Would creating a custom human language be stronger protection than encrypting a known human language?

Question

There are some languages that people haven't been able to read (Cretan Hieroglyphics for example).

Does that mean that you can better protect data by making your own language than by encrypting an existing one?

Or are all human languages really just different mappings of symbols to the same or similar ideas meaning that decryption and translation are the same thing?

Answer 1

$$\texttt{NO}$$

In short;

$$\text{We are living in a world where that is ruled by Kerckhoffs's principles}$$

The above may not be totally clear as indicated by Maarten in the comments. The We is the Modern Cryptographers, the World is used for the Cryptography,... In other words, and better;

$$\text{Modern Cryptography keeps to Kerckhoff's principles}$$

---

Decoding lost languages is not related to cryptography, though some techniques may be related like frequency analyzing. That is people tend to give names to some common objects like the sun, earth, man, women, child, etc.

Does that mean that you can better protect data by making your own language than by encrypting an existing one?

Go on build one, then;

• On day one, one will sell the decoding. Then you need to change your language and then you have to teach it, everybody. In Kerckhoffs's principle, the system relies on the secrecy of the key, not the system. If the key is sold or compromised, we can change the key, not the system. Of course, we assuming that the encryption system is designed to stay against cryptography attacks like AES that stands for almost 20 years and expected to stand more.

• The more you talk the more information you will give to decode. You can get the notion from the researcher who goes into amazon forest to communicate with people who had no contact with other humans in centuries.

• Actually, you build a large permutation cipher, that is breakable.

Or are all human languages really just different mappings of symbols to the same or similar ideas meaning that decryption and translation are the same things?

That is not exactly. For example; some languages don't have a name for the color pink or other. Some missing other words for describing things/actions etc where it exists in some languages. Culture, living area affects the language. This is out of the Cryptography context.

Would creating a custom human language be stronger protection than encrypting a known human language?

NO

---

Let finish with the Kerckhoffs's principles

1. The system must be practically, if not mathematically, indecipherable;
2. It should not require secrecy, and it should not be a problem if it falls into enemy hands;
3. It must be possible to communicate and remember the key without using written notes, and correspondents must be able to change or modify it at will;
4. It must be applicable to telegraph communications;
5. It must be portable, and should not require several persons to handle or operate;
6. Lastly, given the circumstances in which it is to be used, the system must be easy to use and should not be stressful to use or require its users to know and comply with a long list of rules.

Answer 2

In addition to what Kelalaka said - there are also a few notable problems with your idea:

1. Imagine that you have a friend that is willing to translate for you 2 messages of your choice from your language, to that secret language. Now, if the translation is deterministic, then next time you see one of the two translations, you know the "decryption" to your language. An encryption scheme (a good one) that allows you to encrypt, but only allows the key holder to decrypt (a public key encryption scheme) doesn't have this weakness (read about chosen plaintext security (CPA)).
2. Similarly, imagine that you can give to your friend a list of sentences to decrypt and he will decrypt them to your language for you. Now your friend goes on a vacation to an island without phone reception, and you suddenly see some "encryption" in the secret language. Since you have some decryptions of your choice from the past, this might help you decrypt the new encryption. However, some encryption schemes are chosen ciphertext secure (CCA) and you don't get any advantage for decrypting a new message based on past decryptions of messages.
3. If you somehow saw an encryption of a message in the past, and now you see it again - you know that the same message was sent just now. However, a good encryption scheme would not allow this kind of thing.
4. An encryption scheme should have deterministic and 1-to-1 decryption, so a ciphertext should not be able to be decrypted to multiple plaintexts. In languages it's usually not the case...

Answer 3

There are already two excellent answers from the approach of formal threat assessment, but I think there are a few more key ideas about usability worth raising.

Number 1, your cipher must be able to encrypt your message. If the message is English text, you're probably ok, but this can't work with most data we encrypt in practice. The famed nerdy comic xkcd makes that point here.

Number 2 is how you and your confidant learn the code. It is simply a huge amount of work to learn a language to fluency, especially in secret.

Number 3, your cipher should resist picking apart in pieces. There's an observation quipped by language students that French homework is translation of "Where is the beach?" and Latin homework is "My gladiator is mighty with a trident." The key observation is that although your whole dictionary might take an attacker a long time to work out, if they only care about messages on a certain topic then they don't even need to bother.