PreprintPDF Available
Preprints and early-stage research may not have been peer reviewed yet.

Abstract and Figures

Web3 is the most hyped concept from 2020 to date, greatly motivating the prosperity of the Internet of Value and Metaverse. However, no solid evidence stipulates the exact definition, criterion, or standard in the sense of such a buzzword. To fill the gap, we aim to clarify the term in this work. We narrow down the connotation of Web3 by separating it from high-level controversy argues and, instead, focusing on its protocol, architecture, and evaluation from the perspective of blockchain fields. Specifically, we have identified all potential architectural design types and evaluated each of them by employing the scenario-based architecture evaluation method. The evaluation shows that existing applications are neither secure nor adoptable as claimed. Meanwhile, we also discuss opportunities and challenges surrounding the Web3 space and answer several prevailing questions from communities. A primary result is that Web3 still relies on traditional internet infrastructure, not as independent as advocated. This report, as of June 2022, provides the first strict research on Web3 in the view of blockchain. We hope that this work would provide a guide for the development of future Web3 services.
Content may be subject to copyright.
Exploring Web3 From the View of Blockchain
(Tech Report)
Qin Wang?4, Rujia Li?2,3, Qi Wang2, Shiping Chen4,
Mark Ryan3, Thomas Hardjono1
1MIT Connection Science, Massachusetts Institute of Technology, USA
2Southern University of Science and Technology, China
3University of Birmingham, United Kingdom
4CSIRO Data61, Australia
Abstract. Web3 is the most hyped concept from 2020 to date, greatly
motivating the prosperity of the Internet of Value and Metaverse. How-
ever, no solid evidence stipulates the exact definition, criterion, or stan-
dard in the sense of such a buzzword. To fill the gap, we aim to clarify
the term in this work. We narrow down the connotation of Web3 by sep-
arating it from high-level controversy argues and, instead, focusing on its
protocol, architecture, and evaluation from the perspective of blockchain
fields. Specifically, we have identified all potential architectural design
types and evaluated each of them by employing the scenario-based ar-
chitecture evaluation method. The evaluation shows that existing appli-
cations are neither secure nor adoptable as claimed. Meanwhile, we also
discuss opportunities and challenges surrounding the Web3 space and
answer several prevailing questions from communities. A primary result
is that Web3 still relies on traditional internet infrastructure, not as in-
dependent as advocated. This report, as of June 2022, provides the first
strict research on Web3 in the view of blockchain. We hope that this
work would provide a guide for the development of future Web3 services.
Keywords: Blockchain ·Web3 ·Internet of Value ·Architecture
1 Introduction
Web3, also known as Web 3.0 or decentralized web, hits the cryptocurrency
markets and blockchain communities from 2020 to date [1][2]. It has become the
most prevailing term in the recent period of blockchain prosperity. The concept,
proposed by Wood [3], promises to provide distributed internet services with-
out trusted third parties (TTP), thereby offering users more control over their
data. The very primary principle shared by Web3 applications is that users can
hold the data with full control, covering identifiers/tokens/ownership/etc., rather
than being managed by centralized organizations as in Web1 and Web2 (cf. Ap-
pendix A). With the emphasis on decentralization, Web3 moves data away from
?These authors have contributed equally to the work.
Affiliations are ordered by the text length.
arXiv:2206.08821v1 [cs.CR] 17 Jun 2022
these central authorities and establishes applications and services surrounding
blockchain technologies. Ideally, Web3 developers do not need to build applica-
tions on top of a single server (processing business logic) or database (storing
user data). Instead, Web3 applications are deployed on decentralized networks
such as blockchain platforms or related distributed systems hosted by many
peer-to-peer (P2P) servers.
The conversion from central authorities to the blockchain, more than handing
over the ownership back to users, brings many non-functional benefits of being (i)
open: Web3 data is stored in an open network developed by public communities.
Also, Web3 applications are executed in a global view towards the public, making
data permanently visible to all the participants; (ii) trustless: a user can build
connections or exchange assets to an unfamiliar user without the reliance of
a trusted third party; (iii) permissionless : users’ identities are no longer tied
to any specific platform and users’ activities are free, which do not need an
authorization from a governing entity; (iv) anonymous: users can obtain partial
anonymity through using multiple pseudonyms or off-chain storage; and of (v)
high availability: Web3 provides a high availability architecture, which reduces
the probability of the server crash or single point failure; (vi) compatibility:
Deployed services and applications are not limited in the Ethereum ecosystem,
but applicable to all competitive blockchains [4,5] like Avalanche [6], Solana
[7], Binance Smart Chain [8] or Oasis Network [9], indicating that Web3 can
integrate both existing public-chain ecosystems and incoming systems.
Web3 also improves user experience in using web stack technologies and thus
promotes the development of blockchain technologies. Web3 helps to build a
complete user experience that can better serve newcomers. Users, from the cur-
rent view, only need to connect the wallet to their targeted sites. These sites
contain the activities that a user would love to spend time on, which can be
specified in different types, such as asset swaps (in DEXes [10]), games [11], and
trades (of NFTs [12]). Meanwhile, users can obtain a portion of revenues by
keeping active in these projects. The contributions are measured in a wide range
covering spent time, interactions, deposits, staking, etc. Users can receive fair re-
wards as well as add more liquidity to the Web3 ecosystem. In this sense, Web3
has greatly extended the boundary of blockchain, incentivizing more users to
participate in the games. The evidence is, as in Dec 2021, the total locked value
(TLV) of Web3-related smart contact contract has reached a peak with 193.14
billion USD (more details in Sec.5.1). Such a high capitalization market further
stimulates the prosperity of techniques surrounding blockchain (cf. Tab.3).
However, despite the fact that Web3 has drawn much attention, clear statements
of what is Web3 and how Web3 is designed are even absent. Several studies
provided their investigation on the consensus level [13] but failed to cover a full
view of other components and architectural designs that are equally important to
Web3. Unclear definitions and non-agreed consensus indicate that Web3 either
is just a hyped concept without practical development or, has more than one
single direction for the development. In this technical report, we avoid such high-
2
level discussions, narrowing down the scope to the Web3 architecture and its
relation with blockchain. We dig into its building process, investigate its usage
in existing projects, and deconstruct the protocol into separated components.
With this in the arm, we analyze and evaluate state of the art Web3 solutions
from perspectives of their design patterns and properties. We further extend our
scope to the entire Web3 space (still rooted in blockchains) by discussing their
impacts, opportunities, and challenges. To our knowledge, this work provides the
first in-time study on Web3, reviewing and exploring the wild Web3 solutions.
- We investigate existing Web3 projects and extract a succinct backbone model
with participating roles and operating workflow to demonstrate the major
processes. This provides a loose general model for easy adoption, as well as
highlights the core steps to establish a Web3 service.
- We classify a total of twelve architectural designs of Web3 to reflect the
operating mechanism of typical Web3-based applications. We decouple a full
Web3 service into three-layered components according to the data workflow.
Data in each component can be either operated by the ways of on-chain,
off-chain or hybrid. The identified design types can well present all potential
combinations, covering a wide range of Web3 applications and services.
- We accordingly evaluate each architecture based on different property met-
rics that are grabbed from classic blockchain systems. The evaluation anal-
yses the architectures from multi-dimensional properties (cf. Tab.1) when
adapting to real applications. We also discuss which participating entity can
gain the most benefits under different types. The evaluations directly help
software architects to evaluate and compare Web3 solutions and further de-
sign their architectural frameworks.
- We extend our scope from architectural design to the entire Web3 space by
discussing its impacts, opportunities and challenges to predict the potential
future directions. We show several promising fields that may be inspired by
Web3-related technologies, and also point out the open challenges needed
to be addressed in the long term. Meanwhile, based on our analyses and
evaluations, we answer several questions that are frequently asked by the
communities as our closing results to conclude this work.
- Further, we provide much more surrounding knowledge that is related to
Web3, covering its basic primitives, referenced token standards, and adopted
programming languages. Necessary primitives cover both architectural de-
sign in conventional web services and layered components in classic blockchain
systems. The knowledge lays the foundation of Web3, presenting the common
components in different blockchain-based Web3 projects.
