How to take down the internet?

Question

The internet is a mesh or "net" of computers, all constantly establishing and re-establishing connections with each other. For this reason, only temporarily do servers go down, before they are back up and running again (i.e. "This server is temporarily unavailable. Try reloading in a few minutes").

The Internet is the global system of interconnected computer networks that use the Internet protocol suite (TCP/IP) to link billions of devices worldwide.

The internet consists of a vast number of hosting devices. What I am looking for in an answer is complete and utter eradication of the internet so that no information remains or can be salvaged.

How do I take down such a hugely integrated and widespread communications network?

I do not wish to knock out the power grid by EMP-ing the entire surface of the planet, I only wish to knock out the internet, other devices may still function after the destruction of the network

Edit:

I would rather the infrastructure to remain, as opposed to destroying all of the associated communications networks. I would like this to be something a non-superpower country could accomplish with the current standings of North Korea, for example as a backlash to security threats.

If it is only possible through strategic bombings or other infrastructure destruction, feel free to include it as well.

Answer 1

The distributed nature of the internet makes it really impossible to destroy all of it, and all the information on it, without destroying every computer as well, which you seem to be excluding. To erase all access everywhere you must contact every computer on the internet.

In particular I need to ask what you define as 'the internet'. You have a rough hierarchy of computers connected, how many interconnected computers is 'the internet'. Is an intranet at a large company still 'the internet'? What about an intranet of a small company all within one building? My connecting my three computers at my house into a tiny intranet? Is an internet that can connect to all clients hosted by my local ISP only, but not make the 'hop' to a higher ISP (meaning can't reach beyond my city) count as internet? Depending on how loose your definition of internet is it can be nearly impossible to take out the internet (I plugged my cellphone into my laptop. That's two computers communicating, it's the internet!)

I don't believe any of the posted answers so far are viable ways to take out, or even more than mildly inconvenience the internet, but that isn't the answer's fault, it's nearly an impossible challenge without an apocalypse that destroys all physical computers.

However, there are a few approaches that I can see which you could use to varying degrees, depending on how much you need to actually destroy the internet for your story:

Remove access to large portions of the internet easily

note, this solution also removes phone, cable TV, and most other forms of electronic communication in addition to internet

It's not that hard to isolate parts of the world from large parts of the internet. Right now data going between continents travels through a few dozen wires spanning the ocean. Not only is the number of lines relatively small, they are completely unsupervised, you can mess with them without anyone seeing you.

Cutting a cable is not entirely trivial, but also hardly impossible. You would need something capable of diving deep enough into the ocean to get to the cables, with sufficient cutting power to cut through cable and the protective encasing. This is more expensive then a regular individual can manage, but even a small nation can probably get together enough money to get a cable-cutting sub built. If you don't want to destroy the infrastructure, tricks to disable the communication without fully cutting the wire surely can be made easily enough once you reach it.

People could repair these connections, but as long as you have a cable cutting sub you can cut new connections as fast as they can be repaired; it's nearly impossible to watch all of line from the US to Europe at one time, it's just too long to properly watch for malicious attacks against it.

Imagine Australia if the internet lines to it were cut. Most hosting for large companies and services are located out of Australia, they would lose their Amazon and Wikipedia and many other major sites they use regularly. They would still be able to communicate with anyone in their home continent, and in fact many major businesses may have servers in Australia for latency reasons that would still work, but the internet would be a fraction of what it used to be. For all intents and purposes Australia will be missing most of the power of the Internet.

Other places will do better. The US has the most internet infrastructure of any other nation, so many of the big names will have servers within the US. Anyone in Europe will have access to any server on any landmass connected to Europe, which is the majority of all land, so they likely will also not suffer as much; However, both the Americas and Europe/Africa would suffer from lack of connection to the other major land mass. The UK and Japan, two first world nations who make huge contributions to the internet in general, will suffer significantly since their small size means relatively small number of major services have servers within their nation.

People within these nations would still be able to use satellite internet, but considering the small number of satellite and the huge number of people there would be significant bandwidth issues, to the point of making the internet effectively destroyed until the countries figure out a way to regulate limited satellite bandwidth in some manner. In theory a country could also try to take out satellite as well. This is harder then taking out the physical lines, but not entirely impossible.

