This repository contains a paper on IBISS-CIA scheme that addresses the problem of computational integrity proofs (correctness of the actual computations) made by some non-trusted (anonymous) third party without repeating the whole volume of the actual computing. This issue is usually addressed by use of probabilistically checkable proofs (PCP), however, they leave untouched the question of coupling the payments for the consumed computational resources with the actual proof. The purpose of our protocol is to scale distributed calculations by obtaining results of nearly perfect (measurable) quality, without the need to repeat calculations or check the computational integrity proofs by the client. This is achieved by providing meritocratic economic incentives to all participating economic agents, so that their interests are aligned with the network's success. As a result, the need for repeat calculations diminishes, and costs of calculations become lower and competitive.
Ability to run PCP schemes without trusted checking party provides a way to codify the actual protocol in terms of smart contracts, so it can be used in a very compact form with existing economic blockchain layer. Our approach to protocol design aims at the smallest possible size of the shared system state so it can be implemented with any non-Turing complete smart contract system with storing nothing else then the account balances for the involved parties. For example, the protocol can be implemented with the existing Bitcoin script functionality with only two on-chain transactions per whole protocol computing cycle. In such system the PCP proofs are run the second layer on top of actual Bitcoin blockchain, reducing the footprint, price and increasing scalability for the real-world business cases for the computing.
The proposed solution is twofold. Firstly, we introduce a more meritocratic approach, where network participants need to assume responsibility by putting something at stake if they want to receive gains from the network. The greater the rewards that a participant opts for, the greater is the stake that will be lost if the participant does not produce contributions of acceptable quality. Secondly, we develop a process of arbitration, which is essentially a process of quality control and network protection applied specifically to distributed ledgers. Arbitration is about re-checking random parts of work done by network participants and resolving conflicting opinions whenever they occur. We also develop instruments that help to estimate the probability and costs of attack on the network, whether for personal gain or for disruption of its services. Finally, we attempt to align the incentives of all network participants so that the network would be able to use quality control sparingly, e.g. checking only 10%-15% of all work done, and still achieve high quality results.