The rest of this paper is organized as follows. Section 2 presents a typical Web3-
based protocol, together with the inducted security model. Section 3 presents
our architecture extraction and design, as well as gives the criterion used to cap-
ture the features of Web3 related applications. Section 4 details our evaluation
3
towards architectural designs, providing practical adaption under different sce-
narios. Section 5 discuss current impacts, opportunities and challenges. Finally,
Section 6 gives our closing results and Section 7 concludes this work. Notably,
we provide more knowledge of Web3: Appendix A gives fundamental primitives
surrounding Web3 and an abstract Web3 workflow. Appendix B and C sum-
marize existing token standards and EVM-operated programming languages,
respectively. Appendix D shows a small corpus of in the wild Web3 projects.
2 Web3 Protocol
This section extracts the protocol, security model, and a running example of
Web3 systems between user clients and the blockchain infrastructure (together
with the protocol workflow, referring to Fig.3 in Appendix A.3). Before our
investigation, we clarify the entities that participate in Web3 protocols.
Involved Entities. The Web3 protocol mainly consists of three types of roles,
namely, the Web3 user,service provider, and the blockchain maintainer. The
Web3 user is a data owner who can propose the request by sending a transac-
tion from the client. The service provider offers the user an on-chain interface
and the service that can both process the request and interact with blockchain
platforms. The blockchain maintainer provides an operating environment for
smart contracts that take the task of processing business and data storage.
General Construction. To launch the Web3 service, a Web3 user needs to
first establish his distributed identity by creating an address or account on-
chain at the client. Then, the user login the applications using his address-based
identity. Subsequently, the application will interact with the blockchain for data
computation and data storage. Here, we present the detailed protocol.
Identity Creation (Π1). This algorithm takes as input the security param-
eter κ, and outputs a blockchain address/account addr. The address provides
a public and open-source index that can be associated with an existing iden-
tifier. Also, this address is used as an identifier for sending or receiving the
cryptocurrency. It covers three sub algorithms, private key generation,public
key generation and address generation.
sk KeyGensk(1κ),
pk KeyGenpk(sk),
addr AddressGen(pk).
The algorithm is run by a local Web3 client (e.g., client-based wallet, browser
extension-based wallet). Note that the sub-algorithms KeyGensk,KeyGenpk
and, AddressGen have different manifestations. For example, for AddressGen,
Bitcoin addresses use the base-58 encoding1, while Ethereum addresses adopt
the base-16 encoding algorithms.
1https://tools.ietf.org/id/draft-msporny-base58-01.html
4
Transaction Generation (Π2). This algorithm takes as input the user’s
private key sk, the transaction metadata, the transaction payload, and ac-
cordingly outputs an enveloped transaction T x.
sig Sign(sk, metadata, payload),
T x TranGen(sig, metadata, payload).
This algorithm is run by a Web3 client that receives the request from users. sk
is used to generate a signature. metadata refers to transaction-related data,
such as the transaction receiver and transaction nonce. payload points to the
contract method to be involved and the data to be stored.
Contract Execution (Π3). This algorithm takes as input the transaction T x,
and current state s, and smart contract contract, and accordingly outputs the
transferred state s.
sContractExec(s, T x, contr act).
This algorithm is run by the blockchain maintainers in the network. The
transited state sincludes a reward or payment used to incentive the participat-
ing nodes. The contract contract contains the logic of upper-layer applications,
which defines how the state changes happening on the blockchain.
State Consensus (Π4). This algorithm takes as input the transaction T x, the
smart contract contract and the current state sto be transited, and outputs
the confirmed state s0, and the confirmed transaction T x0.
(s0, T x0)Consensus(s, T x, contract).
This algorithm is run by blockchain maintainers. The term confirmed indi-
cates that the proposed block (containing sand T x) has been agreed upon by
sufficient maintainers in the network. The threshold of confirmation depends
on specific algorithms (e.g., 51% in PoW [14], 2/3 in BFT algorithms [15]).
State Retrieval (Π5). This algorithm takes as input the user’s address addr
and outputs the transaction T x00 and the related state s00.
(s00, T x00)Retrieval(addr, contract).
This algorithm is run by a Web3 client who aims to retrieve the state.
Secure Web3 Protocols. Based on the extracted general construction, a secure
Web3 protocol is informally defined as follows.
Definition 1 (Secure Web3 Protocol) A Web3 scheme Sis secure, if for all
blockchain address addr and the initial state soutputted from Phase Π1to Phase
Π5, it holds that,
5
sig Sign(sk, metadata, payload)
T x TranGen(sig, metadata, payload)
sContractExec(s, T x, contract)
(s0, T x0)Consensus(s, T x, contract)
(s0, T x0)
(s00, T x00)Retrieval(addr, contract)(s00, T x00 )
where the output satisfies:
(s0, T x0)=(s00, T x00)
The definition states that a Web3 protocol is deemed to be secure if a user can
successfully retrieve the correct state and transaction on the blockchain at any
time after the block confirmation. Our security model is based on the assump-
tion of a robust blockchain [16] and the corresponding security is guaranteed
by persistence and liveness [17]. The persistence means whether different nodes
have the same view at a specific height of blocks, while the liveness focuses on
whether a block can be eventually buried deep enough (without the possibil-
ity of being reversed) in the valid longest chain. We omit other types of chain
structure, such as the directed acyclic graph (DAG) [18].
Fig. 1: An Instance of Web3 User Case
A Running Example. We provide an instance to describe how a Web3 sys-
tem works. The case is straightforward: assume that a user wants to sell his
self-created NFT painting on-chain (cf. Fig.1). Firstly, he needs to connect his
wallet to the related website and send the request to mint an NFT on his targeted
blockchain platforms [12]. The action of mint requires invoking APIs that en-
able interactions with smart contracts. The contract is executed according to its
predefined specifications and token standards (ERC721, see details in Appendix
B). After processing the logic, the raw picture is stored in an external decen-
tralized storage network, e.g., IPFS [19], due to its high-resolution format. IPFS
6
separates the raw data into pieces and distributes them into different nodes,
labelling each of them with a content identifier (CID). The returned identifiers
are recorded on-chain for further queries from users. Once completing the entire
logic, the seller can trace the transaction and his NFT on the blockchain. The
buyer with willingness can buy the seller’s NFT on Opensea [20], and pay the
prices as stated. The ownership of this NFT will be automatically transferred to
the buyer once completing the payments.
3 Architectural Design
The architecture is the foundation of software systems. It defines the system
structure and external-world interface via underlying components and their re-
lationship. This section provides our architectural design for Web3 systems.
3.1 Design Principle
We stay aligned with the mainstream recognition of Web3 and accordingly struc-
ture our report by progressively presenting the ways to build a Web3 service and
the featured properties in different designs. Based on plenty of investigation to-
wards in the wild Web3 projects, we classify them into a compact framework (cf.
Fig.2), which thus allows the audience to explore the conceptual design space
and evaluate different design options.
Specifically, we delineate two basic axes, namely, decoupled component and tech-
nical route, to explore the subtle differences in each cross from orthogonal di-
rections. For the former, we abstract three critical components, including client,
compuation and storage, as our characterisation. Our decoupling is based on the
data workflow, which contains three corresponding phases (A-C in Fig.2), data
access,data computation and data storage, to describe how data transfers under
a normal operating logic. For the latter, we extract three technical routes of im-
plementing a blockchain system: on-chain processing, off-chain processing, and
their combinations (hybrid).
By interweaving two axes, we can find that each component can be categorized
into sub-items according to its on-chain/off-chain processing route. Based on
that, we identify a total of twelve types of architectural designs to show different
purposes and usages among a large bunch of projects. Moreover, we capture
several essential properties that are frequently mentioned in classic blockchain
systems to evaluate the proposed design types. This framework captures major
architectural characteristics and related properties of each type, helping Web3
users and software architects to choose the proper design for their products. We
give the detailed definitions of each item as follows.