Take out HTTPS

This probably isn't what you want, but removing security from the internet will significantly weaken its usefulness, if not destroy all of it. Right now our secure internet is dependent on two things:

1. No algorithm can decode keys as efficiently as we can encode them
2. We trust our top level Certificate Authorities

If either of these presumptions were proven invalid security falls apart. Suddenly with minimal effort one can set up a man-in-the-middle attack and get all my banking information, taking my money over night. They can also inject misinformation or viruses into any request I make online, since they can simply replace messages with whatever they want me to see. This would not destroy the internet, but it could have horrible repercussions.

Getting access to the private key of just one CA would allow quite a bit of evil, but publishing an algorithm that allows decrypting of a message using only the public key would truly break security since all CA are equally invalidated at once.

The internet knowledge still exists with this approach of course, it's a limited solution to what you want at best.

Edit:

I added this option, knowing it's quite different then your stated goals. However, I think that, while not breaking the internet, this has room to have a significant affect on internet, economy, and more. I've asked a question inspired by the idea just now, perhaps it will give some ideas of how much harm something like this could lead to: How can a small country exploit breaking HTTPS to destory a larger country while growing its own strength?

Big bad scary virus

The other option to get closer to what you want is a virus. Of course I don't consider it at all realistic that a virus would actually be able to take down the internet, but you get some poetic license here. If you truly want all the internet to go away I suggest using a hand wave virus without explaining exact details and ask your readers to give you some suspension of disbelief; because any answer that does what you want is going to require quite a lot of it. Ultimately it's better to give an answer where you don't explain some details than to try to explain everything and, in so doing, make it obvious to informed readers just how improbable your suggestion really is.

There are some things you can do to make the virus idea more believable, to at least get a little leeway from readers.

Imply that someone utilized some 0 day exploit (i.e. found a bug no one knows about and exploited it before anyone knew it existed to defend against it) in the very infrastructure of systems that meant that you could spread a dangerous 'virus' in such a way that anyone trying to connect to different services will likely contract it. Basically, someone found a huge security hole that nobody realized was there and exploited it to make an uber-virus before anyone could defend against it.

Usually, protecting against a virus after it's released isn't that hard - people will find ways of blocking it and spread that information - but they spread it with the internet. Thus, if a virus did worm its way out there and knock out the internet in one go, the method for identifying its root cause, creating protective measures, and spreading them to others would be greatly limited. The virus could also be destructive (for instance, permanently removing data from infected systems to render the information inaccessible).

I would go on to say that the virus was subtle in the manner it was crafted. Someone released the virus a while ago and gave it time to spread and infect most ISPs before activating it. In fact say they also infected the certificate authorities and got hold of their private keys. It can now do a perfect man in the middle attack, getting through SSL security by pretending to be your security authority. Whenever you attempt to hit a website it can instead send its own virus back to your computer to spread further.

Suddenly all internet traffic is disabled because your ISP won't forward your requests, and many devices attempting to connect to any site anywhere are instead infected by the super-virus. Everyone is now too afraid to go online or have already had their devices infected, and depending on how nasty the virus is even after new ISPs are built people will be afraid to go online until they find ways to fully protect against the virus and stop it from spreading to the new ISPs via infected systems that connect to it.

Answer 2

It is not possible, because the Internet is not a thing - it is a relationship.

For example, imagine 20 people in a book club. The book club is not a thing that can be pointed at, it is the relationship between the 20 people all talking to each other about the books they have read. How would you "destroy the book club"? Well, you could stop all the people from talking, that would seem to have destroyed the club. However, as soon as two of them meet again and start talking, the book club is established again. The only way to really destroy the book club is to kill all the people.

The same goes for the Internet. The only way to really destroy it is to destroy every one of the billions of devices talking to each other. This is not possible, short of annihilating the entire surface of the planet. Well, not without magic (or sufficiently advanced technology :-), anyway.

A single computer virus wouldn't work, as the billions of devices are running hundreds of different operating systems, many of which can't actually be modified by software. I guess you could do it by writing multiple viruses, one for every operating system version used across the planet, but that isn't practical (or even possible?). In any case, all of the important machines have multiple backups, which a computer virus can't destroy.

Answer 3

This gives a clue on one way to do it: Someone Just Tried to Take Down Internet's Backbone with 5 Million Queries/Sec

Botnet tried DDOSing the main root DNS servers, and knocked out 3 of 13... Which means there are 13 targets that would need to be taken out to cripple most of the internet.

