# The internet without computers - 1940s era

Alternate 1900s

Imagine that we are in an alternative Earth in the early to mid 1940s. There has been no World War 2. In this world there are analog computers used rarely in laboratories for specialised purposes. There are no digital computers whatsoever.

Radio technology (still valve based) is however all the rage. Telephones work by having human operators who connect one caller to another by manually plugging them in at the exchange.

Question

If someone in this era invented "the internet", how closely (in terms of functionality and speed) could it approximate today's internet without digital computers? You can assume that memory functions can be carried out by people with notepads. Or, you can invent a non-electronic alternative or banks of unreliable valves or electromechanical relays kept on permanently.

Assumption

Transistors can be considered impossible in this world.

Power consumption and inherent unreliability of components must be taken into account.

Note

• You've got "the internet" in scare quotes here, so lots of clarification needed. What features of the Internet would you want to reproduce? – jdunlop Aug 19 '20 at 18:54
• "Telephones work by having human operators who connect one caller to another by manually plugging them in at the exchange": automatic circuit-switching telephone exchanges were invented towards the end of the 19th century, and rotary telephones started being installed in the first decade of the 20th. By 1920 all large cities in Europe and North America had automated telephone exchanges. But there were small villages which continued to have manual exchanges well into the days of the Internet, and direct international dialling did not become universal until the end of the 20th century. – AlexP Aug 19 '20 at 19:55
• Note that digital computers didn't start with transistors, but with vacuum tubes, aka valves, so you've already torpedoed yourself anyways. Mechanical computers already had multiple ways to record/read data, so no need to involve people for working memory. That aside, there's not much reason to just make a machine available - they were often purpose-built, and took a lot of resources to operate; depending on the kind of data you were looking for, the person on the other end could do it faster manually. – Clockwork-Muse Aug 19 '20 at 23:18
• Relevant: IP over Avian Carriers and RFC 6214. – Martin Schröder Aug 20 '20 at 15:46
• Sidenote: Internet is just connections. telegraphs handled that easy. What your question is really about is the world wide web and other services that use the Internet. – Mooing Duck Aug 20 '20 at 17:09

There were not less than four worldwide messaging systems in widespread use before the Internet. Your choice.

1. The worldwide postal system became available, if expensive, since the late 17th century. Before the establismnent of the Universal Postal Union, correspondents in countries which didn't have direct postal agreements relied on specialized mail forwarders.

2. The worldwide telegraph network was slowly stitched together in the second half of the 19th century. The first successful transatlantic cable was laid in 1865, by the life-time efforts of Cyrus Field.

3. The worldwide telephone network was an interbellic development due mostly to transnational conglomerates such as the British Cable & Wireless, or the American International Telephone and Telegraph.

4. The worldwide Telex network rose to prominence in the years after WW2, and remained in continuous use until the Internet was able to take over its main function of transmitting authorized-ish written messages in almost real-time.

In addition, the shortwave amateur radio community, which boomed in the inter-war period, could function as the message carrier for non-commercial activity, similar to how the internet operated non-commercially in its formative years.

• +1 for mentioning the telegraph network in particular. One might argue that the telegraph office was the internet cafe of the 19th & early 20th centuries – elemtilas Aug 19 '20 at 21:38
• See "The Victorian Internet" (en.wikipedia.org/wiki/The_Victorian_Internet) which describes the telegraph in exactly these terms. Also for fun see "Going Postal" by Terry Pratchett, which has the Discworld version. – Paul Johnson Aug 20 '20 at 17:11
• From 1876 : a couple getting married by telegraph. What's more Internet than an LDR?newspapers.com/clip/31562878/married-by-telegraph – user_1818839 Aug 20 '20 at 22:45
• Y0u might also mention amateur radio. "Hams" have been talking to each other around the world since the invention of the radio. – workerjoe Aug 21 '20 at 11:55
• @workerjoe: Added a reference to the non-commercial internet of the 1980s compared to the amateur radio network. – AlexP Aug 21 '20 at 13:23

# Someone invented the Internet more than a hundred years ago

It didn't work out because of the technical problems. It was called the Mundaneum. With powerful enough analog computers, digital counterparts of the like Charles Babbage had already designed, and monied backing neither Charles Babbage nor Paul Otlet ever had, you could potentially engineer and build custom analog computers that would run the Mundaneum.