3.2 Decoupled Components
We first investigate how data flows from a user to its backend server. In such
a procedure, we have totally extracted three major processes to normal system
7
Fig. 2: Framework. We have first identified three major components according
to the data workflow: access, computation, and storage (cf. left Square). Then,
we extract three technical routes that reflect the ways of data processing in
blockchain systems: on-chain, off-chain, and hybrid (medium Square). Based on
the two principal dimensions, we summarise a total of 12 types of potential Web3
architectural designs (bottom Square). With this in the arm, we analyze every
type in terms of different property metrics (right Square ) and point out design
trade-offs. Notably, we also provide a guide map by using blue dotted line.
operations: data access,data computation and data storage. Such procedures
describe a typical workflow of blockchain systems.
-Data Access. To start a Web3 service, a user first needs to send his request
through a client. Typically, the request is formatted as a transaction, and
accordingly, the client is instantiated as a wallet1. Users enter the network
by connecting their wallets to a specific website.
-Data Computation. After a successful connection, the requests from users
are parsed into different pieces of logic, being transmitted into backend
blockchain platforms. Blockchain operates the logic and decoded methods
via smart contracts according to their predefined specifications.
-Data Storage. The executed data, at last, should be permanent storage.
In normal cases, small size data can be directly recorded inside transactions,
whereas in some cases, large-scale data requires external storage. The key
to maintaining the data integrity in off-chain storage is to add a hook that
connects the on-chain (such as a hash) and off-chain data.
1We omit the sub-categories of different wallet types, such as whether utilizing SPV
or hardware. More details refer to [21][22].
8
Then, based on descriptions of data flow, we abstract three fundamental compo-
nents in a full-functional Web3 service by decoupling existing solutions. Merging
with our different data processing routes, we categorize the decoupled compo-
nent with on-chain,off-chain and hybrid options (selected by two or all of them).
We give their definitions as follows (and more details in Sec.7). To be noted, as
Web3 is featured by decentralization, at least one of the components should be
operated on-chain, where no gatekeeper can fully control the network.
Client. Clients are used to receiving requests from users. Classic blockchain
systems, on a small scale of requests, can use a single browser as the client to
connect with the wallet. However, if the request sharply increases within a short
time frame, an agent is required to process an instantaneous flow of requests.
-Browser-based Wallet. A browser-based wallet is an intuitive solution for
users to adopt the Web3 service. Users only need to install an extension tool
in the browser and import their private key to this embedded wallet. When
browsing a Web3-supported website, users can directly put the button of
connect the wallet, and all the clicked functions in this website will invoke
the backend methods through APIs under the user’s account.
-Agent-based Wallet. An agent solution is to enable batch processing when
confronting a high-density situation of requests from users. Similar to tra-
ditional Web1/Web2, users should first grant a trusted agent with proper
permission. The authentication procedure is executed once users have for-
mally started to register with the agents.
Computation. Operating a Web3 service built on smart contracts requires great
computing power as the running environment. Purely on-chain computation is
instant but costly. A hybrid computation that moves partial computational tasks
off-chain is another solution.
-Contract-based Computation. All the computations are performed on-
chain through smart contracts. Here, variables in smart contracts are allowed
to be revoked and modified based on the predefined contract instructions.
-Hybrid Computation. A partial proportion of computations are executed
on-chain, while the rest of the computations off-chain. Generally, verifica-
tion mechanisms are required in this method in order to guarantee integrity
between on-chain and off-chain data.
Storage. Similar to any services applied to Web1/Web2, applications and ser-
vices in Web3 also require enough space for storage. Data in small size, such
as ordinary transaction payload, can be stored directly on-chain. In contrast,
for the large size, or complicatedly formatted data, like streaming video or raw
audio, it should be (partially) stored to external deceives or providers. In this
sense, we summarize three types of data storage.
-On-chain Storage. All the data is stored on-chain, as a state of the smart
contract. Any changes will be publicly recorded and can be verified by the
9
entire network. Theoretically, this is the best way to prevent malicious nodes
from destroying data due to its total transparency and accountability.
-Off-chain Storage. It refers to storing data in off-chain networks where
only the hash roots are recorded online (e.g., IPFS [19] or Swarm [23]).
The solution requires an additional process for verification, such as proof-of-
existence, to prove data integrity.
-Hybrid Storage. This is a mixed way that combines both on-chain and off-
chain storage, which means storing raw data off-chain (like video and audio)
and light metadata (such as account history and certificates) on-chain. The
proportion of these solutions is adjusted according to specific scenarios.
3.3 Architectural Designs
We have identified totally twelve types of architectural designs towards Web3
services (cf. Architectural Type 1 12 in Fig.2). Specifically, we denote each
design with Type, and it is made up by a three-element tuple (Aa, Bb, Cc) where
a, b {1,2}, and c {1,2,3}. Here, the elements A, B, C separately specify
the components of client,computation and storage, while the subscripts 1,2,3,
respectively, represent on-chain,hybrid and off-chain as defined before. Based on
different combinations, we provide the potential architectural designs as follows.
-Type1 (A1, B1, C1), Type2 (A1, B1, C2), Type3 (A1, B1, C3), which are shown
in the black line at the left square in Fig.2;
-Type4 (A1, B2, C1), Type5 (A1, B2, C2), Type6 (A1, B2, C3) (brown line);
-Type7 (A2, B1, C1), Type8 (A2, B1, C2), Type9 (A2, B1, C3) (blue line);
-Type10 (A2, B2, C1), Type11 (A2, B2, C2), Type12 (A2, B2, C3) (teal line).
In each design type, smart contracts play the most essential role in entire ser-
vices that are responsible for on-chain calculations. Type1 is the most simplified
architectural design amongst those combinations. A typical instance is known as
today’s Ethereum platform. To fit for more complicated cases, external resources
should be equipped to the system design, such as the agent in Type7-12 or the
off-chain storage in Type2/3. Notably, the running NFT example in Sec.2 is a
based on the Type2 design.
4 Architecture Evaluation
In this section, we present the evaluation criteria and evaluate each type of
architecture. Our evaluation covers the properties that are summarized from
classic blockchain systems and benefits to different participating parties.
10
4.1 Evaluation Criteria
In the past decades, many evaluation frameworks have been proposed. Among
them, the architecture trade-off analysis method (ATAM) [24] is a well-known
one. We partially adopt ATAM to assess the Web3 architecture, integrating with
the blockchain properties and beneficial parties. In particular, we first select a
concrete scenario, refine quality attributes into this scenario, and then present
potential concerns related to the quality requirement.
ATAM Method. The architecture trade-off analysis method is initially used in
software development to recommend the most suitable architecture for a specific
system. The method majorly discusses the trade-offs between different design
types and their sensitivity points for risk mitigation. A classic ATAM process
consists of three critical aspects: extract quality attributes,identify beneficial
parties, and analyse in specific scenarios. Quality attributes are non-functional
requirements to specify how well software should be done. In the context of
Web3, we adopt the properties extracted from classic blockchain systems as
quality attributes. Then, we analyse the system entities regarding their potential
benefits and losses. Meanwhile, we still use a similar NFT instance (compared
to the running example in Sec.2) as our analysed scenario.
Property Metrics. We capture several major properties that are used to eval-
uate classic blockchain systems. These metrics depict the architecture from mul-
tiple dimensions, and we summarize them as follows.
-Performance. Performance is used to show the maximum rate of confirmed
transactions during a specified duration. In the context of blockchain, per-
formance is measured by transaction per second (TPS), which is a widely
adopted way to describe the throughput of a system.
-Scalability. Scalability refers to the ability to process transactions along with
the increased scale of networks, which is reflected by the increase or decrease
in its performance (measured in transaction per second–TPS).
-Cost. Each transaction in VM-engined blockchain platforms requires certain
transaction fees. This is necessary because a total zero friction (free of fees)
will result in a loop logic error [25] in smart contract operations.
-Security. Security is a broad term. We limit the scope of security to data-
level security. The data should be protected for its integrity and prevented
from unauthorized access or faking, or compromising.