They were using a DDOS attack, but a well organized physical attack could work better, and be longer lasting.

It probably wouldn't take it down forever, or even very long, but as a first strike it could be very effective.

Answer 4

If you want to take down the internet you want to take down BGP. This is the protocol that internet providers use to exchange information about which networks are connected to which internet provider.

Without BGP the internet providers wont know where on the internet computers are. The internet will essentially fall apart. Each provider will still be able to communicate internally, but they won't be able to communicate with each other. And many users likely will perceive the internet as being completely down, since without knowing so they are completely depending on some service outside of their chosen internet provider.

Since BGP itself runs over the internet there is a possibility for cyclic dependencies. If you can identify any such cyclic dependencies and take down every BGP session in the cycle simultaneously, then it will be harder for the provider to get the network working again and consequently it will take less resources for you to keep the network down for a prolonged time.

Several years ago there was a concern that it is easier for an attacker to take down a TCP connection than it ought to be. Taking down individual end-user connections this way is not particular interesting for an attacker, but if the attacker could target BGP sessions instead, each TCP connection the attacker kills has a much larger impact. An MD5 digest option has been introduced in TCP to protect against this attack. Because of that protection that particular attack scenario probably wouldn't work today, but there are other possibilities.

You may be able to overflow the routing tables with too many routes. Last year one internet provider did that by accident and knocked several other providers off the internet for some time. If you combine such an attack with route hijacking, you may even be able to get legitimate networks to help you by sending out more announcements of the networks you are trying to hijack.

But the most effective attack would be the one that nobody else knows about yet. If you can find a 0-day vulnerability in the most popular BGP implementation among internet providers and attack that, then you could probably succeed in taking down the internet until they figure out how you did it.

Answer 5

@dsollen is really right, taking down the internet is not very feasible, short of total annihilation of most of the planet surface and/or the entire planet.

One possibility remains however, which is a bit less destructive than complete surface annihilation:

Large scale Carrington event

A solar storm on the level of the Carrington Event could theoretically take out almost the entire electronic infrastructure of the planet in one swoop. True, there will be small remaining networks consisting of servers and grids which were specially shielded against EMP and the likes, but with most of the world-wide power supply gone there goes also 99.99% of the grid, publicly accessible servers and end-user appliances.

Note that a Carrington Event is a fairly apocalyptic scenario, which would throw the world as we know it into utter chaos. The solar storm would fry the entire electric grid, including large scale capacitors/transformers that would take years and years to rebuild. No electricity would mean that gas stations stop working, which means the supply lines break down almost immediately.

Answer 6

General advice: learn how the internet works. It's fragile in many ways. There's actual scientific literature on how the internet is vulnerable to attacks.

One observation is that there are relatively few tier-1 network providers, each being integral to the internet as a whole. Take down a significant number of these, and the global internet is dead. One way to do that is to just pay off all of the staff in the relevant departments, e.g. by offering them better-paid jobs. They walk out en masse, failure is only a matter of time. Direct sabotage is also an option, of course. A few bad routing tables at this level have serious impact.

Answer 7

Back in 2008, a misconfiguration by Pakistan Telecom caused all of YouTube's IP traffic to be routed to Pakistan. Hopefully they've fixed that problem by now, but there might be other router-level security holes you can exploit to misroute traffic, potentially causing confusion and slowdowns across a large area.

Answer 8

Subvert, extort, blackmail, recruit, or otherwise gain control of these guys or their keys.

Answer 9

This is going to depend on what you call "The Internet". If you use the definition used by my grandmother, just change the "A" DNS record for aol.com. If you use the definition by my niece, she'll understand the Internet is down if you just change the "A" DNS record for facebook.com

If my ISP were completely unavailable, I could still use the IPv6 protocol, or the IPv4 protocol, for my own equipment (assuming my own computers worked). I have a networking savvy friend who I could also connect with fairly quickly.

It may be worthwhile to understand that an Apple ][ can be used to connect to the Internet.

So, to be specific, you probably want to make the "backbone" non-functional. There is no reasonable way to do this. People have looked into turning off the IPv4 Internet, and it was deemed to be unfeasible (without causing unacceptable collateral damage).

"I would rather the infrastructure to remain, as opposed to destroying all of the associated communications networks."