The Mundaneum is simply a physical location that is an analogue to the ephemeral location we call the Internet. Its basic functions are essentially identical. With the Internet, I type "difference between dreadnought and parlour guitar", the request is sent into the workings of the search engine, whose automated algorithms return what (I desperately hope) are relevant documents, images, and videos that relate to my question. With the Mundaneum, I send a telegramme to their central receiving office who would then translate my request into the relevant catalogueing scheme, a research team would, using their own algorithms, research patents & makers & catalogues of musical instruments. They might write a short, well referenced report and telegraph that back to me. I don't know if visual data can be sent over telegraph systems, but they could easily mail a microphotograph of the pertinent visual data, along with their written report, which I would then have blown up to its original size.

While it is true that the Internet is a communications network, it is also more than just that. It is also true that the telegraph system through which the real world Mundaneum would have done its work is a communications network. The real world Mundaneum is simply the "cloud storage" part of the fictional world Internet. While the relays and wires and people sending the telegrammes are the "communications network" part of the fictional world Internet.

# What is the Internet, exactly?

What makes the Internet different than television? Or radio? Or the printing press, or the mail? The Internet is:

1. A way to transmit information
2. That allows any node in the network to send messages to any other
3. In real time
4. And the contents of said messages can contain arbitrary data

Television is an audiovisual feed only; radio is audio only. Both are single-source many-receiver. Telephones let anyone talk to anyone; but that specifically means only talking. The mail isn't real time.

The Internet can do the job all these things did, and more. That's why the Internet is replacing them all. Already many "landline telephones" behind the scenes are transmitting calls over the internet.

# But there's another layer built on top

The internet technically existed in the late 90's. It was nothing like what we have today. But what we have today is nothing more than a few extras layered on top of "the internet":

2. Organizations began using this technology as an interface to build and design ever more complex services.

You can think of any given website as a really big encyclopedia. A URL is just a page number. When you go to the site, you send the site's bank of servers a request. The site's computers go find the document for you, and send it back.

Nowadays, of course, the documents are actually generated on the fly; when you go to hotels.com and search in the search box, their computers receive the request, decipher it, look up the relevant info in their databases, then craft a reply. Google doesn't precompute search results for every search into an enormous encyclopedia; it has a queryable database, it runs your search, it spits out an answer.

# What Paul Otlet invented

Paul Otlet invented Wikipedia in 1910. He called it the Mundaneum. He couldn't actually build it, of course, but oh how he tried. For decades, he tried. But like Charles Babbage, his idea was simply so far ahead of its time he just couldn't make it happen.

He anticipated having a network of "electric telescopes" (computers) that could all talk to each other. That's basically the Internet. He anticipated using links to connect the documents into a giant interconnected web.

# What could be done without electronic computers

Charles Babbage already had a design for the Difference Engine. His technology doesn't use transistors. It isn't even electric. We know that it works, if only you can build one, because in 1991 someone did. The Computer History Museum in Mountain View has another it uses for demonstrations.

Charles Babbage designed what is, for all intents and purposes, a machine computer. He just was never able to get the money or resources to build such a ludicrously intricate piece of machinery. But his Analytical Engine was indeed a bona fide computer.

In your world, transistors simply don't work. Analytical Engines will. They'll never be anything like as powerful or as cheap to operate as electrical computers were even in the 80s.

# Hook Babbage and Otlet up with deep pockets

If Babbage had successfully been able to make the Analytical Engine a reality, then by the time Otlet comes into the picture, what we would call a modem might have existed: Imagine duct taping a telegraph to an Analytical Engine.

In which case, Otlet's grandiose vision can actually be realized. The Mundaneum comes to be a thing. In order to use it, you have to be at a place that has a connection to a telegraph line and an analytical engine. But that's still a pretty big change; libraries and other place (like very up-scale hotels) will want one.

In turn, others will eventually start offering their own special services that you can '[tele]graph into'. Said hotels will already have most of the equipment, so it's just a matter of hiring an Analytical Engineer to make the machine automatically respond to telegrams inquiring about reserving a room, or checking on the status of an existing reservation. Banks will find having access to an information sharing network very lucrative, so they'll want in as well.

