This paper presents a protocol for secure multiparty computation where at least 2n/3 participants are honest, ensuring unconditional secrecy without relying on computational assumptions. The protocol uses authenticated secrecy channels between participants and relies on secret sharing and verifiable secret sharing (VSS) schemes. It ensures that no subset of less than n/3 participants can gain information about others' inputs, and that honest participants can correctly compute the result even if up to 1/3 of the participants cheat. The protocol is optimal and does not require cryptographic assumptions for secrecy or correctness. It is based on the work of Chaum, Damgård, and Van de Graaf, who introduced the concept of unconditionally secure blobs. The paper also discusses related work, including the Byzantine Generals Problem and secret sharing schemes. It describes the use of a "cut-and-choose" procedure to ensure the correctness of the protocol and the construction of robust double blobs to handle faults. The paper concludes with a discussion of generalizations and open problems, including the possibility of extending the protocol to tolerate more unreliable participants by using cryptographic assumptions.This paper presents a protocol for secure multiparty computation where at least 2n/3 participants are honest, ensuring unconditional secrecy without relying on computational assumptions. The protocol uses authenticated secrecy channels between participants and relies on secret sharing and verifiable secret sharing (VSS) schemes. It ensures that no subset of less than n/3 participants can gain information about others' inputs, and that honest participants can correctly compute the result even if up to 1/3 of the participants cheat. The protocol is optimal and does not require cryptographic assumptions for secrecy or correctness. It is based on the work of Chaum, Damgård, and Van de Graaf, who introduced the concept of unconditionally secure blobs. The paper also discusses related work, including the Byzantine Generals Problem and secret sharing schemes. It describes the use of a "cut-and-choose" procedure to ensure the correctness of the protocol and the construction of robust double blobs to handle faults. The paper concludes with a discussion of generalizations and open problems, including the possibility of extending the protocol to tolerate more unreliable participants by using cryptographic assumptions.