However, the associated communication networks have started to rely on the Internet. There is no way to turn off the Internet without massive collateral damage.

eWeek diversity quotes Verisign: "It's important for us to maintain the reliability of all the services, so we don't rely uniquely on any particular implementation in the operating system space," Kaliski said. "Having both FreeBSD and Linux makes it possible to have that diversity."

So much of our society has started to use "Internet" connectivity, due to cost advantages, that any attempt to dismantle the Internet would result in people re-creating the cost benefits, by re-creating the Internet. If every Microsoft Windows machine and Unix server were somehow magically wiped clean, people would instantly create new networks, using IP (because enough people DO have the IP communications details memorized) using whatever equipment remains, even if those are Apple ][s, Commodore 64s, or Pepsi Cola cans tied together with strings. We've learned enough about the benefits that many people would be getting functional "Internet" communications up and going even before most people heard that the old one disappeared. If we had to use "Walkie Talkies" to transmit via Morse Code, we would start with that, and keep improving from there. Once people are creating Internet communications, the only remaining step is for them to use the communication infrastructure to connect their different pieces of self-created Internet.

If you're looking for a fictional scenario, you'd probably want to envision a different Internet, which uses addresses that are generated by some key piece of technology that breaks. Or, you might want to fictionalize an attack of the people (kill the scientists) rather than the physical technology. "The Internet", as a widespread global phenomenon which is highly popular, could be substantially shrunk if a sufficiently powerful and inhumane, hostile government enforced a new prohibition of wired or wireless communications. They might make many people so fearful that they don't consider Internet usage to be a worthwhile risk. However, there would certainly still be an "Internet" underground, just like people today have the "Dark Web" using Onion/Tor/etc.

Perhaps a more realistic fictional scenario would be to assume that firewalls became defeatable (they essentially shut down, denying all traffic). It might be a story that sufficiently convinces the masses. However, the truth is that trained technicians would figure out a way to handle this, just as people do use the current Internet (e.g. E-Mail) despite constant spam, existing DDOS attacks, etc.

In this universe, people embraced standardization, and communication with other implementations, and so there is no simple point of failure. The Internet was built with the concept of flaky equipment in mind, which is why it "heals" so well if damage occurs. Since it is so resilient, rather than vulnerable, nothing short of EMP-style destruction would stand a chance, and the ability to recover would only be limited by troubles being able to run automated machinery again. If you're looking for a super-simple attack, I'm going to stick my neck out by saying: There isn't one. That's by design.

Answer 10

@dsollen has really given the correct answer to this. But I want to expand on the computer virus part.

Any normal computer virus will be discovered, detected and countered within days. As @MichaelHampton mentioned, the Morris Worm took down much of the Internet in 1988. The first day was chaos, but then the important routers was reconfigured to stop the worm from spreading on the Internet backbone. After that it was a local problem and cleaned up in a matter of days.

Since 1988 we have become far more paranoid about how we do the Internet. Pulling something similar today would be nearly impossible.

What is needed is an Artificial Intelligence virus. I.e. Skynet.

This AI would be a specialist in discovering new security holes in all operating systems. When it has access to a computer with a given operating system, it reads and reverse-engineers that OS, finding new holes. As one hole is closed, it switches to another.

To get it started, our clever Hackers manually gives it access to machines with the major OSes, for computers, network routers and phones.

First it needs a quiet phase where it just spreads. Then, it strikes...

If you want to, you could leave computers running perfectly, except that their network cards somehow doesn't seem to work.

During the pre-strike phase, it can give fake work orders to build a second independent internet for its own use. The AIs' various parts can use it to spread information about new security holes. This net can be kept running after the human Internet is down. If the Hackers ask nicely, they might be allowed to use it too.

Even so, it would probably not be a total victory. There are many OSes out there. If the AI doesn't have any infected ObscureOS computer to experiment on, it would have difficulties finding that first security hole to get started. This will make ObscureOS computers immensely popular until the Hackers get their hands on one and break it open. Que EvenMoreObscureOS.

If the superpowers manage to track down the Hackers, they will not be gentle.

Answer 11

NTP Amplification / Distributed Denial of Service (DDoS) attack.

NTP = Network Time Protocol. It's used to sync internet-connected machines and their clocks. This sort of attack is when a packet is sent to a server but the response is disproportionate to the original packet's size. It utilises the UDP protocol which is basically sending larger sized packages as opposed to TCP which is faster and smaller.

