Why is 2PC (2-Party Computation) not used as any other primitive like Diffie-Hellman?

I'm not talking about the general secure computing, but some weaker security requirement or special case schemes.

A number of tricky methods for GC (Garbled Circuit) have been proposed like free-XOR, half-gate and so on. And recent promotions like TinyGarble, LEGO seem practical.

Is there any irreplaceable need for such primitives in the real world? Or is the GC the only way for MPC (Multi-Party Computation)?

  • $\begingroup$ I don't quite understand what your question is. Are you asking why we do not replace special purpose protocols like DH key exchange with general purpose 2PC protocols? $\endgroup$ – Maeher Nov 11 '19 at 12:33
  • $\begingroup$ The general MPC protocol is far away from practice, just like the multiparty KE. While the DH can be the indispensable part of our network security. The optimizations of GC make the 2PC seems practical enough. But I haven't heard any 2PC scheme adopted in the real world. $\endgroup$ – scv10086 Nov 12 '19 at 2:00
  • $\begingroup$ What's your question then? $\endgroup$ – Maeher Nov 12 '19 at 9:52
  • $\begingroup$ If the GC is efficient enough, what's the main obstacle/need for 2PC application? If not, to what extend would the cost of GC be tolerable? Or the GC tech itself is too complicated to deploy? $\endgroup$ – scv10086 Nov 13 '19 at 6:20