bitpro가 원문에서 글자만 모은 내용이에요. 표·그림·수식은 빠지거나 깨질 수 있어요. 전체 4쪽 중 1쪽
Anoma: a unified architecture for full-stack decentralised
applications
Christopher Goes, Awa Sun Yin, Adrian Brink
Pre-release, August 21, 2022
Abstract
Programmable settlement architectures do not enable
counterparty discovery and solving, both of which are
necessary to build the majority of interactive multi-
party applications. The architectural constraints of
programmable settlement result in contemporary ap-
plication protocols that have at least one Web2 com-
ponent, which becomes the centralisation point. We
present Anoma, a unified architecture for full-stack de-
centralised applications. Anoma is designed following
the principles of intent-centricity and homogeneous
architecture / heterogeneous security, together con-
stituting a declarative paradigm for building decen-
tralised applications. In this paper, we first outline
the Anoma architecture, provide an intuition for the
design rationale, and describe how Anoma disentan-
gles the choices of protocol and security. We then
definetheAnomaapplicationprogrammingmodeland
enumerate several existing and novel decentralised ap-
plications that can be built using the novel primitives.
Finally, we outline the current components used to
instantiate Anoma and list future research directions.
1 Background and motivations
The release of the Bitcoin protocol in 2008 marked
the beginning ofscriptable settlement, a category of
distributed ledger architectures that is suitable for
cryptocurrencies with discrete properties and mon-
etary policies. Although it is not Turing-complete,
Bitcoin Script[1] is able to support applications be-
yond currencies, such as Namecoin and Colored Coins.
As discussed in the Ethereum Whitepaper[2], while
applications built on scriptable settlement are func-
tional, this architecture requires too many trade-offs
that resulted in constrained properties and usability.
The introduction of the Ethereum protocol in 2014 set
the precedent forprogrammable settlement, a new cat-
egory of architectures for constructing decentralised
applications that leverage Turing-complete virtual
machine execution, which adds substantially more
expressivity to the settlement layer. Programmable
settlement paved the way for improved versions of
applications that scriptable settlement is not able to
support, suchasfungibletokens(ERC20)orEthereum
Name Service (ENS), which are today well-established
versions of the Colored Coin and Namecoin ideas,
respectively – in addition to many other desirable
applications, such as non-fungible tokens (NFTs), De-
centralised Autonomous Organisations (DAOs), or
the recently introduced Soulbound Tokens (SBTs)[3].
Proposed and deployed blockchain protocols since
Ethereum’s release have brought significant im-
provements to specific architectural components,
for instance: consensus mechanisms (Tendermint[4],
Avalanche[5]), Sybil-resistance mechanisms (proof-of-
stake, proof-of-storage), scaling solutions (sharding,
rollups), and cryptographic schemes (zero-knowledge
proofs) – but these improvements to constituent prim-
itives do not change the basic architecture of pro-
grammable settlement.
While programmable settlement is sufficient for cer-
tain applications, many contemporary applications
have further requirements. Settlement suffices when
the involved parties have already decided what and
with whom to settle, but contemporary applications
1
often also require infrastructure for helping potential
counterparties discover each other and decide with
whom and on what to settle. As a workaround, exist-
ing applications have usually adopted an architecture
that relies on one or many permissioned or centralised
components (such as provers, solvers, or sequencers),
usually implemented as Web2 services, in their stack.
Examples include decentralised exchanges for fungible
assets (0x, CoWSwap, Uniswap), for non-fungible
assets (Wyvern, LooksRare, OpenSea), novel vot-
ing/funding mechanisms (quadratic voting/funding,
Gitcoin), and rollups (Optimism, Arbitrum, Starknet,
zkSync) – their architectures involve at least one
centralised component that often results in a loss
of permissionlessness, fault-tolerance, or censorship-
resistance.
One emerging approach for applications seeking to
avoid centralisation points in their architecture is to
deploy an application-specific sovereign chain to re-
place a specific component in the stack. Even though
this approach can solve the immediate centralisation
problem, it comes with substantial trade-offs, such as
the loss of network effects (application composability
and software re-use) or the addition of disproportion-
ate complexity to developers and users, who need
to reason about multi-layered security, and latency
domains.
