[filup]

I don't understand the hype around FHE. FHE sounds like a fancy way to say my data is only partially encrypted and we can still gain all the insights we want without technically unencrypting it.

I don't want my encrypted payloads to betray me in any of the ways FHE wants it too.

[perching_aix]

> FHE sounds like a fancy way to say my data is only partially encrypted and we can still gain all the insights we want without technically unencrypting it.

Does it just sound like it or is it? Cause it sure as hell didn't "sound like that" to me last I checked, so that's 1:1 so far.

[Stefan-H]

Consider 2 researchers, Alice and Bob. Lets say that Alice has developed a cool way to analyze gene data, and she uses it on her gene data and gets cool information, so naturally Bob would like to do the same analysis. How does Bob securely get his data analyzed with Alice's intellectual property (which she wants to keep secret as well), enter homomorphic encryption! Bob can encrypt his data in such a way that Alice can run her analysis on it, without Alice ever knowing the content of Bob's data. Alice can get neither Bob's data nor the analysis of it.

[filup]

So, maybe we should change the name from Fully Homomorphic to something else. The schema your describing sounds like leaving the key under the mat for only parts of the data.

Fully sounds like Alice could process the data in absolutely anyway she would like. This schema sounds to complex to become useful for anything but a narrow set of capabilities. It sounds like it would be more effective for Alice and Bob to sign an agreement with each other than for bob to shape his data in a format useful for Alice to run her processes on it.

Why do we need to muddy the water of what encryption means to make the FHE schema work.

[baby]

Am I paranoid or does this comment feels like what an LLM would write to imitate an HN comment?