Ranter
Join devRant
Do all the things like
++ or -- rants, post your own rants, comment on others' rants and build your customized dev avatar
Sign Up
Pipeless API
From the creators of devRant, Pipeless lets you power real-time personalized recommendations and activity feeds using a simple API
Learn More
Comments
-
atheist99852yI'm too drunk to understand maths rn but I think sober me would find this really interesting.
-
Interesting topic. The math is sound - but you have to know beforehand which operations you need to support on which data and there is a limit to the amount of operations performed before the result becomes undecryptable - which basically means, that it is almost guaranteed that there will be really hard to find data loss bugs in practice.
It certainly is easier and less error prone to implement a hash-chain of signed commands instead. The original value is not exposed in any form. Commands of what to do with the field are added to the hash chain and can eventually be applied by a central authority to the original data.
That way, you can add support for new operations at any time and there is no inherent limit on the amount of operations that can be enqueued to be applied to a single field. Also you can ignore invalid commands in the middle of the list without having to discard all following valid commands too. -
cprn17752y@iiii AFAIU by just looking at the infographics, you need to encrypt your data with your public key in a special way before evaluation (not just `gpg -e`) and that special way allows the evaluation algorithm to generate results without decrypting the data itself - your data remains encrypted with your public key throughout the whole process from input to output so only your secret key can decrypt the result.
-
cprn17752y@iiii My guess would be it's not entirely scrambled? It's more like variables. You scramble only the values but you know which variable represents what. Then the algorithm moves things around and derives some equations that you get as output and decrypting process does the most of the calculation. I might be totally wrong, though.
-
@Oktokolo the limited number of operations (called SHE) can be used to bootstrap the unlimited version. (FHE). I wish I better understood the fundamentals because I think its pretty fucking cool.
Want to comment on what you wrote, when I have a more time. At working at the moment. -
atheist99852y@iiii some mathematical operations are commutative, eg addition, a+b=b+a, and if your encryption was to "add c", then your decryption was to "subtract c", then the order of operations wouldn't matter. The decryption step can't be reordered, but the encryption step can. If you construct your encryption algorithm in such a way that certain operations are preserved, but done in a way that still sufficiently obfuscated the data, then you get this. Which is quite interesting.
Related Rants
Fully Homomorphic Encryption (computing addition and multiplication of numbers WITHOUT decrypting) is fucking cool. That is all.
https://bit-ml.github.io/blog/post/...
random
encryption
math