In this paper we present Anoma. Anoma is a uni-
fied architecture for full-stack decentralised applica-
tions – characterised by its intent-centricity, decen-
tralised counterparty discovery and computational
outsourcing of NP search problems to solvers which
compute valid state transitions. With this architec-
ture, contemporary applications can be built without
compromising permissionlessness, fault-tolerance, or
censorship-resistance.
Anoma’s architecture also exposes novel primitives,
such as multi-chain atomic settlement, which allows
users and applications with different security pref-
erences to obtain atomicity. These and other novel
primitives pave the way for the development of ap-
plications that cannot be built with existing architec-
tures, several of which we enumerate inSection 5:
Applications.
2 Anoma deployment roadmap
2.1 Galileo: Deploying the Anoma dis-
tributed OS (operating system) to
Ethereum (including other EVM
chains) and further systems (SVM,
CosmWASM, ...)
• Galileo enables developers to use all the func-
tionality of the OS on each supported plat-
form/architecture. Specifically it allows devel-
opers
– to write applications in terms of intents
rather than transactions
– to write an application once and deploy it
everywhere
2.2 Dagon: Enabling multi-domain
concurrency and settlement
• Dagon networks all the individual deployments of
the OS into one whole distributed system. Specif-
ically it means that
– users can use intents from any specific do-
main on every other domain (i.e. cross-chain
trades)
– developers can leverage different underlying
systems for different use cases (i.e. DA from
Celestia, ordering from Ethereum)
2.3 Ahra: Scale-free local and global
consensus for hyper scaling of the
Anoma world computer
• Ahra finalizes the world computer. It allows
for any networked device to become part of the
distributed OS.
• It enables the deployment of local consensus in-
stances, such as enterprise chains or local trading.
• At this point Anoma fully implements the dream
of the world computer.
• It is faster than Solana (local consensus will al-
ways be faster than global consensus) and more
decentralized than Bitcoin (consensus could in-
volve every single human on the planet).
2
3 Token utility
XAN,Anoma’snativetoken, istheglobalcoordination
mechanism that powers the distributed OS that is
Anoma. It faciliates the coordination between all
involved actors, from users over solvers to consensus
providers and is used in everything from a simple
intent to a complex ZKP proof. It does not just
capture a portion of the economic value generated on
a deployment like traditional base layers have done,
but rather captures a portion of the total value that is
coordinated anywhere on the distributed OS (Anoma)
and as such its value capture dynamic are superior
to existing comparables. Fundamentally the utility
comes from powering global economic coordination,
whether it happens on Ethereum, Solana, Plasma, a
sovereign corporate chain or as part of local consensus
between 3 phones.
• Users, whether manual or programmatic, include
fees alongside their intents in order to pay for
their execution
• Solvers, which can perform both traditional
matching, similar to market makers or also other
roles such as ZKP proving or data storage, charge
fees for those services.
• Consensus providers, which provide ordering, fur-
thermore charge fees.
• Other types of nodes, such as DA, can also con-
figure their individual revenue model.
Specific Anoma instances can choose whether to allo-
cate the fees to a strategic reserve, distribute them to
service providers or burn them. Most likely economi-
cally it is the most advantegeous to do the latter and
as a result most deployments are likely to choose that
route.
Besides coordinating all the actors involved in the
economic flows, XAN also faciliates the continuous
evolution of the distributed OS, it’s many deployments
and the core software powering it. As such the token
can be used to: * signal the desire to upgrade to
a specific new software release. * move value to an
upgraded deployment. * perform actions to upgrade
local or global instances without requiring them to
build and deploy their own governance system.
To summarize, rather than only using the native asset
for gas payments in a single base layer, with Anoma
the token is used by every actor in the distributed
OS and fee capture can be proportional to all the
economic flows facilitated by Anoma without being
limited to a specific deployment. It allows the token to
capture a portion of the total value that is coordinated
anywhere on the distributed OS (Anoma) and as such
its value capture dynamic are superior to existing
comparables. Fundamentally the utility comes from
powering global economic coordination, whether it
happens on Ethereum, a sovereign corporate chain or
as part of local consensus between 3 phones.
