be give a cluster of N objects, initially together, random velocities subject to the following constraints: Total linear and angular momentum is zero. Resulting velocities are still gravitationally bound. Every time two objects come within 1 R of each other both objects split in 8, R is replaced by 1/2 R, and mass is replace by 1/8 mass, the kinetic energy is reduced by half.

    Under these conditions, do you still get ejection events? My guess is, ‘yes, but rarely’