In a world set many years (At least a generation or two) after a near extinction level nuclear event, how could a recovering population set up a internet or internet like service?

Key points I'm interested in:

Hardware / Software (if applicable) needed

Infrastructure needed to be put in place to accommodate.

Limitations of this type of internet service (Wired only? Wireless to Xm?)


This question asks for hard science. All answers to this question should be backed up by equations, empirical evidence, scientific papers, other citations, etc. Answers that do not satisfy this requirement might be removed. See the tag description for more information.

  • $\begingroup$ Why do you need the Internet to connect a few thousand people? This was a near-extinction event, after all. The benefits at this small a scale may not outweigh the investment to set it up. If the population is geographically localized, the benefit is even smaller. $\endgroup$ – Nuclear Wang Jul 19 '18 at 18:22
  • $\begingroup$ What if the survivors are determined to never let another internet happen to help prevent the catastrophe from happening again. Maybe if people hadn't had to view other people broadcasting **** on the internet, there wouldn't have been a nuclear war. $\endgroup$ – Hosch250 Jul 19 '18 at 18:31
  • 1
    $\begingroup$ The use case for post-apocalyptic that I see is that there are well provided pockets of survivors, with electricity, computers and data libraries. Those pockets want to share their data without the need to travel with USB drive. $\endgroup$ – Alexander Jul 19 '18 at 18:33
  • 7
    $\begingroup$ What exactly do you want your people to have? You say "Internet or Internet-like service", but that's on one hand very broad if we take "Internet" to mean what it means to most people, and on the other probably doesn't mean what you think it means if we take the technical meaning of "Internet". Having basic computer-to-computer networking, such as terminal sessions to access files on one computer from another, for people who are willing to jump through a bunch of technical hoops, is very different from having Facebook or Google or Stack Exchange accessible to everyone without effort. $\endgroup$ – a CVn Jul 19 '18 at 18:56
  • 2
    $\begingroup$ I definitely second @MichaelKjörling's comment, TheBlueTurban. We need to know what sort of services this internet could provide, and what criteria it would need to meet, especially for a hard-science question. $\endgroup$ – HDE 226868 Jul 20 '18 at 14:10

It only takes a generation or two to build an internet from scratch

Think about it: One of the key predecessors of the Internet, ARPANET started in 1969. Look where we are today - or even 20 years ago. It did not take long to go from "nothing" to the Internet as we know it. That was starting with:

  • Low speed modems
  • No (or very minimal) networking protocols
  • No web servers
  • No FTP servers etc.

As I often say, Knowledge is the Key. Even if there was NOTHING left except the very basics (e.g., even a few old 16-bit computers would be enough to get started) and some determined hackers, it would be possible to start essentially from scratch and build up to full 21st-century level internet in a matter of a few years (guesstimating 1 - 30 depending on the hardware issues).

About the only thing that will be really hard to do is to create a chip fab. So a key will be salvaged hardware to get through the first few decades.


Even assuming that all large equipment on-grid and most cell phones and other wireless devices were "fried" by the apocalypse due to EMP and other effects, there would likely be some equipment that survived. It only takes one server with a full LAMP stack to get things jump-started as far as software. A few computers, routers and other equipment (and really a computer can be a router and a router can be a computer - as with the first days of ARPANET, there is no fundamental difference) salvaged from the rubble somewhere and you have all the hardware you need to get started.

DNS and some other pieces will not be so easy. But if you are in a small group then you essentially run as one large local network and, if necessary, hardcoded addresses. Not hard to do with a little bit of surplus equipment.


In addition to hardware, your new Internet will require software. Easiest solution by far is one Linux machine with everything on it. Which is not that unusual. Everything else is copies of copies plus a few hackers that start building useful web sites. No Amazon for a few years - but only because there are no warehouses full of goods and fleets of trucks to deliver them. But basic web sites for social networking, barter & sales, government, etc. can all be done pretty quickly.

If you have to start truly from scratch then it still won't take a generation. The simple knowledge that it CAN be done, together with some basics of how to do it will lead to a few programmers putting together something resembling a web server in a few years. Very different from 1969 when most of us had no idea what we could accomplish.