• Fascinating. I had no idea. Analog computers aren't good at memory I believe. How would that work? – chasly - supports Monica Aug 19 '20 at 19:28
• Please explain what the Mundaneum was and its limitations. I don't believe "the Internet" could have been built using what it was intended to be. – JBH Aug 19 '20 at 20:02
• If Ton has brought all of that and maybe a bit more into his/her answer, I wouldn't have downvoted it. – JBH Aug 19 '20 at 21:50
• @AlexP in a now-deleted comment to this answer I pointed out that the mundaneum was more analogous to a web browser than the Internet. Still, it's an interesting piece of history that contributes to Chasly's post. – JBH Aug 19 '20 at 22:03
• Hi, I did kind of gloss over a bit much - more than I thought - cause I had to get to work. I'll elaborate more when I get a chance but basically - the internet is a standardized way to send messages, and most consider "the internet" synonymous with "the web", which is very similar to the Mundaneum. – Ton Day Aug 19 '20 at 22:56

The Internet was imagined in the 1940s as Vannivar Bush's "memex"

Vannevar Bush was the head of the US Office of Scientific Research & Development during the war, and he wrote a famous essay called "As We May Think" that has been cited as a major inspiration for the developers of digital computers and the Internet. The essay addresses the problem of the growing body of knowledge, so large that no one scientist can get a complete view of even his own field, and the inadequacy of library filing systems for helping people find relevant knowledge. The machine he envisions called a memex makes use of analog technologies then emerging, to allow people to create links (he calls them "trails") between objects and to share the data as well as the links. Although in our world those ideas led to a digital Internet, if you want to create a fictional analog Internet, the suggestions in his article should be adequate enough for you to depict them in your world.

It can be done: with a serious The Wild Wild West steampunk flavor

Please note that "what if some critical branch of technology wasn't invented?" questions are always problematic because everything needed for that branch will still exist — so why wasn't it invented?

Anyway...

Given enough electricity and big enough buildings the entire Internet (but not its speed) can be implemented using telegraph wire and relays, which is all a switching transistor is. A "bit" of memory is simply a specialized switching transistor. I can even create static memory by creating solenoid-driven cams that would "lock" a value into place until the solenoid (or its resetting solenoid) was activated. Thus, if the power went down, the "memory" would remain.

Do we have analogs of this behavior?

Oddly, we do: electromechanical pinball machines, where the "state" of the machine must be known and remembered during the process of play. The only thing missing is the state-locking mechanism that would let you turn the machine off, then back on, to continue the same game. (Why do I know this? Because I own a 1973 Williams Triple Play machine, which is an awful lot of fun to play — and just as much fun to maintain.)

So, telegraph wires connect two cities — three wires, reference/ground, data from and data to (sounds like an old Modem, doesn't it? we want that third wire to allow simultaneous bidirectional transmission). On each end are the listening and sending apparatus of a telegraph. The difference is that they're connected to solenoids that allow an analogue computer to do something with the transmissions.

Memory is achieved with massive banks of what I'll call latch-lock relay assemblies that mimic SR Flip Flops. One solenoid affects a SET, the other a RESET. The mechanical latch allows the memory to be maintained indefinitely (well, let's assume no earthquakes) even without electricity. They're activated using diode AND gates (1900's tech, right?).1 Thus, when a single column and a single row are line-active, only the one relay at the cross point is SET or RESET.

I did mention this would be painfully slow, right? I didn't? OK, it's going to be PAINFULLY SLOW.

The semblance of a digital computer can be created electromechanically with relays and diode logic.2

Frankly, the real problem is the human interface. So I would expect everything to be done using punch cards. (However, since typewriters have been around since 1868, you could develop an output device that typed. I believe 1900s tech could come up with large rolls of paper, in effect, the Telex machine.)

Conclusion: Yes, the entire Internet could be conceptually implemented using 1900s technology. Even IP addressing could be done. It would be power hungry, slow as molasses, LOUD! REALLY REALLY LOUD! and would have an incredibly low MTBF3 (mean time between failure). People would be constantly monitoring and replacing relays and other components. But, it could be done.

Heck, you could even have a 1900s version of the Dark Web where people illegally tap those telegraph wires and hijack IP numbers to set up physical-location-unknown data processing nodes. (Although the idea of hauling all that equipment, including power generation, around in horse-drawn wagons is a bit daunting.)

If you're interested, author Terry Pratchett implemented something like this called the "Clacks," based on semaphore tech. He even conceptualized a "ghost in the code." I'm a fan.