-Anonymity. Anonymity refers to hiding the identities of acting persons.
In the context of blockchain, anonymity mainly indicates utilizing crypto-
graphic schemes to break the linkage between physical entities with virtual
public addresses [26]. An adversary cannot learn useful information about
any specific parties related to the on-chain assets.
-Confidentiality. Confidentiality refers to hiding the sensitive data that is
presented on transactions, including their asset values, useful information in
11
payloads, or contract states. Confidentiality focuses on states rather than
entities. Notably, privacy covers both anonymity and confidentiality, where
an adversary cannot learn any useful knowledge from public transactions.
-Availability. Availability is used to measure the probability of a system run-
ning abnormally with the existence of potential failure. It indicates the sta-
bility of a blockchain system.
-Usability. Usability in the context of Web3 follows the same connotation as
the general computer system, which measures the satisfaction of a specific
user in a specific context when using a specific product.
Stakeholders. A major task in ATAM evaluation is to analyse a system’s ca-
pability for satisfying the stakeholder requirements [27]. The stakeholders in the
context of our Web3-based evaluation indicate the potential beneficial parties
in the system, being aligned with involved entities stated in Sec.2, rather than
the person holding assets in proof of stake (PoS) protocols [28]. Three main
stakeholders are identified as our evaluation target.
-Web3 user. A Web3 user is an entity that owns the data. Also, the user is
an entity to initiate transactions. The user should accordingly obtain a fair
proportion of rewards in the forms such as tokens, badger, etc.
-Service provider. An application owner uses blockchain to provide publicly
accessible services for users. The service provider may obtain revenues from
both sides: user payments and blockchain platform rewards.
-Blockchain maintainer. Blockchain maintainer provides a running environ-
ment for executing the smart contract and storing the blockchain data. The
main reward for maintainers comes from the cryptocurrency rewards.
4.2 Architecture Evaluation
Then, we start to evaluate each architectural design. We give a specific sce-
nario to narrow down the scope of the high-level description and analyze each
architecture regarding its properties and stakeholders.
Scenario. Non-Fungible Token [12] is a blockchain-based cryptocurrency that
allows proving the ownership. In our scenario, an NFT designer, Alice, wants to
sell her NFT to Bob by paying cryptocurrencies. Even though this application
focuses on a specific domain, we argue that the core requirements are likely to
stand across different Web3 applications. We apply different architectures to
this scenario, discussing both advantages/disadvantages and trade-offs, to help
interested readers consider their final decisions on Web3 architectures.
A1: An Ideal Web3 Architecture. An ideal Web3 infrastructure should elim-
inate all the middle man, without centralized databases or computation devices.
The blockchain provides a decentralized platform for data computation and stor-
age while bringing a native token-based economy. Under this architecture, Alice
12
uploads the raw NFT data, the price information, and NFT auxiliary informa-
tion to a blockchain-based smart contract through a wallet. Then, the blockchain
nodes reach an agreement on the data. Next, Bob sends a transaction to invoke
the contract for buying NFT. Finally, the NFT ownership transfers to Bob, and
Alice obtains the payments. The architecture puts all the computation and stor-
age on the blockchain, bringing security and availability for Web3 users. The
blockchain operates on a collection of decentralized nodes, which guarantees
the availability of on-chain data. Meanwhile, once blockchain nodes reach an
agreement for the received transactions, data becomes immutable. The related
architectural design in Sec.3 is Type1.
However, this architecture confronts some drawbacks, which mainly center around
being impractical when adapting to today’s Web3 infrastructure. Firstly, upload-
ing and replicating the data to all blockchain nodes in a peer-to-peer network
is time-consuming. Website interactions are slower where the back-ended state
transitions need confirmation and propagation throughout the network. Mean-
while, scalability is a long-term issue existing in blockchain systems due to in-
trinsic decentralization. The architecture is closely integrated with underpinned
blockchain systems, which, as a result, will face a similar issue. Moreover, the
computation and storage in the blockchain are expensive regarding monetary
costs. Worse still, the fees increase with the growing size of payload data.
Beyond that, this architecture raises problems about data privacy. On the one
hand, the code and state of smart contracts are transparent by default. Any
changes in the contracts are immediately visible, not just to the blockchain nodes
but also to anyone outside blockchains. The over-publicity issue hinders the de-
velopment of Web3 applications, making them hard to be adopted for privacy-
critical applications. On the other hand, users’ identities lack anonymity. With-
out an explicit approach taken to protect the user’s addresses, virtual addresses
on blockchains and physical identities are linkable with the help of analytical
tools and big data, even if users can use multiple pseudonyms.
A2: Agent-login Architecture. The ideal Web3 solutions, singly relying on
the original blockchain technology, confront the issues of poor scalability and
high cost. These shortcomings force the user to adopt the alternatively central-
ized computation or storage outside the blockchain. We gradually proceed from
layer to layer. Firstly, we address the congestion issue at the access layer. The
agent-login design provides a scalable approach to handling many transactions
by merging multiple executions. A group of users can access the blockchain plat-
form with one agent account. In particular, in our scenario, multiple NFT-selling
and buying transactions are bundled into one transaction. An agent anchors
this transaction to the blockchain later for the final agreement. Centralized ex-
changes, such as Coinbase [29] or Binance [30], act as agents similar to this
design. Many Web3 participants rely on custodial wallets for easy access and
management. This architecture can concurrently process transactions without
waiting for the blockchain’s confirmation, which brings high performance, good
scalability, and low cost. The related architectural designs include Type7-9 . The
13
main drawback of this solution is centralization, which may cause unnecessary
issues. For example, a malicious agent may hide the user’s transactions without
sending them to the blockchain. Worse still, such action is hard to trace due to
the lack of transparent evidence.
Table 1: Web Architecture Evaluation
Property Stakeholders
Architecture
Performance
Scalability
Gas Cost
Security
Anonymity
Confidentiality
Availability
Usability
Web3 User
Service Provider
BC Maintainer
A1,B1,C1Type1
A1, B1, C2/3Type2/3 + - -
A1,B2,C1Type4 + - -
A1,B2,C2/3Type5/6 ++ - - - -
A2,B1,C1Type7 + - -
A2,B1,C2/3Type8/9 ++ - - - -
A2,B2,C1Type10 ++ - - - -
A2,B2,C2/3Type11/12 +++ - - - - - -
Baseline, Property enhance, Property decrease; (Compared to Type1 )
A3: Hybrid Computation Architecture. This architecture shares the same
workflow with A1 and A2. As discussed, on-chain computations are both time-
consuming and money-consuming. Hybrid computation architecture solves this
issue by bringing TTP with high computation capabilities. These TTPs process
the data off-chain, and merely put a few computation tasks on the blockchain.
In our example, the on-chain computation tasks may cover the token payment
and NFT the ownership transfer. This design improves user usability because
off-chain computations have better performance and scalability. The related ar-
chitectural designs include Type4-6 . However, it still assumes that the TTP
who provides computation power is honest. A TTP in the real world may act
maliciously or fail to provide computations due to hidden interests or being
compromised. Again, such malicious computations are hard to be detected.
A4: Hybrid Storage Architecture. This architecture stores the raw data off-
chain while the data pointer is stored on the blockchain system. In our example,
only NFT identifiers and payment information are stored on-chain, while the
raw NFT data is stored off-chain. Correspondingly, in the real world, most NFT
artists rely on centralized platforms like OpenSea [20], or Solanart [31] to store
the raw NFT data. In this architecture, the choice of how to store off-chain
data and how to reveal the contents of their off-chain data are left to users,
which brings certain privacy and over-publicity issues. Meanwhile, the architec-
14
ture improves the performance and saves the cost. Due to these benefits, the
design has been widely adopted by some storage-intensive applications, such as
blockchain-based streaming media, such as Theta [32], Audius [33] and Livepeer
[34], which depends on the combination of on-chain storage and off-chain storage.
The related architectural designs include Type10-12 .
However, the solution is partially centralized with the risk of single point failures.
