Skip to content

Anoma Specification

Anoma is a privacy-preserving, distributed, trust-aware operating system. Like a typical operating system, Anoma is a platform on top of which applications can run. It provides memory isolation, inter-application communication, and execution. Unlike a typical operating system, Anoma is built for multi-party intent-centric applications. An intent, so named because it conveys an intention, is a message sent by a user describing a preferred state of the system. After a user sends an intent, two phases take place: counterparty discovery and settlement. In the counterparty discovery phase, network participants examine and match compatible intents, forming transactions which enact particular state changes. Transactions then enter the settlement phase for ordering, execution, and confirmation by consensus, after which users can read the updated state. Anoma assumes a model of heterogeneous trust, where users express trust preferences in their intents, only intents with compatible trust preferences can be matched, and consensus rounds are performed on demand to facilitate settlement. To applications, Anoma provides a host environment and state model that handle all of the complex parts of intent-centric, privacy-preserving, and distributed operation. Anoma also comes with a compiler toolchain which allows developers to write Anoma-compatible programs in high-level languages, abstracting the particulars of compilation to the cryptographic primitives required for privacy-preserving operation. This standardisation allows developers to focus on their application-specific logic and enables applications to interoperate in both the counterparty discovery and settlement phases.

Anoma's multi-party intent-centric architecture is designed for applications concerned with coordination of socio-economic processes dealing with resources, distributed capabilities (freedom of action, and control over resources), and agreement between agents (users) under conditions of heterogeneous trust. Intents are computable predicates over possible states of the ledger which convey underlying human intentions or preferences. Anoma supports both arbitrary fungible measures of value (e.g. currencies) and unique (non-fungible) objects, so users can choose the representations and level of precision most appropriate to model aspects of the world which they care about.

Anoma provides a substrate for information flow control, giving users fine-grained control over and the ability to reason about where, when, and to whom information may be disclosed, subject to whatever trust assumptions they are willing to make. In order to provide this substrate, Anoma uses many cryptographic constructions, including public key encryption, one-way hash functions, succinct non-interactive zero-knowledge proofs, distributed key generation, threshold encryption, and homomorphic encryption. Anoma's construction abstracts the underlying primitives by their information-theoretic properties, so that new primitives may be swapped in over time.

These documents describe Anoma. They are intended to be complete, in that enough is said to define precisely what a valid implementation of Anoma must do, and minimal, in that no more is said than that.

Anoma is free, in both the senses of "free speech" and "free beer". The source for these documents is on Github, permissively licensed, and they can be forked or edited as you like. At present, this particular repository is stewarded by the Anoma Foundation. Contributions are welcome.

Note

For clarity, the word "Anoma" is used to refer to three separate but related entities: - The Anoma protocol, which is defined in this specification. - The Anoma network, which is the network of nodes using the Anoma protocol. - The Anoma Foundation, which is a Swiss foundation (Stiftung) chartered with support & coordination of the Anoma protocol, network, and surrounding ecosystem.

These documents describe the protocol only, not the network or the foundation. To learn more about the foundation, go here.

In these documents, the word "Anoma" alone refers to the Anoma protocol.

Note

What precisely we mean by "protocol"? As the words are colloquially used, Anoma is closer to a "protocol architecture" or "protocol topology", in that it defines a class of protocols which are unique up to structural isomorphism. If a "protocol" in the TCP/IP sense can be said to consist of a structure and an encoding, where the structure is constrained by the acceptable assumptions and desired properties, but the encoding makes arbitrary decisions about symbolic representation, a "protocol" in the Anoma sense fixes the structure but not the encoding. Any program preserving this structure can be said to implement the Anoma protocol. Standardisation of an encoding is likely in practice, but strictly speaking not even required for distributed consensus - agents must only agree on the encoding and decoding functions on a per-message basis in order to "understand" each other. This disctinction may seem arcane, but it is potentially of practical importance, for a few reasons:

  • A third party observing the cleartext part of the communications between two agents may not easily be able to detect whether or not they are even speaking the Anoma protocol if they do not know the specific encoding in use.

  • The Anoma protocol may be "disguised" as any sufficently general existing protocol (e.g. HTTP) in a way which cannot be automatically detected.

Note

These documents are not yet complete. If you're reading this page right now, there's a high chance you might be interested in Namada.

Happy reading!