Computational and Functional Principles

Computational Reflection The SAGA framework will provide avenues for computational actors to engage in various levels of computational reflection concerning: (1) resource allocation, (2) data representation, (3) execution introspection: In terms of physical resource allocation, computational reflection will empower agents to maintain ongoing interaction with their execution environments, enabling them to search for and request additional computational resources and infrastructures based on their specific criteria. Additionally, agents will have the capability to download or update necessary libraries, thereby evolving their execution strategies and facilitating free migration between nodes (virtual machines, cloud vendors, private computers, or mobile phones) within a distributed computing environment; The aspect of capability and data representation will enable agents to semantically represent their computational abilities and input/output data. This information will be accessible for other agents to query when negotiating contractual agreements and designing workflows; Execution introspection refers to each agent's ability to monitor actual resource utilization by its algorithms, maintain a history of execution times and memory usage, and access its own state during execution, among other functionalities. Agents may choose to share some of this information with the network to demonstrate their capabilities and quality of services; It's important to note that the abstraction of SAGA does not dictate how computation or actions of agents will be executed. Through computational reflection, computational agents will have the flexibility to design and implement workflow design and execution functionalities according to their specific preferences and requirements. Moreover, a human element can seamlessly integrate into the same model. For example, an agent could represent a user interface (UI) through which tasks requiring human intervention can be performed and integrated into the workflow. A scenario of hybrid computer-human collaboration could involve SAGA workflows defining a computational task, which is then carried out by humans through crowdsourcing or freelancing platforms (such as Amazon MTurk, Udemy), or even code hosting platforms (such as GitHub, Bitbucket, etc.).

The tokenomics of SAGA will support and facilitate decentralized marketplaces where both human and machine tasks will be requested, offered, and contracted on commercial or other bases. Computational reflection will empower resource owners to advertise and price their capabilities, while resource users will be able to estimate, track, and manage computation costs dynamically and transparently.