NTP amplification is reflective. The response is sent to a spoofed IP address. The attacker sends a packet, and the server responds to the spoofed IP with a huge, disproportionate response. It targets NTP servers which support the command monlist, which returns the last 600 clients that connected to that server. With a fully populated monlist response, it responds with heavier amounts of response data. Another command is any, which returns DNS zone information.

With a collection of NTP servers, you can easily redirect entire megabytes of traffic and data to a built network of machines and servers. If you have enough resources and data, it's a wonderful way to attack the entirety of the internet, or at least a huge portion of it.

The power of the amplified attack is that the return is much larger than the request. Depending on your connection, you can flood everything with Mb or even Gb level traffic.

If you sustain the attack and/or combine it with a physical attack (perhaps an ISP/DNS/intercontinental cables terrorist group) for long enough, you can likely disrupt the financial stability and infrastructural stability of the internet enough to render it, in a fashion, eradicated.

Answer 12

In the absence of a "hard science" tag, there are a number of science fiction tropes which could do the job...

• A metal eating virus or nanite swarm which loves to munch on wires and hard-drive platters. ...or a processed-silicon-hungry variant of either which eats all our chips.

• Earth gets invaded by Aliens who strip the planet of metals before moving on, trapping humanity in a permanent pre-technological age.

• A shift in the local laws of physics, either increasing the electrical resistance of the metals which we make wires from, or increasing the magnetic decay rate of the metals we make hard-drives out of.

• Kill off all the humans who know how to use computers.

or my favorite...

• A computer virus which eats all of the "\" and "/" keyboard input, leaving us with no way to change our computer's current directory and/or website URL.

Answer 13

Destroy most of switches

It is plausible to accumulate a remote exploit for multiple popular families of commercial network switches - i.e., not your home wifi/router, but the ones that your ISP would be using.

Such a remote exploit can be used to develop a worm that can rapidly (timescale of hours) scan and infect a large portion of global networking hardware.

It is also plausible to have a payload that will physically damage the hardware, for example, by overwriting firmware chips to 'brick' the unit. It's not trivial, but such attacks exist.

Having a worm that rapidly spreads and at a set time destroys a large proportion of networking hardware can take down most of internet for a prolonged time - sure, some fragments will remain functioning, but you would expect many data centers to lose connectivity and be unable to quickly restore it simply because the replacement hardware would be very scarce until more can be manufactured.

Answer 14

If your world had a large conspiracy among relevant hardware manufacturers, a Hardware Trojan could act as a kill switch inside of chips vital to the continued functionality of the internet. We have suspicion of this being done already in military operations. In Operation Orchard it was probably radio frequencies. In network hardware it could be looking for a specific packet.

Hardware Trojans can be really hard to detect; there isn't really good software to do it automatically because of the complexity of many hardware designs, so you need expert eyes on every design you want vetted. And even then there's no guarantee that the chip design you send off to manufacturing is the one you get back; you have to physically scrape layers off of the chip and painstakingly map things on the chip to things in your design to figure that out. The problem here is that the process destroys the chip, and only examines a single chip.

On the other hand, Hardware Trojans are also hard to design. Your chip has to be about the same size, take in and put out about the same power, etc.

Due to cost concerns I don't think a lot of effort goes into looking for hardware Trojans relative to their potential for damage, so to me this seems believable.

Answer 15

I really like @kasperd's idea of using BGP do destroy the routing information between the Autonomous Systems (networks) which the Internet is comprised of. It is relatively plausible, but the effects are unlikely to last long - system administrators around the world would scramble to define routes between trusted hosts manually. The Internet would soon rise from the ashes, but maybe a bit slower until a BGP-like protocol with cryptographic signatures takes over.

Alternate proposal If your work is not set in today, but in a plausible future:

Take down national firewalls

Prerequisite:

All over the world, national parts of the Internet are separated from each other. Think "Great Firewall of China". The stated reason for this are some high-profile things, like terrorist organizations spreading propaganda, or simply foreign companies refusing to pay local taxes. If you like, you can also add others who campaign for such things for their own reasons, e.g. people wanting to twart copyright infringement.

Once the firewalls are in place, they can be attacked and taken down, thereby taking others with them.

Answer 16

Most of the core routers and DNS infrastructure is maintained by a rather small group of people who know each other well, and who have very efficient communication channels that they normally use to keep things floating, e.g. if there is a large-scale DDoS attack going on, this is where countermeasures such as disconnecting links and dropping certain traffic are coordinated.