In particular, the off-chain storage is controlled by a TTP. Any services built
on top of this type can only be processed when a TTP is available, making the
data access to relevant information becomes a privilege. Fortunately, some stor-
age systems rely on replication to back up files for data integrity. For example,
Storj [35] utilizes the proof-of-redundancy mechanism, where every file is stored
in at least three locations to avoid files being destroyed. The system operates
on the Ethereum platform and stores the metadata in Satoshi format. Similarly,
Sia [36] is a distributed storage system that relies on storage proofs. These proofs
consist of a list of publicly verifiable root hashes from the submitted file and a
fraction of the original file, and users can verify them easily from on-chain data.
InterPlanetary File System (IPFS) [37], with more complicated proof mecha-
nisms, establishes a fully distributed peer-to-peer file system. Leveraging proof
of time/space, IPFS ensures data integrity from both time and space.
Statistical Results. The investigation tries to answer the question which ar-
chitectural design is the most prevailing? Existing Web3 projects are designed
based on different scenarios. Accordingly, their solutions target different compo-
nents. We expand our research from mature systems (ranked in the market) as
well as newly released whitepapers that claim to launch the projects. Based on
our investigation (inevitably select examples in a small corpus from the entire
project pool), we find that most of the teams adopt the Type1 (A1, B1, C1) de-
sign with a straightforward blockchain-based architecture. These projects either
put focuses on merely one functionality, such as setting connections from Web2
to Web3 [38], building decentralized identities [39,40], or implement the very ba-
sic infrastructure [41,42,43] that can better serve existing blockchain ecosystems.
Above reasons made us to select Type1 as the baseline. For other types, we can
observe that the options of external techniques are insufficient. For instance, if a
project aims to move the storage off-chain, the options are limited to IPFS [19],
which has been implemented for years with comprehensive instructions and user
guides. Affected by little attention to external technologies, other design types
(even the sum of the rest ones) only occupy a small share of the entire picture.
A Concise Summary. As shown in Tab.1, we provide detailed evaluations of
each architectural design. Adding auxiliary techniques (agent, off-chain compu-
tation, off-chain storage) will impact the properties from different sides. Posi-
tively, performance, scalability, gas cost, and usability can get improved at dif-
ferent levels depending on how many layers have been modified. This is because
external techniques can support much more volume of data and participating
parties. From the view of single users, they can obtain better services due to
faster transaction processing time and cheaper gas costs. In contrast, negatively,
15
security, anonymity, and availability have decreased, compared to the baseline
(Type1 ), due to the in-transparency of off-chain processing procedures. Here,
the singly applied agent at the client component will not affect the anonymity
because either personal address or custodian addresses are equivalent towards
adversaries. Similarly, confidentiality, security, and availability are majorly vul-
nerable at the computation and storage components, with little relation to the
access component. Finally, for stakeholders, we can observe that Web3 users can
obtain benefits that are consistent with previous reasons (better service). In op-
posite, both service providers and blockchain maintainers are disadvantageous
since the managing costs accordingly increase.
5 Extending to the Web3 Space
In this section, we extend our scope from architectures to the entire Web3 space.
We present the Web3 impacts to current markets. Then, we point out the promis-
ing research directions as well as potential barriers on the road.
5.1 Web3 Impacts
We first discuss the impacts of current Web3 projects. We approach it from two
sides. The one is to estimate their influenced market value, while the other one
is to see the scale of Web3 applications or services. We give details as follows.
Value Estimation. Supported by mainstream blockchain platforms, Web3 has
gained an unexpected breakout. Measuring the quantitative impact is impracti-
cal due to its generality and vague bounder. No statistical data can be directly
used by different institutions. But we still find indirect ways to reflect its influ-
ence. The first way is to investigate the usage of smart contract languages. This
is because building a Web3 DApp requires defining logic and functions (e.g.,
ownership/transfer/connect) through these languages due to its close relation
with user interfaces. The total value locked (TLV) by smart contract languages
reaches up to maximally 193.14 billion USD (as of Dec, 2021)1. However, these
languages only show a partial picture of the entire Web3 ecosystem, while many
competitive programming languages are proposed following the same targets (cf.
Appendix C). The real impact of Web3 will greatly outpace the stated data singly
observed from languages. Another way to reflect the popularity of Web3 appli-
cations is based on fundamental index methodology [44] that covers different
types of underlying valuation figures such as users’ paid fees on applications and
individual votes for proposals. Fees, in this track, are different from transaction
1The TLV of $251.877b is made up by $193.14b from Solidy (76.68%), $28.19b from
Rust (11.19%), $27.56b from Vyper 10.94%, $1.57b from Ride (0.62%), $1.07b from
Cairo (0.42%), $193.47m Bitcoin Script, $152.45m from C#, $3.69m from Python.
Among them, Solidity still occupies the largest proportion, indicating that Ethereum
and EVM-compatible platforms play the most important role in the Web3 area. [Data
source: https://defillama.com/languages].
16
fees paid to miners in Ethereum. Instead, they are used to represent the cost that
people are willing to spend on related decentralized services. For instance, users
have spent 253,015 USD (per 30 days) on Arweave [45] which is used to query
or store data on the network. It provides an estimator of a network’s value and
activities by an in-time track from the view of the demand-side, which provides
a distinguished way of investigating the status of the Web3 market.
Practical Development. Besides its high values, Web3 DApps have promoted
a rapid evolution towards a wide variety of scenarios. A promising paradigm
shift is to move the centralized authority from trusted third parties (TTP) to
distributed participants, which is also known as the Decentralized Autonomous
Organization (DAO). DAO removes the formal management roles (e.g., dele-
gated authorities) and physical entities (company/office) [46], by instead a suite
of contracts residing on the Ethereum blockchain. DAOs have been adopted with
many instances like Aragon [47] and MetaCartel [48]. Another significant shift
lies in their way of authenticating identities. Unlike asking for sensitive infor-
mation through methods of email+password or OAuth in traditional networks,
identities in Web3 applications are replaced by anonymous addresses, where the
web establisher cannot obtain any useful knowledge of their users. The design
further stimulates the widespread of decentralized identifiers (DIDs) [49]. Users
can fully control assets and metadata under their DIDs, and optionally release
them to service providers by personal preferences. Moreover, Web3 also instimu-
lates the prosperity of development tools and supplementary suites surrounding
blockchain technologies. These tools cover APIs, statistic indicators (The Graph
[50]), distributed storage (Sia [51], Arweave [45]), edge computing (Helium [52]),
etc. They fulfill the blank left from previously isolated components, enabling
them to seamlessly integrate together and well support each other. We further
provide a look into Web3 applications that are stated in Tab.3 in Appendix D.
Unfortunately, constrained by our bandwidth, we only select a small group of
Web3 projects to demonstrate its wide adoption in this report.
5.2 New Paradigm
We then discuss the forthcoming wave of Web3 that goes beyond the initial
use case like cryptocurrencies. Web3 can promote technical integration across
different domains, ranging from establishing decentralized self-governance orga-
nizations to facilitating the progress of DeFi and Metaverse.
Technical Integration. Web3 is a novel approach to delivering internet archi-
tecture in a decentralized way. As discussed, Web3 covers every layer of the web
architecture, from the front-end to the back-end. It indicates that technical inte-
gration may occur in multiple areas. For instance, Theta Labs’ [32] decentralized
video solution aims to stream video on Web3 via customized APIs. Audius [33]
is a music sharing platform with the target to decrease the dependency on a
record label. Radicle [53] is an open-source and distributed platform for code
collaboration. Arweave [45] and Sia [51], similar to IPFS [19], are to establish a
17
decentralized storage network that allows users to store data. Deeper Network
[54] intends to build a hardware-powered VPN ecosystem. All these attempts
to modify, integrate and improve the existing infrastructure from centralized
networks are promising, providing educational experiences on both success path
and failure cases for the following developers.