1If you insist on no semiconductor devices at all (you only said no transistors, so I'm cheating a bit), then the diode logic for memory could be replaced by a three-state voltage system where columns would transition between mid-volt and logic-high-volt and the rows would transition from mid-volt to logic-low-volt, the result is that one and only one relay could actually activate (having the correct potential across its terminals). Nothing else would because the voltage difference wouldn't be enough to fully energize any other solenoid. However, the power loss would, itself, power a small city.

2Again, if you insist on no semconductors at all, I can implement full logic using only relays. It's just slower and more power intensive.

3As I've thought about it, I believe you could even set up a rudimentary form of JTAG Boundary Scan that could be used to lower the repair time.

• Regarding state locking, you really only need a single relay - though controlling it becomes more difficult. Use something easily magnetized; magnetize it to hold the relay in one position, then demagnetize to allow the relay to move back to resting position. The solenoids used in battery- and solar-powered irrigation valves operate this way. One example is opensprinkler.com/gardena-1251-using-osbee ; rayshobby.net also discusses it. – user51614 Aug 20 '20 at 19:05
• @user51614, that only works if you have power 100% of the time. The OP is expecting early 1900s technology. That's a very tall order. – JBH Aug 21 '20 at 2:25
• that only works if you have power 100% of the time huh? These need power only when changing state. They use that power to magnetize or demagnetize something, and the magnetism or lack thereof maintains the state indefinitely without power. Solar-powered devices do not have power 100% of the time, and battery-powered devices would rapidly die if they had to keep a coil energized. If in doubt, look at the opensprinkler link - there are oscilloscope traces showing the power pulses. – user51614 Aug 21 '20 at 4:20
• Latching relays exist already. Some have two coils: one to latch on, one to latch off. Some have one coil and the direction of current flow makes them latch on or off (and when there's no current they don't move). – user253751 Aug 21 '20 at 10:09
• @user253751, This is a traditional latching relay. I know you can do the same function by complicating the mechanics and using both the forward and reverse solenoid actions, but the power complications of the early 1900s are bad enough without having to figure out how to reverse the voltage potentials. That's why I ignored that path. – JBH Aug 21 '20 at 13:12

The Internet does not require computers at all. Most of the protocols are simple enough to perform by hand, and even the protocols that do require some processing power are only designed that way because we have computers.

In a world without computers, those protocols could be designed differently, to only require calculations that an average operator can do in their head.

There is a (joke) April Fool's RFC that specifies IPv4-over-pigeon: RFC 1149 – A Standard for the Transmission of IP Datagrams on Avian Carriers. It was actually implemented and successfully tested by the Bergen Linux User Group in Bergen, Norway.

Pretty closely, except for speed.

The core components of ‘the Internet’ are mostly conceptual. DNS lookups, communication protocols, common data exchange formats, resilient pathing algorithms: all of these are possible with analogue technology. A DNS lookup is just a complicated automated exchange, or can be run by humans. They existed without digital computers. Resilient pathing is basically a series of linked exchanges. All data transfer can be done with analogue audio signals and a known (standard) format. Output/input is a matter for the end user. If they use a pen drawing the waveform on paper or a person listening to the bios and transcribing: it’s the same as long as the protocols are being properly followed.

Where this falls over is speed and bandwidth. Even the best analogue exchanges take time to perform routing and need physical space and have a limited number of possible transmission channels per wire (barring optical transmission technology like fibre-optics) to enable the connections. This limits the number of concurrent connections down any given path between two nodes. Any protocol that requires handshakes or error correction packets will be limited by the speed of both parties and every node in the communication pathway between the two. Without considerable research (which we’ve never had to do because digital computers are very good at this kind of thing), you will be severely limited by available connection lines.

Pull messages vs. push messages.

With the http protocol, the client computer queries a server, possibly half a globe away, for information. This query is relayed from a local server to that distant machine. The query might include information that is incorporated into the anwer.

The earlier usenet system would push newsgroups from server to server. You dial up to the nearest one and look what they have. So imagine that the local telegraph office allows people to browse incoming "general interest" telegrams, and to send telegraphs "to whom it may concern." The telegraph office also forwards such messages to the next office down the line.

To that add mailing lists. You send a telegram to subscribe, and then you get any telegrams by others to that address.

• So magazines... – user253751 Aug 21 '20 at 10:10

There are no digital computers whatsoever.