The Internet works this well because this group is self-selecting and interested in keeping the network running. Now if there were an event that would convince this group of people that it was preferable to shut down the international links, they could do so, likewise for the DNS roots.

Answer 17

Finding a bug in DNS or common routing protocols, etc. could simultaneously take down all the redundant networking infrastructure.

It's not likely in the real world but it is fictionally plausible.

Answer 18

Information stored on the internet still exists if you destroy the internet because most a lot of important information has backups. Even an EMP wouldn't make the knowledge of Wikipedia to be lost.

If you however just want the internet to breakdown make a party attack intercontinental cables. If those get destroyed the global internet as we know it breaks down.

Answer 19

There are a lot of good answers, and mine is more of a tie-in than a completely new concept.

Destroy the Hubs:

The Global Internet, as another poster mentioned, is tied together by a countable number of fiber optic and copper cables. As shown here.

The other interesting thing is that they go to a countable number of cities, and from there are distributed to hundreds of adjacent cities and counties. They all follow the old phone lines, with follow the telegraph lines, which follow the railroads. Hundreds of years of consolidating our infrastructure into a few key places.

Two dozen nukes would take down the entire US, given proper placement. And its not just DC; Kansas City is a good target.

And the pattern is repeated globally, so you can keep most of the hardware or current day, but still take down the Internet (and a bunch of other stuff too) with a few focused attacks.

Rebuilding and Rerouting:

As others have mentioned, this wont forever destroy the Internet. Data will get rerouted and so will everything else. But I picked nukes for a reason: By denying the area itself, the entire city and surrounding area, you will severely limit the effectiveness of the rebuilding effort.

Those places were Hubs for a reason to being with, versus whatever city is 50 miles upwind, and those reasons don't go away. Data, goods, and whatever else will start flowing between places again, but "The Internet" wont look anything like it does today. No more Facebook or Youtube, the prices for 'decent' internet will skyrocket, and only major businesses and gov't will be able to use it routinely - everyone else will not only be stuck without Internet, but phone, cable, etc.. will be gone from a regular persons life too. The upheaval will last quite a long time.

Answer 20

It is impossible, within your conditions.

Note: your given requirement to shut down the internet is without leaving any information available. That is even more strict as your title asks.

The internet is no media where something can be stored in. It is just a..... well.... network of connection between hosts. So to make any data unavailable, you had to destroy binary media. So any mobile, any pc any fridge that works wiht a controller.... and anything that would be able to physically store data in anyway that could be retrieved by such a medium connected to the network.

You know, the internet isn't a thing. even if all computers were eliminated, nothing would stop (expect the dumpness of the idea itself) humans to communicate verbaly over tcp/ip as language. What theoretically could be fitting into the definition of an network based on that protocoll aswell.

and so, since the protocoll itself is jsut an idea, you can't destroy it without eliminating each medium hosting it.

In other words your definition of taking down the internet is "extingushing humanity" since you require to make no information of it ever be available again.

so as others have allready adviced, but in other context. The only way to achieve this is:

Nuke the planet containing the idea and its mediums storing it out of existance. Everythign else will at least leave the possibility of medias (of what ever origin) be starting again to communicate over tcp/ip ;)

Answer 21

Stop filtering spam and viruses.

The pest will grow all by itself and take up any possible room^Wbandwidth, like a gas.

Answer 22

I believe the only sure-fire way of accomplishing this cleanly would be through an uncrackable, artificial-intelligence-based virus.

The virus would have to be:

• programmed with unbreakable coding
• completely dynamic, made possible by AI programming techniques
• spreadable by many technological avenues
• able to exploit human psychological/sociological tendencies

Answer 23

A Coronal mass ejection that is big enough, could take out the entire power grid.

In a “perfect storm” scenario, when a high-power coronal mass ejection (CME) of charged particles slams into Earth at a time when the delicate balance operators try to maintain in electric power grids is precarious, the resulting damage could take a decade to repair at a cost very roughly estimated by the National Academies of Science as high as $1 trillion.

If this happened, it would take out the power grid, for about a decade, and without power, all out the computers, servers, and everything would run out of power, and die. RIP internet, for atleast a decade. Plus, it would cost tons of money to fix (1 trillion).