# What would a society without cryptography look like?

In this world you have access to computers powerful enough to decipher messages as long as the method used to encrypt it has at least one theoretical weakness (those computers cannot defeat an impregnable encryption method).

In other words: if it's feasible then the computers can solve it in less than one second.

Therefore, you can't easily transmit ciphered information (keys, passwords, etc.) and you can't hide anything but with physical vaults.

• What would this society look like?
• (Edit to narrow the subject) How could those people still transmit secured messages? Is it possible?

I asked a related question in Physics. See Could quantum computers break any code for more information.

Edit: I removed the part which stipulate my world could have a law forbidding people from ciphering information here.

• To not have crypto you'd not need to pass on written secrets in the first place – ratchet freak Jul 16 '15 at 12:56
• i don't think that setup is feasible. I think it should always be possible to create a cryptographic setup that for the available computatioonal power is not decipherable in a reasonable time span. And unless your computational power increases by an order of magnitude or two almost instantly, the two will co-evolve. – Burki Jul 16 '15 at 12:59
• Repeat with me: passwords (or "shared secrets") are useful for some forms of cryptography, and for authentication. Without cryptography, passwords are still useful for authentication (but their transmission and storage would become more complicated) – SJuan76 Jul 16 '15 at 13:56
• I fixed my mistake but I find your "repeat with me" rude. – Ephasme Jul 16 '15 at 14:15
• It's always helpful to link to other questions (both one's own and others'), on the site or on different sites, if it helps provide context for what one is asking. In this case it might help by providing context to the statement "access to computers powerful enough to solve any cryptographic problem in no time". If in doubt, it's often better to link but also make the context of that link clear. It's easier for you (as the OP) to edit a link out if it turns out to not be helpful, than to expect everyone else to run around the network looking for other possibly-relevant questions. – a CVn Jul 16 '15 at 15:25

If there were no (known) programs to do hard encryption with a small internal state and/or amount of computation ( modern cyphers like DES, AES, Twofish, implement a well-shuffled permutation, without needing to keep a full list of the order of 2^n (n=58, 128, 256,...)) then one-time pads could be useful. Another thing is a stream cypher, and if there were none (known) that were usefully strong, a large bank of key material could be still be stretched out more instead of using each value once, via a stream cypher concept. But instead of one key used for an arbitrary run of the stream, one hard key could generate a small number of stream bytes.

But... how could there not be? They might not be as good and refined as we have now, but pseudorandom generators can shuffle a list, and make for a somewhat reasonable stream cypher. I remember in the 80's using a pseudorandom number generator (written up in DDJ) that used a table of 256 values, and each iteration changed one of them as well as returning a result value.

If poor-quality cyphers ran on consumer computers could be broken by big computers,

• it would not be known to common users whether a particular novel encryption system was any good at all, so they would worry or assume that they were not good.
• use of more hard key material and more time can make stronger cyphers.
• A proper study would identify how much it takes to be secure against breakage for what length of time, and good scholerly work could identify algorithms that were sound.
• but, the same scholerly work could lead to discovery of the missing cryptography knowledge of block cyphers and strwam cyphers that are many orders of magnitude harder to break than to employ.

It is more reasonable to suppose that asymmetric encryption was not known (after all, it was only publicised in the 80's) or if known in principle was not fully worked out or not trusted by the users.

You might end up with something a little like a toned down version of Times Eye (https://en.wikipedia.org/wiki/Time%27s_Eye_%28novel%29) where there could be no secrets (even inside vaults).

Problem is, being able to exchange secrets is just too damned useful.

The 2 approaches you outline"solve any cryptographic problem" and "illegal to cipher information" have 2 very different effects. The first would cause a switch to 1 time pads (which can't be cracked by any computer) while the second would lead to crypto still being used in any venture where the punishment is worse than the punishment for using crypto.

Steganography is going to be a huge thing in this world and depending on how you implement your anti-crypto measures crypto might still be a thing.

If you can hide a message inside a hollowed out walking stick or encode it in the least significant digit of the pixels of a photograph then you can bet that travelers with their family photos will still carry secret messages.After all, these days one microSD card hidden inside a coin can hold a scan of every document from the Manhattan project.

• +1 for steganography. It would be cool to see that idea fleshed out. – Brendan Adkins Aug 24 '17 at 21:54

It means a lot more things would be done in code. Encryption is making something unreadable by a specific process and anyone with the right 'key' can change it back, this means if someone can find or make a key they can understand the message.

Codes however, have no direct 'mechanical' conversion. Take the Navajo code talkers during WWII. No one new their language outside of the US so they could speak to each other without an enemy having a clue what they were saying. On top of that they spoke cryptically in code so even if some German had spent time in the American southwest learning the language, they still wouldn't know what was actually meant.

Code however, is much more difficult to implement on a computer, since it is generally based on shared knowledge between two or more individuals. Kind of like Cockney Rhyming slang.

• :s/German/Japanese/ – AlexP Mar 18 at 22:53

I think in reality a society without cryptography would have to look one of two ways.

1. Access to technology is tightly controlled. The average citizen does not have access to it for private use and likely doesn't know what it is or understand how it works. You would almost have to have a Stargate (the movie) type scenario where a portion of the population is super technologically advanced and they use that tech to keep the remainder in line acting like it is magic.

2. The second alternative I can think of is a world where people simply do not or cannot steal. As impossible as it sounds if things go missing or are stolen to the detriment of the owner then security will be needed.

Obviously in either of these settings an open network would not be a smart place to store any important data.

