Why we can't build perfectly secure multi-party applications (yet)
โ General ๐ 2025-01-06 ๐ค enrico ๐๏ธ 168
Warning
I just published a post exploring why building secure multi-party applications is impossible today. These applications aim to compute a function over inputs from different parties while keeping those inputs private. We use Multi-party Trade Credit Set-off (MTCS) as an example, showing how technologies like MPC and FHE fall short of perfect security due to a fundamental tradeoff between security and liveness. The current solution involves a delegated security model, but itโs not ideal. Are there any robust patches to this problem, or will we have to wait for indistinguishability obfuscation? Read the article and share your thoughts here
๐ท๏ธ FHE ๐ท๏ธ MPC ๐ท๏ธ iO
atomictag 2025-02-01 ๐ 1 ๐
In my view the โinteractiveโ nature of most (practical) MPC protocols is always going to be the bottleneck. Hence the appeal of FHE, which, at least in principle, removes that constraint at least to some extent. Perhaps there should be more research, as far I can tell, on practical, concretely efficient NI-MPC protocols, or โlowโ interaction ones. Promising directions are Mr NISC and related/derived works (which try to circumvent the IO problem) or perhaps YOSO-like MPC like SCALES. None of the above are practical by any definition today - but I would love to hear your thoughts on these