Oh yes there are. You aren't the first person to make the mistake of thinking that "digital equals transistor", and no doubt you won't be the last. But all the others were flat wrong, you're flat wrong too, and your world will have digital computers, because digital computers were in widespread use across the world decades before the invention of the transistor. In historical fact, if this is supposed to be our 1940s then they didn't even have transistors then.

You've forgotten about relays, switches, motors, punch cards, punched tape, and all the many mechanical devices which are either on or off and embody digital state and processing. Relays are actually highly reliable to hundreds of thousands of operations. Motors and switches can be similarly reliable. In 1940 these mechanical digital processors were in the process of being superseded by valves, but mechanical and valve processing were both very much alive. The Enigma and Lorenz cyphers were broken by digital computers using these.

For your internet, I would even submit the Lorenz as being an example of what you could do in that direction. Before Lorenz, the operator typed messages into an Enigma machine, wrote down the encoded message, and sent it themselves using Morse. Another operator then reversed the process. With Lorenz though, the machines communicated with each other directly. The operator typed in the message, the machine automatically encrypted it, and the machine broadcast the encrypted message using the Telex product. On the receiving end, the other machine received the Telex, automatically decrypted it, and printed out the decrypted message. If you're reading this and thinking "mechanical SSH session", have a cookie. :)

I could give you dozens of examples of pre-transistor digital computers, but you can do the same if you JFGI. I don't mean to be overly harsh, but this is a bad question because you have even the most basic general knowledge of the history or technology. Not knowing is fine, of course, but I would expect some trivial googling before asking your question, and even a cursory search and a minimum of thought would reveal that the concept is fundamentally broken. Your question even explicitly says that valves exist, for goodness sake!

Then there's comms. I already mentioned this was a Telex. People had already invented digital communications. The phone network was already there. The phone network was still using operators, but the early automated exchanges were purely mechanical. And the phone system was interconnected around the world.

Frame challenge then. Your basic concept of "no digital computers at that time" is wrong. Your basic concept of "no digital communication at that time" is wrong. Your basic concept of "no global communications network" is wrong. All your prerequisites are there, exactly as we had. So you're asking us to answer a question which can simply be answered by "look at our world", and that has no place in worldbuilding.

• Thanks. Don't forget that I used the tag alternate-history. I also used the phrase "alternative Earth" in the opening paragraph. This gives me a certain license to change things. For example in this world Charles Babbage was famous for designing a widely-used drainage system and Alan Turing didn't have a war to make him interested in computers. He was a moderately successful lecturer in mathematics who died of old age. – chasly - supports Monica Aug 20 '20 at 21:14
• I just brought up Babbage's work in my elaboration on the Mundaneum answer. – Ton Day Aug 21 '20 at 3:50
• @chasly-reinstateMonica Still doesn't work. There were digital computers well before Turing too. You would literally have to eliminate the concept of switches from humanity to achieve this. – Graham Aug 21 '20 at 9:08

Radio technology (still valve based) is however all the rage. Telephones work by having human operators who connect one caller to another by manually plugging them in at the exchange.

We're looking at the wrong technology here. What we should be considering is the telegraph for storing/forwarding and retrieving messages.

We have most of the fundamentals in place.

I'd suspect something like a cross between teletext, where fixed information was sent to terminals (as opposed to interactive terminals). We could have a constant stream of data, encoded on punched tape that's sent. Its probably possible to encode something like a destination, and have a fixed set of routes (maybe updated over telegraph).

Lets say our "routing" table is London -> Bombay -> Madras -> Singapore (practically there may be more hops).

Each site would have a routing table to the next hop to a location - London would know that if they send it to Bombay, Bombay would see its a message for singapore, look up the next table and know it needs to go to Madras and send it on.

As for mechanics - have a standard header, maybe a checksum and end (yes, its overhead) and punch from tape, with something like a code for the destination (SIN01 for singapore for example), write it to a punch card. Throw it into the appropriate bin for the next hop, punchcards are loaded into the machine that sends it to the next hop (so as long as the punchcard bin is full, we keep transmitting). Each punchcard essentially is a packet, and by having 'static' routes we might reduce some of the complexity of routing in a system with relatively few nodes.

This is essentially poor man's UDP over telegraph, and probably entirely doable mechanically. Since the lines are "fixed" and constantly transmitting, its a matter of picking up the message and getting the last mile sorted.

