fully-homomorphic-encryption VS SEAL

Secure computing. This includes Fully Homomorphic Encryption AKA FHE, of which there is a language/compiler which just got released and Google's older FHE compiler. FHE is probably more "compiler" than "type system", e.g. Google's compiler works on C++. Also Security Type Systems which include Oblivious data structures and Oblivious ADTs.

I don't have any advice for you, but I do work on homomorphic encryption at Google and we have an FHE compiler project [1] (though it is likely going to change a lot in the coming year). I happen to have a math PhD, so the transition to this field was not a huge stretch, but before that I worked in supply chain optimization for data centers, and just decided this was too exciting to pass up.

[1]: https://github.com/google/fully-homomorphic-encryption/issue...

+1, and some compilers already exist to do that for you. See, e.g., Google's compiler (which I work on). https://github.com/google/fully-homomorphic-encryption

Once the tools are written, you should be able to take a program written in some language foo and transpile it to a FHE version of foo. See Google's C++ to FHE-C++ transpiler. Thus, you can test/debug your application without FHE before transpiling to something that is FHE.

the protocol implementation built on the opensource SEAL library

FHE has been around for a while now, IBM: https://github.com/homenc/HElib and Microsoft: https://www.microsoft.com/en-us/research/project/microsoft-seal/ both have their own (much more mature) FHE implementations.

It's really great to see more big companies getting into this game, ease of adoption is really the key here.

When it comes to FHE, there are 3 underlying paradigms you can target with compilers:

1. boolean circuits, where you represent your program as encrypted boolean gates. The advantage is that it's as generic as it gets, the drawback is that it's very slow. TFHE is great for that, and it's what is shown here.

2. arithmetic circuits, where you represent your program as a combination of encrypted additions and multiplications. This goes really fast, but you are quickly limited in terms of usecases because you can only do a certain number of arithmetic operations. CKKS/SEAL targets that: https://www.microsoft.com/en-us/research/project/microsoft-s...

3. functional circuits, where you represent your program as a combination of homomorphic functions. Advantage is that you can do very complex things like deep neural network, the drawback being that you have limitations of the bits of precision for the computations. Concrete targets that: https://zama.ai/concrete/

Case 2: the encrypted case. You would be encrypting the data itself, so each cell in your database would be encrypted, the filesystem, the tables, the database, and every other aspect of postgresql is left exactly the same. As a matter of fact, you would install this the same way you would install every other postgresql extension. Just run "make install". The only dependency is Microsoft SEAL(https://github.com/microsoft/SEAL) which is open source and you download it and install it from source as well. That is it.