4 Architectural design philoso-
phy
Anoma’s architecture is driven by two design princi-
ples: first, intent-centricity; second, a homogeneous
protocol architecture with a heterogeneous security
model. Beyond these two design principles, all other
architectural choices are a matter of modularisation
and runtime configuration parameters.
4.1 Intent-centricity
Anintentis an expression of what a user wants to
achieve whenever they interact with a protocol, for
instance “transfer X from A to B” or “trade X for Y”.
Practically, an intent is an off-chain signed message
that encodes which state transitions a user wants to
achieve. Unlike transactions, intents are partial, so
one can think of intents as parts of transactions that
require other direct or indirect parts as complements
in order to form a final balanced transaction which
satisfies all users’ constraints.
Existing protocols are designed withtransactionsas
their most fundamental unit. Anoma takes a radi-
cally different approach: the architecture of Anoma
is centred around programmaticintents.
An intent-centric architecture is necessary to enable
counterparty discovery, which is crucial for compelling
applications, since they require multiparty coordina-
tion and to enable full-stack decentralised applications.
3
Anoma vertically integrates counterparty discovery,
solving, and settlement, and is able to interpret and
process intents natively and generically. Contempo-
rary applications, as described earlier, require both
counterparty discovery, solving, and settlement. In-
tents are the point at which users interact with such
applications, and an intent-centric design captures
the requirements of applications which need these two
processes to work in tandem and satisfy censorship-
resistance, and fault-tolerance properties.
Intent-centric design also constitutes adeclarative
paradigmfor building applications, since Anoma is
designed to settle intentsas definedby the users – an
intent is either settled as defined, or not settled at
all. This declarative model gives users a significantly
higher degree of control,withoutrequiring them to
understand the underlying protocol primitives and
execution flows, which is crucial in order for decen-
tralised applications to reach mass adoption. This
paradigm presents a radically different approach as
compared to existingtransaction-centricarchitectures
that default to animperative modelfor applications.
In the latter, users are required to understand the
full execution trace to benefit from security and data
locality guarantees, because instead of authorising
a specific state change, they authorise specific exe-
cution paths. In practice, this is so difficult that
users commonly interact with applications without
understanding the risks.
For application developers, Anoma’s intent-centric
architecture enables them to buildsafer by construc-
tion applicationsby leveraging the combination of
intents andvalidity predicates. Validity predicates are
an architecture for smart contracts which separate
out cleanly the task of computing state transitions
and the task of verifying correctness of state transi-
tions, as compared to message-passing VM execution
models (pervasive in current programmable settle-
ment architectures) which interleaves computation
and verification. Validity predicates allow applica-
tion developers to reason about the invariants which
they would like their application to satisfy without
worrying about how other applications interact with
it, since the validity predicate of their application
expresses these invariants directly.
4.2 Homogeneous architecture, het-
erogeneous security
The Anoma protocol, just like the TCP/IP protocol
stack, follows the principle of homogeneous architec-
ture and heterogeneous security. In TCP/IP, the
various layers of the internet protocol are standard-
ised, but the choice of whom to connect to and what
data to entrust them with is left to the user, and
different users can make different choices while using
the same protocol stack. In Anoma, the various layers
of counterparty discovery, solving, and settlement are
similarly standardised, but the choice of what security
domains to trust and what data to send to whom are
left to the user, and different users can make different
choices while using the same protocol stack.
In this framework, protocols can be analysed along
two dimensions: architecture and security.
• Architecture: the abstractions and relations
constituting the structure of a system. An ar-
chitecture is syntactical, possessed of properties
and syntaxes but with no particular semantics in
relation to the exterior world. Convergence on a
singular architecture saves time and verification
costs without constraining users to particular
choices.
• Security: the choice of whom and how to trust in
the operation of a distributed system. Security is
a decision inseparable from the particular seman-
tics of a specific context of use. While security
can be economically abstracted to a certain de-
gree by limiting the information available to and
consequent choice-making capabilities of system
operators, operators will always have choices of:
how and from whom to accept messages; when
to elect to include them in blocks or other aggre-
gations over which they vote; and when to cease
voting or otherwise alter normal operational pro-
cedures in response to exceptional circumstances.