There are 4 types of connectivity involved. Each has advantages/disadvantages:

  • Wired

Wired connections (e.g., 100 Meg. or 1 Gig. copper) are the easiest. Salvage some cable (every destroyed building is filled with it) and you can put together a network 100 meters at a time very easily. Longer connections are a little tougher because you can't order a line-driver/extender on Amazon, though a few years of work and that can get resolved too. But in the short term, start with a basic 100 Meg. network and you can extend it around town using every switch and router you find.

  • WiFi

WiFi is also pretty easy. Get a few salvaged access points running and you're all set. Actually, WiFi will probably work better than it does now because, initially at least, there won't be any interference! But WiFi is range-limited, and again you won't be able to easily get top-quality longer-range access points for a few years.

  • Fiber

Fiber is the best. But unless you can find a good cache of cable and top-quality tools, fiber will take longer to get online than copper. You can't just splice together bits & pieces the way you can with copper.

  • Cellular and other long-range wireless

This will be the hard one to do. The cell phone network was destroyed in the apocalypse, and even if you can find an occasional functional cell, it is useless without the infrastructure which depends on the phone company's demolished data center. So connectivity will be limited to wires and WiFi for a few years.


Last, but not least, is power. None of this will work without a reliable power source. But everything else in modern civilization needs power too, so that will be one of the top priorities (along with food) for the survivors. Once they have reliable power, the Internet will be easy. While powering the internet will be a lower priority than powering lighting, machinery in factories and many other things, the basic internet will not require much power - a few hundred watts here & there. You won't have big data centers, with their big power requirements, for a while.

  • $\begingroup$ Actually, FTP was one of the earlier protocols. I don't know off hand when a predecessor to FTP was first implemented, but RFC 114 describes "A FILE TRANSFER PROTOCOL" and is dated in April 1971. It's not that hard to follow links back from RFC 959 (October 1985) to there, and RFC 959 describes what we would today recognize as FTP. $\endgroup$ – a CVn Jul 19 '18 at 19:01
  • $\begingroup$ As for power, yes, power might be something that survivors prioritize because of its obvious uses, but power, especially reliable electrical power, is also likely to be in short supply in such a setting. Why waste electricity on running a large number of computers and large amounts of networking equipment, if there are lower-powered alternative ways of accomplishing the goal? (Note that we still don't know what OP is really after; they're asking for "Internet or Internet-like", but we don't know what they mean by that and we don't know why their characters would want it.) $\endgroup$ – a CVn Jul 19 '18 at 19:05
  • $\begingroup$ @MichaelKjörling That is EXACTLY my point. It did not take long from the beginnings of ARPANET in 1969 to FTP in 1971 and much more. Starting with a population that includes a few people with some general idea of how these protocols work (or at least what they can do), it won't take long to come up with a replacement. And if you find one loaded Linux box then you don't even need a replacement, just tweak for your limited network. $\endgroup$ – manassehkatz Jul 19 '18 at 19:05
  • 1
    $\begingroup$ @MichaelKjörling It takes a lot of power to run a modern data center. It takes very little to run a few Linux boxes and consumer-grade routers & switches. $\endgroup$ – manassehkatz Jul 19 '18 at 19:06
  • 1
    $\begingroup$ @MichaelKjörling If "a recovering population" == "a few thousand people" then a "internet or internet like service" could be just a widely dispersed LAN. There are a few aspects to make a network "the Internet". One is wide range - and it only needs to be as wide range as your population done. Another is "web services" - a few Linux boxes can do that quite well done. The third is advanced routing, DNS and other large-scale structural issues. Those will be much harder to handle, but arguably less important in a recovering/newly established small Internet. $\endgroup$ – manassehkatz Jul 19 '18 at 19:17

There are three components to a successful social communication via technical methods:

  • Availability of hardware;
  • Availability and ease-of-use of software;
  • Infrastructure viability and reliability.

I am "porting" a somewhat famous statement of Vitus Wagner that hard-SciFi space travellers would use UUCP. Because TCP would time-out and, well, who uses a packet protocol on a network with hundreds of seconds ping?