Distributed Autonomous Organization (DAO). Web3 will significantly
promote the development of current distributed autonomous organizations. Uti-
lizing a suite of smart contracts improves reliability, as no powerful authorities
can break the rules. DAOs run on a flattened hierarchy where each participant
has the right to vote on specific issues, similar to the way in a conventional execu-
tive board. All the processes, including decision-making, token issuing, option se-
lecting and voting, are transparent due to the on-chain settings. This means any-
one, either internally or externally, could audit the code, which greatly improves
accountability and reliability and avoids misdirecting usage of funds collected
from investors and users. To achieve decentralized governance, every project can
issue its specified votes (or tokens) that stakeholders can put their preferences
on. For instance, Yearn [55] allows users to participate in decision-making and
voting on proposals. Radicle [53], as a decentralized GitHub alternative, grants
stakeholders the right to manage the project. Similarly, many DeFi protocols,
including Uniswap [56], The Graph [50], SuperRare [57], and Audius [33], enable
ownership, participation, and governance via their issued tokens. All these votes
and tokens require a Web3 DApp to get interacted with different DAO partici-
pants. Also, long-existing DAOs, such as MakerDAO [58], have attracted many
developers contributing to crypto-space and Web3 ecosystems. In this sense,
DAO has extended the scope of decentralization, which was previously bound
to machines, to a broader area that involves human beings. The shift of how to
operate a digital organization would pose a great impact on the future world,
more than they are presented today.
Self Governance. Initially, software companies in Web1/Web2 obey the rule
of protecting data, with a simple aim to involve more users for growth. But
eventually, they have to start turning profits by selling or manipulating users’
data, such as training AI models to make better recommendations. Users have
no choice, facing the dilemma between privacy and convenience. In contrast,
individuals in the Web3 space can hold as much personal data, which is more
than ever before. Together with DiDs, users can freely browse the internet as
well as perceive their data without compromising its privacy. Many solutions,
like Ceramic [40] and IDX [39], replace traditional authentication by allowing
users to build self-sovereign identities. The Ethereum foundation makes much
more progress by drafting an RFP [59] for defining a formal specification. By
controlling data and assets, an individual can even earn profits through incentive
mechanisms. This is practically important for building a sustainable ecosystem
that encourages users to continuously contribute to the required infrastructure.
DeFi. This is an intuitive application of Web3 as all the assets held by users
are stored in their wallets. Performing financial-related activities, such as swap-
18
ping different tokens at DEXes, loaning coins from exchanges, buying/selling
crypto-assets from fiat to stablecoins, and designing derivatives (e.g., NFT [12],
contracts, securities [60], share, etc.), becomes common and easier for single users
since they do not need to register to any financial intermediaries. These activities
make up the core of today’s DeFi market [61], which also attracts tremendous
monetary investments involved in this field. Blockchain systems lay the foun-
dation of DeFi protocols, guaranteeing the normal operations of every piece of
logic, while Web3 paves the path to practical usage of these protocols, guiding
users to participate in the games. Users only need to act as they behave in cen-
tralized markets, rather than having hard research to understand the differences
between token standards or blockchain platforms.
Metaverse. Unlike DeFi or NFT specifying one specific direction, Metaverse
is a general term that involves numerous technologies, with ambiguous targets
to describe a virtual digital world for the incoming future. Intuitively, users
in Metaverse will first interact with its front-end and then connect with the
decentralized network supported by interoperable blockchain platforms. Web3
can cover all touchable applications that a user can reach, such as social networks,
search engines (Brave [62]), galleries (Opensea [20]), and marketplaces. A user
can fully control their digital assets, identities, and data, browsing any sector
of the Metaverse just like shopping at the store. Web3 can help to establish
such a front-end environment with a pretty easy one-step connection for the
participating users. In practice, many projects start to establish Metaverse from
different aspects like IoT [52], Games [11], Markets [57][20], etc.
5.3 Open Challenges
In this subsection, we point out the potential challenges from four folds, sepa-
rately from the views of user-level (application ), system-level (blockchain), mar-
kets (economy), and social organizations (legality).
Application. Decentralized applications built on blockchain-empowered sys-
tems are the first gate to individuals. Users have intuitive feelings towards these
applications. We abstract three aspects that may affect the user experience.
-Composability. Web3-based solutions cover a wide range, either from the
type of components, or the products for each component. Data transmitting
in different products are inevitably isolated. The way to make data reused
by different products at the same layer is urgently important. Standardized
APIs may address the problem to some extent, where at least many DApps
with similar designs can invoke the same APIs. Ceramic [40] tries to build ap-
plications with composable Web3 data and enable reusable data for multiple
scenarios. However, making most back-end components composable is still
a challenge. The barriers are whether composable components are compat-
ible with each other and whether data can flow seamlessly across different
components. For such reasons, the data need to be designed in the same
structure for the re-usage by these components.
19
-Accessibility. Web2 networks still cover most Internet users’ activities, in-
cluding social media, shopping, meeting, education, and payments. Users are
accustomed to and enjoy their services due to their super convenience and
easy accessibility. The lack of integration with modern web browsers and
mobile applications limits the wide adoption of Web3 to end-users. An indi-
vidual user will not change from the product that he used for a long to a new
one. How to decrease the migration cost is a challenge. Moreover, interacting
with Web3 applications also requires additional development, education, and
software/hardware. This becomes a huge recognition difficulty for users and
thus impedes its wide adoption.
-Data Recovery. The private key is the most important secret of users when
using blockchain platforms. The entire account of users relies on a credential
that is represented in the form of a complex string or a series of secret recov-
ery phrases. If a user loses his private key, he will never enter his account.
The account becomes a dead account with all data, including assets, being
locked. Methods to recover accounts, or at least the internal data, are an
urgent requirement for applications in the Web3 space. Moreover, applica-
tions that adopt off-chain storage require more strict verification for data
integrity because additional checks on whether the data in external stor-
age matches on-chain hash values are needed. The data recovery in external
pages is difficult due to the absence of traceability and accountability.
Blockchain. Blockchain systems are the most fundamental layer in each archi-
tectural design type, supporting upper-layer applications as well as connecting
underlying storage. Limitations in blockchain systems will significantly constrain
the development of the entire Web3 ecosystem.
-Scalability. Scalability represents the ability to process transactions along
with the increased scale of networks, which is reflected by the increase or
decrease in its performance. This is a long-term issue existing in blockchain
systems due to intrinsic decentralization. Every major step needs time, such
as data update, signature verification, etc. Among them, the consensus pro-
cess has the most impacts since the more mining nodes joined the network,
the more computational tasks were added. Web3 is closely integrated with
underpinned blockchain systems, which, as a result, will face the similar issue
as well. In most cases, transactions are slow on website interactions as state
transitions need confirmation and propagation throughout the network. Un-
derlying blockchain platforms, together with atop Web3 applications, require
conquering the challenges of accommodating rapid growth and the demands
of not compromising performance. Otherwise, users, with a high probability,
may have a poor experience such as extremely slow loading speed of websites.
-Interoperability. Web3 applications, in the foreseen future, will be de-
ployed on many blockchain platforms. This requires interoperable blockchain
technologies to facilitate smooth state transitions, either homogeneously or
heterogeneously. Creating interchangeable communication channels to con-
20
nect isolated decentralized ledgers is still a challenge for the development.
Current interoperability solutions, such as pegged sidechains [63], hash-locks
[64], and trusted relays [65], partially mitigate the problem: they enable
transactions to cross over chains within their specified ecosystems (e.g.,
Polkadot and their para-chain slots [66]), but cannot make nature trans-
actions operated across different ecosystems (Polkadot token on Ethereum).
A prevailing method is to create the wrapped token that anchors the origin
token with equal supplies as an alternative representation, such as the BTC
coin existing in Ethereum with the representation of WBTC (an ERC20
format [67]). However, this makes more and more representative tokens pro-
duced with no actual usage, increasing the waste and complexity.