Other than increasing the cost of encrypted traffic to the point that it would be used more selectively, there is no reason to expect a change.

It is always possible to generate one-time encryption keys physically (monitoring radioactive decay, for instance, or my all-time favorite, lava lamps https://en.wikipedia.org/wiki/Lavarand). At this time, so far as I know, such keys are not amenable to being broken, even in principle, by the application of computing power, and going to quantum computing won't help.

Once this occurs, the "only" problem is secure key management and distribution. The gains in convenience and security associated with private/public encryption algorithms would be lost, and the replacement costs would be considerable, and would require an entirely different security model, but in principle these don't seem insuperable - just expensive, with all the effects which "expensive" imply.

I think part of this would be separating digital/computer cryptography from cryptography-at-large. Things were encrypted long before modern technology, and some of those forms of encryption are still difficult to break. For example, assuming it's not just a medieval prank, the Voynich Manuscript has gone undecoded for over 100 years since it first really came to light.

As with anything digital, computer cryptography follows very rigid rules. Not doing so would make it not terribly usable. Computers can only follow programmed rules, and encrypting information using techniques that don't follow a specific logic can be done.

Take, for instance, a manufactured language based on English, where words are spoken with some form of prefix, suffix, or inserted pattern, but nothing specific - if'n Iyiyiyi ajusted starterin talkspeakingly as'if this - many English speakers would understand more or less what I was saying (If I just started speaking like this). A computer would have a much more difficult time deciphering it. It might start by doing speech-to-text (assuming it was spoken out loud), then trying to match words by removing letters, then trying to match words into pattern groups. In the end, given a very large sample, the most powerful computers would still have difficulty deciphering this, whereas most people would simply understand it.

The above isn't a super strong example, but it shows that how humans "decode" things and how computers do so is vastly different. If you are able to come up with some similar idea, something organic and not rigid, that would be more difficult for other humans to decipher unless they knew something about it, then you could plainly speak to another human in a crowd and not be understood... cryptography (of a sort).

Honestly, I don't see any reasons for quantum computers to eliminate cryptography. Quite the contrary, they bring new possibilities of quantum cryptography.

Comparing with "regular" computers, their advantage is the processing power - quantum computers are really fast. But that doesn't break any core principle of cryptography, since their computing speed still has it's limits.

## P ≠ NP

The P versus NP problem asks whether every problem whose solution can be quickly verified by a computer can also be quickly solved by a computer. Asymmetrical cryptography is based on assumption that P ≠ NP. That basically means that you need much faster computer in order to successfully hack a cipher text. But in a world of fast quantum computers you use another fast quantum computer to encrypt your data. If it takes long enough time (it will, providing the key is long enough), decryption still will be time-consuming.

Even when you don't use a quantum computer for encryption, decryption still could be complicated. There are post quantum cryptography algorithms being developed in order to resist the brute-force approach to decryption.

You have provided with a preformed structure for your world: You can't easily transmit ciphered information (keys, passwords, etc.) and you can't hide anything but with physical vaults.

Instead of arguing whether or not this is practically possible (with current technology), let us concentrate on your questions.

# What would this society look like?

Almost the same as it looks like, now. Only two-key (public and private) type encryption systems would not be possible. The passwords on emails and user accounts would still be as effective as they are now.

Simply, because hacking a password is an extremely long and boring process known as "brute forcing", where the attacker tries every possible word combination possible for a given length of characters. This is the only type of hacking possible when you don't have a direct attack on the information possible (keylogging, server database hack etc).

Yes, encrypted connections between server and client side computers might not be possible in such a world and hence banks would have limited services regarding online transactions and money transfer.

Militaries all over the world would have a very high reliance on codebooks than on digital encryption systems.

# How could those people still transmit secured messages? Is it possible?

Yes it would be possible. People would send encrypted messages to each other without sending the key to decrypt those messages. This is how the system would work:

Person A sends person B an encrypted message Person B applies another layer or encryption on the message and sends it back to person A. Person A removes his layer of encryption and sends the message to person B. Person B removes his layer of encryption and reads the message.

At no point in this protocol does a transmission of keys occur. Yes, this system might be slightly slower in working than the protocols we have in place these days, but it would still work flawlessly for digital communications.

I would suggest that such a world is mathematically impossible, assuming that mathematics works the same in all universes. The mathematical theory behind cryptography relies on computational complexity, which conveniently doesn't depend greatly on the physical hardware.

It's possible that the physical computer in that universe is non-deterministic or something close to it, making RSA very easy to break. (This is actually an issue with quantum computers) But there's no evidence to suggest that other, more advanced, alternatives wouldn't be possible. In fact, some very smart people in the NSA (and some other organizations) are already working on encryption methods that can't be thwarted by quantum computers.

Basically, there will always be something better. Where there's a need, there's going to be innovation and this encryption-free society won't be unencrypted forever.

The only way this could actually work is if math actually worked differently in that universe to the point where all forms of asymmetric encryption were mathematically impossible. It's hard to imagine such a universe, but let's entertain that for just a moment.

Instead of focusing on encryption for security needs, there would be a massive emphasis on physically verifying identity. This is incredibly difficult to do with cryptography, let alone without it.

There would be a lot of undetectable man-in-the-middle attacks, a lot of identity spoofing, and tons of fraud. Passwords would be useless (unless hashing is still possible) and robocalling would be rampant beyond what is already the case. Things would be so bad that the internet would actually be a bigger liability than a benefit and few people would use it since it's essentially the technological equivalent of walking unarmed into a back alley in a crime-ridden ghetto.