UUCP is post-apocalypse Twitter

The post-apocalyptic cyberpunk bands would use UUCP. Either they'd have some specialists from pre-fall era that would correctly identify the best method and provide for it. Or they would build something on their own, similar to the amateur networks such as Fido and the likes.

So, the future internet of post-apocalypse (in my vision) is much more NNTP and much less Twitter.

  1. You need a server for HTTP and the likes. Servers need permanent connection, even worse: they need to be protected and supplied with electricity 24/7.
  2. A distributed infrastructure can crawl from one device to next in a relay manner. No permanent activity or availability is needed.
  3. UUCP and NNTP are old. This requires some ingenuity to even think about them in this context, sure. But these are proven, trusty solutions that worked well when the computers had much less power and resources. This fits post-apocalypse well.
  4. Even if the exact the same protocols are buried in the unholy depth of RFCs no one have read and the pre-apocalypse script kiddies are inventing their own bicycle – similar circumstances produce similar solutions.

    FidoNet was uglier and sort of more convolute than NNTP, but they both function in a similar manner and yield similar results. And a similar culture, by the way.

So, we have allied gangs and small groups that send messages to each other via some kind of an ad-hoc network. The delivery is not instant, but very reliable. No central infrastructure is needed. Strong (and strange) relationships of trust and communication hierarchy ensue.

Sounds like the cyberpunk we do deserve.


Forget wires, get a cell tower running and use mobile data.

If you can get a tower running, with a surviving technician's help, or by reading the manuals, you could reset its systems and tell it that you're its new owner. You could be your own phone company.

Whatever surviving phones you find in the region will mostly all be compatible with that region's tower. You might think that 4G is a ridiculous way to share all data, but that's just because human population density is currently straining the tower's maximum bandwidth, and existing phone companies are throttling 4G and making it expensive. Your phone can actually transfer plenty of data very quickly over 4G connections, it's just that the tower is over-burdened and talking to your phone slowly these days. With most humans gone, and almost none of them having working phones, that tower will have bandwidth for miles. You'll finally be able to transfer data over 4G at the maximum rate your phone is capable of, which is pretty high for modern smartphones. One mobile-hotspot could provide for a post-apocalyptic village. And as the new owner of a cell-tower, you can give your people some very favorable mobile-data plans. Finally, additional data fees will not apply.

  • 2
    $\begingroup$ I'll buy the assumption that one cell tower survived - or more likely that it was under repair at the time so the equipment was disconnected and didn't get zapped and could get reassembled easily. But a cell tower by itself won't provide data to your phones unless you first figure out how to spoof or replace the dead data center that stored the billing records. A phone without a "plan" can only dial 911 - no data or regular calls. EVERY phone would be effectively without a plan because Verizon/AT&T/T-Mobile all went offline the day the Big Button was pressed. $\endgroup$ – manassehkatz Jul 19 '18 at 19:37
  • 1
    $\begingroup$ I'm a programmer by trade, so I'm thinking of this on a systems level. I'm talking about being your own phone company. You don't need Verizon at all. If you've got admin priveleges on the tower's system, you can tell the tower it works for Post-Apocalypse-Wireless now, and that your phone has an unlimited data plan. $\endgroup$ – Jared K Jul 19 '18 at 20:25
  • 1
    $\begingroup$ I'm a programmer too. I just think it will take a lot of work to break into the cell tower system. Then again, the last tech. may have left a post-it note on the cell tower just inside the access door. $\endgroup$ – manassehkatz Jul 19 '18 at 20:32
  • 3
    $\begingroup$ If Fallout is to be believed, the superuser password is written in a diary clutched in the hand of the technician's skeleton somewhere in the basement. $\endgroup$ – Jared K Jul 19 '18 at 20:36
  • 2
    $\begingroup$ I agree it would take a lot of work. But I think the technical work would be about the same as getting a regional ISP back up and running. The physical work of repairing wires would be spared though, which makes it a better solution than repairing or creating miles of coax and fiber. $\endgroup$ – Jared K Jul 19 '18 at 20:42

Ad hoc wireless networks seem like the correct networking strategy here! There are dozens of designs out there, with various pros and cons, and they can be built with a wide range of wireless underlying technologies.https://en.wikipedia.org/wiki/Wireless_ad_hoc_network is a good starting point. Note that "wireless" is often mistaken for meaning "Wifi", and they're really not at all the same; different frequencies of radio waves have different properties. I'm not a radio geek so can't say too much more than "you can get some wireless technologies with crazy range". Keep in mind also that a network can be built with different parts using different technologies; so a lot of low-power wifi nodes can link to a high-powered central node that connects over cable, satellite, high-powered radio waves that can travel long distances to other such nodes... take your pick.