-Contract Vulnerability. The security of smart contracts [68] will directly
affect its connected Web3 applications. Smart contracts contain the business
logic and operation specifications, which are key to the upper layer appli-
cations. Meanwhile, smart contracts can act as autonomous agents [69], de-
creasing the power of centralized service providers in combined protocols and
securing on-chain data from users. Vulnerabilities existing in smart contracts,
caused by design flaws or implementation errors, may result in thousands of
monetary loss. Examples include the integer underflow/overflow attacks [68],
DAO attacks [70][71] and Parity Multi-Sig Wallet attack [72]. Even worse,
the scripting nature of contract programming languages (cf. Appendix C)
and the non-updateable feature of smart contracts will significantly limit
the growth of Web3 applications.
Economy. As one of the major differences compared to Web1/Web2, users in the
Web3 space can automatically obtain rewards according to their contributions.
Users holding both digital assets and metadata in their wallets can freely trade
them to earn profits. But there are still many concerns about disparate incentive
mechanisms, operating costs, and technical debts.
-Incentive. Users adopt Web3 applications largely due to their considerable
potential revenues. They can earn extra profits by conducting activities in
Web3 networks: browsing websites (Basic Attention Token [62]), providing
online storage (Arweave [45]), playing games (Axie Infinity [11]), or selling
self-created products (Opensea [20]). Even for the current stage, developers
can obtain airdropped tokens from the project teams by testing their demos
on the testnet. However, as more and more users participate in the game,
the threshold of obtaining rewards becomes extremely high. Thus, designing
a positive incentive mechanism that can cover as many users is crucial to
attract new players joining the network.
-Cost. The high gas fee has already become a major hurdle in using Ethereum.
Sending an transaction will cost more than hundreds of dollars (executions on
smart contracts cost more). As a result, applications with complicated com-
putations cannot be deployed on-chain. This is the reason why many com-
petitive public blockchains can co-exist in the market. Moreover, blockchain
21
provides very constrained capabilities of storage, where many DApps can
only put a small portion of code on-chain. A potential solution is to adopt
more off-chain solutions [73] that can take over the workload from on-chain
to local servers without significantly breaking decentralization.
-Technical Debt. Many Web3 applications are designed in a limited way
to facilitate the entire software development cycle. This would cause many
costs of additional rework in the latter processes, which is denoted as the
technical debt [74]. A suitable approach that aims for a long-term proposal
can save many more costs because a bad design will accumulate interest
just as it is in monetary debt. Developing more in an inappropriate route
can improve the difficulty of making updates. The development of a project
falling into technical debt will reach a point where it is no longer possible
to implement the protocol improvements that align with its initial vision.
Developers should be sufficiently prudent when designing their products,
even for the initial proof-of-concept implementations.
Legality. Governments and official organizations have a lot of concerns due to
the rapid shift and huge change in the cryptocurrency world. Individual users are
eager to have their non-infringeable rights and are afraid of an anarchy state. It is
pretty difficult to reach the balance. Here, we stress the two discussed challenges
in the view of the social side.
-Governance. Since anyone can launch Web3 projects, an increased number
of applications will inevitably make the market segmented and unsuper-
vised. Traditional authorities such as official organizations and governments
become less influential than it is today. A large proportion of power is dis-
tributed to DAOs, which are made up of individuals. However, anyone who
uses Web3 can arbitrarily establish DAOs without strict authentication or
KYC (know your customer) steps. This causes instability in society [75].
For instance, (web3) tweets on decentralized social media platforms would
be uncensorable without facing the risk of being punished for spreading ru-
mors. Moreover, some illegal trades, such as porn or drugs, might be abused
in unsupervised networks as no explicitly compulsive laws can forbid them.
-Taxable Difficulty. The intrinsic property of Web3 is to give the incentive
back to users. Individuals who contribute more will obtain higher rewards.
Their contributions can be in any form, such as deposited stakes (PoS-based
chains [28]), activities (many projects airdrop tokens based on this), atten-
tions (e.g., BAT [62], Cirus [76]), or followers in traditional social media
(Twitter, Discord, Facebook). If most of the current network users move to
the Web3 world, collecting taxes from, at least, technology companies and
Internet practitioners becomes extremely difficult. Even worse, hiding assets
in accounts can help users directly make investments in financial markets
and earn profits. The governments can never know what has happened and
when this has occurred, nor for associated evidence used for taxes.
22
6 Closing Results
This section collects several popular questions that are frequently mentioned in
communities. We accordingly answer them with our investigated results.
What is Web3? Web3 is an umbrella term used to describe the next gener-
ation of internet services. It incorporates a wide range of components in the
computer infrastructure. We have, in this work, investigated in the wild solu-
tions and found several common design patterns. We decouple Web3 services
into three major components as in Sec.3. With this, an application/service,
in the context of our methodology, can be categorized into a Web3 space if at
least one component is decentralized. Based on the current view, an ideal Web3
application, operates all the services on-chain. However, this design can only
support services with lightweight computations and storage. Supplementary
services require external techniques.
What is the cornerstone of Web3? Following the previous discussion, a nat-
ural question is that: how to implement the decentralization of each component
for building Web3. We review the identified twelve architectural designs and
find an interesting result: every component centres around continuously oper-
ating blockchain platforms. The front-end wallet needs to connect with the on-
chain operation, while the external storage also requires on-chain information
for item searching and valid proofs. Therefore, the characterization through
access,computation and storage highly rely on the service of blockchain: ac-
cess to the blockchain, computation in the blockchain, and storage surrounding
blockchain. These integrated components can establish a Web3 service thanks
to the support from the blockchain. In this way, Web3 holds the property of
decentralization, obtaining benefits from the blockchain.
Is Web3 sufficiently decentralized? Web3 is not as decentralized as it ap-
pears to be [77], in which building a Web3 application still highly relies
on a small corpus of companies. For instance, centralized exchanges that
can trade cryptocurrencies are majorly controlled by Binance [30] and Coin-
base [29], while wallets are affected by MetaMask [78], NFT products [12]
by OpenSea [20], and stablecoins by Tether [79]. Meanwhile, a similar phe-
nomenon also occurs in DeFi markets: Uniswap [56] has the most Total Value
Locked (TVL); Dai [80] has dominated the decentralized stablecoins; Chain-
link [81] has greatly outcompeted others in the price oracle. Moreover, many
infrastructure-related blockchain companies that provide programming inter-
faces and development tools are concentred on a few companies like Alchemy
[42], Infura [82], and Ankr [43]. The high density of consolidation in the cryp-
tocurrency field will inevitably result in a partial monopoly, where only a small
group of oligarchs takes most of the resources.
Is Web3 secure? The answer would be “No”. Web3 services rely on a suite of
composable components to seamlessly work together. Problems in any single
component may lead to fail. Come back to our NFT example, the raw data
of the NFT may be erased due to the offline insecure storage by adversarial
23
attacks or compromised managers. This is merely an example from the storage
layer. In the real world, issues happening in other blockchain components
will as well result in severe problems, such as transaction congestion due to
poor scalability, slow confirmation caused by probabilistic consensus, or logical
loopholes in smart contracts. The security of Web3 is closely related to correct
operations of the entire system, which should be an all-level stack evaluation.
Has Web3 addressed privacy worries? Web3 services cannot protect user
privacy as it claimed. We give discussions from the website, which is the entry
of a user to Web3 applications and services. Web-side privacy issues cover a
set of design pitfalls and malicious attacks launched from the front-end, such
as browsers or mobile applications. In many cases, an adversary may act like
a normal operator in the system. For instance, the centralized authority can
play the role of a tracker with abilities to record Ethereum addresses over a
wide range of users, resulting in privacy violations with the help of its script
embedding techniques [83]. Once a user leaves evidence of using an address
at a DeFi site, the malicious browser, which also holds your identities, can
map to your physical entity. Besides, other traditional web attacks may also
threaten the Web3 sites due to the shared engine.
Fact and Fiction - Truth to be Told of Web3
Based on our investigations from both technical and economic perspec-
tives, Web3 cannot fully replace current web services. Instead, Web3 will
still highly rely on the existing Internet infrastructure, including program-
ming languages, communication protocols, agents, and storage. Fortu-
nately, Web3 has reshaped conventional finance markets and facilitated
individual self-governance. Users at least begin to pay much more atten-
tion to their digital assets covering both virtual data and cryptocurrencies.
