[lifthrasiir]

> Have you got an experience in data encryption for databases or related ?

Yes, I worked on online game servers, encrypted protocols and of course encrypted databases (required by the law). There were no keys stored in the database nor in our code, we had a dedicated encryption server that holds them. (Nowadays we would use AWS KMS or similar services.) Our keys were far longer than what is brute-forcable as well. In fact if something is brute forcable your response should be increasing key length (at the very least 128 bits or more), not changing algorithms to some obscure home-made one.

[kronxe]

I see and I understand your point but I want to say the cllasical approaches are more stable but more unsafe. By the way, there is no something that non brute-forcable today, increasing proccesing power and quantum computing allows it. I think we look some different specific fields. 'Home-made' algorithms certainly gives much more safety because people cannot attack if they dont know how it works. For example, in my algorithm I used ideas of prime numbers and also changing the order of the characters but maybe some other people use turn bytes into bcd values and than make some change for complexity. My algorithm is one of the example of trying to use some other method different than standard encryption with key.

[lifthrasiir]

> By the way, there is no something that non brute-forcable today, increasing proccesing power and quantum computing allows it.

See, you don't know anything about cryptography. Quantum computing (which is currently very experimental) aside, 2^128 is not something you can brute force today and even in a near future. Let's see why.

Bitcoin is something very close to the most performant globally distributed computing system, and its hash rate is about 3 x 10^20 SHA-256 hashes per second (since Bitcoin PoW uses double SHA-256, a commonly cited hash rate is a half the actual hash rate). Therefore we can reasonably assume that we can do the order of 10^22 decryptions per second today. Note that this hash rate is increasing, but now in a roughly linear rate (currently about 10^20 hashes per second per year) so this assumption should be not too off. Given 2^128 / 10^22 = 3.4 x 10^16 seconds = 10^9 years, it is clearly not brute-forcable today.

Quantum computing is also not a magical sauce. QC poses a problem to the cryptography mainly because some cryptographic algorithms relied on currently hard problems like integer factorization and they can be efficiently solved by quantum computers. Post-quantum cryptography (PQC) thus seeks for alternative problems that would be still hard for quantum computers. More importantly though, symmetric encryption does not make use of such hard problems, so the potential speedup is only possible with Grover's algorithm that searches N records in sqrt(N) time. Quantum computing thus does make 128-bit keys unsafe (since it will only take the order of 2^64 operations to brute force). But by then we can simply double the key length to restore the difficulty.

> For example, in my algorithm I used ideas of prime numbers and also changing the order of the characters but maybe some other people use turn bytes into bcd values and than make some change for complexity.

Your algorithm, as I can see, is equivalent to a Vigenère cipher with implicit character mapping generated from prime numbers. The practical cryptanalysis of Vigenère cipher (Kasiski examination) appeared in mid-19th century. Using prime numbers doesn't make your algorithm automatically safe.

[kronxe]

I dont want to discuss this issue too much but maybe you can check that what is the ongoing problems about bitcoin's p=np problem, is 2140 a realistic target and is your calculation still applicable for today.

Also, I recommend you to read Google's quantum paper and quantum attacks on some sort of cryptography things.

If we think about today, we cannot catch up new techs, we should think about future. That is why technology is growing exponential.

[lifthrasiir]

Yes I haven't claimed that it will last forever. But you have claimed "there is no something that non brute-forcable today" haven't you? 2140 is not "today" nor a near future. In fact it is more likely that a flaw is found in existing algorithms before 2140. Do you make your system last that long? Then you would want to avoid algorithms already broken in 19th century.

[kronxe]

I think maybe I cannot explain clearly, I dont have a claim like this my encryption algorithm is safest, fastest, best algorithm on the earth. I tried to make an 'unique' algorithm (as you call home made :) ) in an hour. I shared it in order to show there should be different approaches to encryption algorithm because if algorithm working principles are known and predictable, then algorithm safety level decrease. I think you tried to approach cryptography by an academic perspective (more stable solutions) but I tried to avoid stoling of the data.

By the way, I mentioned N=NP due to show probably in a few years there is solution of this problem and maybe the end of the bitcoin [1].

As your claim about brute-forcability, I want to say quantum computing is experimental right now but there is working machine and proccesing units of them. In other words, it is not only theoretical and Quantum computing age probably allows to crack these all algorithms probably mine as well with Quantum AI. However, in my opinion, if people do not know how your algorithms work, the attacking possibility is dramatically decreases.It is more difficult to target something unknown.

[1] https://gizmodo.com/if-you-solve-this-math-problem-you-could...

[lifthrasiir]

> [...] maybe the end of the bitcoin.

I have only used Bitcoin as a reasonable limit of computation power we can readily make use of. It is completely irrelevant whether Bitcoin is feasible in a long run or not.

> However, in my opinion, if people do not know how your algorithms work, the attacking possibility is dramatically decreases.

There are multiple instances of black-box reverse engineering where attackers were able to determine (and break) the previously unknown algorithm. Your opinion is unsubstantiated. No reason to assume quantum AI at all, which might or might not be helpful for cryptanalysis.

Seriously, I'm okay with learning cryptography by messing with innards, but you need the basic understanding of modern cryptography at the very least, including confusion and diffusion. In particular it is very important to know what not to do, that is critically missing from your claims. Modern cryptographers start learning by breaking existing algorithms known to be faulty by their own. Without this active process you are going to repeat past mistakes.