This paper presents a secure and versatile sampling-based verification protocol, Proof of Sampling (PoSP) protocol, suitable for a wide range of decentralized applications. Our protocol has a unique Nash Equilibrium in pure strategies, which compels rational participants to act honestly, thus fortifying the network's integrity. This can be achieved with manageable computational overhead. When applied to decentralized inference for AI applications, we design spML based on PoSP protocol, which ingeniously amalgamates the strengths of optimistic fraud proof and zero knowledge proof based approaches, the foremost approaches in the domain at present. Moreover, the PoSP protocol can be effectively utilized for designing verification mechanisms within Actively Validated Services (AVS) in EigenLayer and Layer 2 solutions, further broadening its applicability. This innovative approach not only enhances the security and efficiency of decentralized systems but also paves the way for a new generation of scalable and reliable decentralized applications.

翻译：本文提出了一种安全且通用的基于采样的验证协议——采样证明（PoSP）协议，适用于广泛的去中心化应用场景。该协议存在纯策略下的唯一纳什均衡，迫使理性参与者诚实行动，从而强化网络完整性。该设计可在可控计算开销下实现。当应用于人工智能的去中心化推理时，我们基于PoSP协议设计了spML方案，巧妙融合了乐观欺诈证明与零知识证明——当前领域内两大主流方法的优势。此外，PoSP协议还可有效应用于EigenLayer的主动验证服务（AVS）及Layer 2解决方案中的验证机制设计，进一步拓展其适用性。这一创新方法不仅提升了去中心化系统的安全性与效率，更为新一代可扩展且可靠的去中心化应用开辟了道路。