Whomtotrustwiththeseresponsibilitiesdepends
on what the state in the databaserepresentsin
the real world, and alignment with the interests
of users of the database requires mutual interests
beyond the purely economic ones.
4
4.2.1 Analysis of platforms
Consider distributed ledger platforms, from the per-
spective of applications running on top of them, along
these two dimensions: protocol architecture and secu-
rity model, and whether they arehomogeneousorhet-
erogeneousfor different applications running on the
same platform.
Protocol architecture refers to the state layout, vir-
tual machine, language support, sharding mechanisms,
cross-contract messaging model, etc. An architecture
determines what is required to write an application
for a platform, and applications are specific to a par-
ticular architecture.
• Platforms with ahomogeneousarchitecture re-
quire that all applications are written in a certain
format (e.g. EVM bytecode or WASM).
• Platforms with aheterogeneousarchitecture allow
applications to be written in different formats,
perhaps with some agreement at the edges, such
as cross-chain communication protocols.
Security model refers both to securityin theory, such
as fault tolerance properties of the consensus, fork
detection and handling; and securityin practice,
i.e. which miners or validators operate the deployed
instances of these architectures.
• Platforms with ahomogeneoussecurity model
have the same security for all applications.
• Platforms with aheterogeneoussecurity model
have different security characteristics for different
applications.
For illustration,Table 1situates several platforms
on these two axes:
As the table suggests, these dimensions are generally
quite correlated: homogeneous architectures come
with homogeneous security models, and heterogeneous
architectures come with heterogeneous security mod-
els. It is easier to design a system where they are
correlated. If everything is homogeneous, protocols
can be fit together neatly, and functionalities includ-
ing cross-contract communication are easy; whereas
if everything is heterogeneous, protocols just agree
on the edges of interaction, for instance via the Inter-
Platform Architecture Security Model
Bitcoin Homogeneous Homogeneous
Ethereum Homogeneous Homogeneous
Ethereum 2.0 Homogeneous Homogeneous
Polkadot Heterogeneous Homogeneous
Near Homogeneous Homogeneous
Cosmos Heterogeneous Heterogeneous
Multichain Heterogeneous Heterogeneous
Anoma Homogeneous Heterogeneous
Table 1: An analysis of platforms based on their
architecture and security model
Blockchain Communication protocol (IBC)[6], and
handling the complexity of security is up to the users
and application developers.
4.2.2 Why decouple these dimensions?
Anoma’s fractal instance architecture is designed to
decouple these dimensions and build a platform which
is architecturallyhomogeneousand with aheteroge-
neoussecurity model. This is more complicated, but
it separates out the question of what the bestproto-
col architectureis, where there may be a “benevolent
monopoly” (à la Git or TCP/IP), from the question of
what is the bestsecurity model, where there is almost
certainly not.
Applications written for fractal instances can stan-
dardise on the architecture Anoma offers, which is
sufficiently well-defined to allow for complex inter-
operability, automatic scaling, etc., without agreeing
on any single security model. Furthermore, in some
cases, this flexibility of choice can be extended all
the way to users of the applications, who can choose
independently.
User interfaces for Anoma instances can support the
same applications deployed with different security
models, and communicate that latter difference to
users in a way which allows them to choose their trust
assumptions while retaining the network effects of
using the same protocol.
Noteworthily, Anoma’s architecture is not homoge-
5
neous like a straitjacket, as it supports multiple de-
ployment models. The components in the protocol are
layered so that fractal instances can pick and choose
which parts they participate in, even if it involves
leveraging Anoma for specific functionalities, such
as decentralised counterparty discovery and solving,
whilst anchoring the final settlement on another plat-
form, such as Ethereum. Nonetheless, a unified and
vertically integrated architecture allows developers
and users to benefit from standardisation.
5 Architectural topology
Anoma’s architectural topology consists of a set of log-
ical abstractions delineated by their role in dataflow,
independent of particular forms of representation, de-
ployment models, choices of cryptographic implemen-
tation, etc,Figure 1provides an overview of the
architectural topology. Particular instantiations carry
different concrete performance and security implica-
tions and should be chosen according to requirements
of the specific deployment in question. We offer a
sketch프로젝트로 돌아가기