The question of hardware and software is a trickier one. @manassehkatz' answer above about scrounging the existing stuff and building from there is a good one. I'd emphasize if you're really dealing with a wipeout scenario that the processors are going to be your biggest issue. Chip fabrication is hard and requires a lot of prerequisite technology (and some very interesting chemicals and some rare minerals) to implement, and very few people actually really understand everything involved; chip design is a specialized field, and even there, the people working on instruction sets and security features aren't likely to spend much time on the physical chip layout, let alone on the actual fabrication process. And the materials scientists dealing with fabrication are likely to know nothing about the functionality in an x86 architecture.

I'd expect the first several generations of post-apocalyptic computing to be very bulky and slow as a result-- you can create 1980s-level processors with some basic electronic components in a garage without much trouble, and I think the components aren't too hard to manufacture (though i'm not confident), but the smaller and more unified and more powerful they get, the more you go from "need to have power" to "need to have mines, and chemical factories, and skilled workers, and sterile environments with specialized HVAC, and expert processor designers, and..." You will save a lot of the time we as a society took in the development of the computing technology because we can reconstruct and learn from past experience-- especially if anyone has a good library available with high-quality textbooks and manuals-- but you can't get around the industrial base required to build the physical objects, or the complexity of those designs. There's a reason Intel and Apple have thousands of people on the job.


So assuming that some technology still works, and you have proper geeks and greybeards to run it all, there are plenty of options.

First - that which exists already.

IP over ham radio. The entire network was designated for this purpose when IANA first started allocating hunks of IP space. Not the worlds best and some confusion on what you can do with it now (due to FCC rules) but there are active networks using it.

IP over avian carrier. Yes, carrier pigeons. RFC1149 was originally a joke, but it has been implemented several times and has actually been proven to be faster on data delivery than the DSL provided by one South African service provider.

Other ways. Plain old 802.lla/b/g/n can be sent LONG distances with proper antenna at both ends. Get some elevation involved and not a lot of trees, etc. and you can shoot a signal for miles and miles and miles. The issue will be initial set up of routing but setting up a tower chain like Terry Pratchett's Clacks Network on the Discworld would be trivial to do with early 2000 computing equipment.

There are lots different medium that you can communicate over in IP packets - such a plan or possibility is included in the OSI model - physical layer can be anything!

Once you have that, then what language do you speak over the packets? Well, you are looking at low bandwidth stuff. Where possibly everyone is a producer as well as a consumer of content again. So go back to the old protocols. You can finger the next clan over to see if there is a BBQ planned, you can email via smtp or UUCP, you can get web pages or other files via http,etc.


I don't think it would realistically work as a legitimate internet service. I personally think it would have to be a sort of local network. Let's take for example, something like Microsoft Excel. If a group of people were responsible for inputting data, the other might be able to link to the computer. They would have to find a way to connect all devices together like a huge super computer with users that have certain access. I imagine, the system is pretty much in one spot like a library and there are multiple terminals all ultimately plugged into each other. Certain people may be known as scribes and have access to input information into the database. Others would be regular users and only be able to read the data rather than edit it.

  • $\begingroup$ Question is tagged hard science, and this answer is not meeting those requirements. That apart, please read this about excessive self promotion $\endgroup$ – L.Dutch Feb 24 at 3:32
  • $\begingroup$ Welcome to Worldbuilding.SE! We're glad you could join us! When you have a moment, please click here to learn more about our culture and take our tour. Please note that the hard-science tag has a somewhat ruthless mandate. All answers must be backed up with links and/or citations to scholarly articles, documents, etc. to prove the point. No opinions. Thanks! $\endgroup$ – JBH Feb 24 at 6:34

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.