I'll leave ensuring delivery and such as an exercise to the reader, but might suggest it could be fitted into 'spare' space in a "packet"

as an aside Bombay is now Mumbai and Madras Chennai, but I've chosen to use the period appropriate names

• This sounds good. I'll read it again when I have more time. – chasly - supports Monica Aug 22 '20 at 12:14

The Pantelegraph was the predecessor of the fax machine, invented in 1860s. It was followed by the Bildtelegraph in 1880.

It would be possible to Telex a request for information to a library, where they'd look up the information, and either Telex or Pantelegraph the answer back.

Memory and speed are the problems

As others have mentioned, you could technically build an 'internet' without technology at all. Think Great Library of Alexandria, with some system to request documents and add documents to the library remotely.

Let's say there were millions of 'runners' whose job it was to be the transport layer of the network. Want to add a document? Give it to a runner, who takes it to the library and archives it. Want to search for a document? Give your search parameters to the runner, who goes to the library, carries out the laborious search, hand-copies the results and returns them to you - in a few years maybe. Photos? Runnees are trained artists that can draw anything you aant drawn and add it to the library.

The problem is that for the internet to work, it has to do these things relatively fast, with high bandwidth, and it needs a LOT of storage.

We had an 'internet' of sorts in the 1980's with FidoNet, CompuServe, and other competing technologies. But dial-up modems and the cost of memory even then limited its usefulness, as did its fractured nature. If only a small portion of documents you are searching for are even available on the internet, its usefulness declines dramatically.

So if you want a usable internet, you need widespread adoption. That means the amount of data that must be stored is massive.

To give an example of early non-solid state memory, look at the Selectron Tube. This is probably the highest form of non-transister memory tech we had. RCA's version could hold 256 bits of data,and cost (in today's money) about 5300 dollars. If you wanted to store a typical MP3 (say, 3 minutes at 128kbps) in selectron tubes in 1948, You'd need about 3 megabytes, or roughly 45,000 Selectron tubes. So your one-song internet page would require roughly $200 million dollars in storage, and those tubes, their power supplies and air conditioning would fill a building. There was a 4kbit Selectron tube that never saw commercial use. If we assumed those were available at the same cost, we could reduce the cost for your one-song storage to 10-20 million. Before transistors and integrated circuits, there was simply no way to store enough data in a way that could be accessed quickly enough to make a usable internet. Then there's hard drives. The first IBM hard drive in 1956 held 3.75 megabytes, and cost$34,000. That's roughly the size required to hold a single MP3 file. But of course, you'd need lots of solid state circuitry to operate one.

So, the bottom line is that, while you can build an internet out of anything, to build one that has the features we have today (high bandwidth connections, petabytes of data quickly retrievable) in a usable form that would provide real value to the masses, was utterly impossible without a whole range of technologies that were not developed until fairly recently. Not just memory, but things like spread spectrum radio, fiber optics, etc. Those in turn require modern electronics to make and operate them.

• Thanks - very useful and informative. – chasly - supports Monica Aug 23 '20 at 0:26
• MP3 files aren't the real thing to compare. If you want a pre-transistor internet serving up audio, you give everyone a dual connection - telegraph for data (with relays & vacuum tubes doing all the work) and telephone for audio. Hook up a jukebox and poof, instant music-on-demand. – manassehkatz-Moving 2 Codidact Aug 23 '20 at 15:27

An internet could be made, but it would have to be about using animals like pigeons and dogs and even human beings for physical data transfer called a sneakernet: an "internet" where information is transferred with magnetic tape, floppy disks, optical discs, USB flash drives or external hard drives. Magnetic tape, that first item, was first discovered in 1928 as audio tape and was used in World War II. This alternative 1940s internet would be using animals and items to move cartridges of tapes between people and then use certain audio devices/mechanical computers to interpret this information from tapes.Even a system of capsule pipelines can be used to deliver tapes. Someone can even send data over the radio using certain methods and even use recording devices to 'write down' radio signals containing pertinent information onto audio tape. This system allows for radios to be used for wireless transfer of data in the form of audio recordings, these recordings can be written on magnetic tapes as 'information', and this information (music, Morse code for a message to a loved one, code for an equation, etc.) can be interpreted by computers/audio devices while be transported by people/animals.