7 Conclusion
Web3 is an emerging concept prevailing in the entire crypto-world. Applications
and services in the Web3 space, with non-custodial nature, allow users to control
their data and obtain rewards. However, no clear definitions of such a buzzword
have formed. In this tech report, we fill the gap by investigating a large corpus
of in the wild projects titled with Web3. We dig into this topic by decoupling
existing systems supporting blockchain-based Web3 services into separate core
components, and accordingly discussing related features and properties for each
potential combination. In total, we have identified twelve architectural design
types and evaluated them with profound discussions. Based on our study, we
present the new paradigms gained by Web3 and point out the design pitfalls. We
further provide our answers to several questions from communities. To the best
of our knowledge, this is the first research on Web3 from the view of blockchain.
24
References
1. Eric Glen Weyl, Puja Ohlhaver, and Vitalik Buterin. Decentralized society:
Finding web3’s soul. Available at SSRN: ht tps: // ss rn.c om/a bstr ac t=41 05 76 3 ,
2022.
2. The web3 report q3 2021 (consensys). ht tps: // co nsen sys.ne t/ re po rts/ we b3 -r
epor t- q3 -2 02 1/ , 2021.
3. Wood Gavin. Why we need web 3.0. ht tps: // gavofy or k.medi um.c om/w hy- we-
need -w eb- 3 -0 - 5 da4f 2bf9 5a b, 2022.
4. Dappradar. ht tps:// da pprada r.com/ , 2022.
5. Mathdapp. ht tps:// ma thda pp.s tore/ , 2022.
6. Avalanche network. ht tps: // www.av ax .net work / , 2022.
7. Solana network. ht tps: // sola na.com / , 2022.
8. Binance smart chain. ht tps: // bscs can.com/ , 2022.
9. Oasis network. ht tps: // oa sisp ro tocol.or g/ , 2022.
10. Jiahua Xu, Krzysztof Paruch, Simon Cousaert, and Yebo Feng. Sok: Decentralized
exchanges (dex) with automated market maker (amm) protocols. arXiv preprint
arXiv:2103.12732, 2021.
11. Axie infinity. http s:// ax iein finit y.co m/ , 2022.
12. Qin Wang, Rujia Li, Qi Wang, and Shiping Chen. Non-fungible token
(NFT): Overview, evaluation, opportunities and challenges. arXiv preprint
arXiv:2105.07447, 2021.
13. Zhuotao Liu, Yangxi Xiang, Jian Shi, Peng Gao, Haoyu Wang, Xusheng Xiao, Bi-
han Wen, Qi Li, and Yih-Chun Hu. Make web3. 0 connected. IEEE Transactions
on Dependable and Secure Computing (TDSC), 2021.
14. Satoshi Nakamoto. Bitcoin: A peer-to-peer electronic cash system. Decentralized
Business Review, page 21260, 2008.
15. Miguel Castro, Barbara Liskov, et al. Practical byzantine fault tolerance. In The
USENIX Symposium on Operating Systems Design and Implementation (OSDI),
volume 99, pages 173–186, 1999.
16. Juan Garay, Aggelos Kiayias, and Nikos Leonardos. The bitcoin backbone proto-
col: Analysis and applications. In Annual International Conference on the Theory
and Applications of Cryptographic Techniques (EUROCRYPT), pages 281–310.
Springer, 2015.
17. Juan Garay, Aggelos Kiayias, and Nikos Leonardos. The bitcoin backbone pro-
tocol with chains of variable difficulty. In Annual International Cryptology Con-
ference (CRYPTO), pages 291–323. Springer, 2017.
18. Qin Wang et al. Sok: Diving into DAG-based blockchain systems. arXiv preprint
arXiv:2012.06128, 2020.
19. Ipfs: Filecoin. ht tps: // filecoi n.io/ , 2022.
20. Opensea. ht tps:// op ense a.io/ , 2022.
21. Panagiotis Chatzigiannis, Foteini Baldimtsi, and Konstantinos Chalkias. Sok:
Blockchain light clients. Cryptology ePrint Archive, 2021.
22. Kostis Karantias. Sok: A taxonomy of cryptocurrency wallets. Cryptology ePrint
Archive, 2020.
23. Swarm. http s:// ww w.eths warm .org/ , 2022.
24. Rick Kazman, Mark Klein, Mario Barbacci, Tom Longstaff, Howard Lipson, and
Jeromy Carriere. The architecture tradeoff analysis method. In Proceedings.
Fourth IEEE International Conference on Engineering of Complex Computer Sys-
tems (cat. no. 98ex193), pages 68–78. IEEE, 1998.
25
25. Sukrit Kalra, Seep Goel, Mohan Dhawan, and Subodh Sharma. Zeus: analyz-
ing safety of smart contracts. In The Network and Distributed System Security
Symposium (NDSS), pages 1–12, 2018.
26. Rujia Li et al. Sok: Tee-assisted confidential smart contract. The 22nd Privacy
Enhancing Technologies Symposium (PETS), 3, 2022.
27. Len Bass, Paul Clements, and Rick Kazman. Software architecture in practice.
Addison-Wesley Professional, 2003.
28. Aggelos Kiayias, Alexander Russell, Bernardo David, and Roman Oliynykov.
Ouroboros: A provably secure proof-of-stake blockchain protocol. In Annual In-
ternational Cryptology Conference (CRYPTO), pages 357–388. Springer, 2017.
29. Coinbase. http s:// ww w.coinba se.com / , 2022.
30. Binance. ht tps:// ww w.bi nance.com/ , 2022.
31. Solana art. ht tps: // sola nart .io/ , 2022.
32. Theta network. ht tps: // www.th etat ok en.o rg / , 2022.
33. Audius. ht tps:// au dius .co/ , 2022.
34. Livepeer. ht tps:// li vepe er.o rg / , 2022.
35. Shawn Wilkinson, Tome Boshevski, Josh Brandoff, and Vitalik Buterin. Storj a
peer-to-peer cloud storage network. 2014.
36. David Vorick and Luke Champine. Sia: Simple decentralized storage. Retrieved
May, 8, 2014.
37. Filecoin: a decentralized storage network. http s:// fil ec oin.io / , 2022.
38. Web3auth. ht tps: // web3 au th.i o/ inde x.html , 2021.
39. Idx: Identity protocol for open applications. ht tps: // idx.xy z/ , 2022.
40. Ceramic network. ht tps: // ce ra mic.ne tw ork/ , 2022.
41. Truffle suite. http s:// tr uf fle suit e.co m/ , 2022.
42. Alchemy. ht tps:// ww w.al chem y.co m/ , 2022.
43. Ankr. ht tps:// ww w.ankr .com / , 2022.
44. The web3 index. ht tps: // web3 in de x.org/ , 2022.
45. Arweave. ht tps: // www.ar we ave.or g/ , 2022.
46. Robin Fritsch, Marino uller, and Roger Wattenhofer. Analyzing voting power in
decentralized governance: Who controls daos? arXiv preprint arXiv:2204.01176,
2022.
47. Aragon. ht tps:// ar ag on.org / , 2022.
48. Metacartel. ht tps:// ww w.meta cart el.o rg / , 2022.
49. Reed Drummond, Sporny Manu, Sabadello Markus, Longley Dave, and Allen
Christopher. Decentralized identifiers (DIDs) v1.0: Core architecture, data model,
and representations. ht tps: // www.w3 .o rg /T R/ did- co re / , 2021.
50. The graph network. ht tps: // thegra ph.com /en/ , 2022.
51. Sia network. ht tps: // sia.te ch/ , 2022.
52. Helium network. ht tps: // www.he lium .com / , 2022.
53. Radicle. ht tps:// radi cl e.xyz/ , 2022.
54. Deeper network: The decentralized gateway and infrastructure for web3.0. https:
// ww w.deep er.n etwo rk/ , 2022.
55. Yearn. ht tps: // year n.finan ce/ , 2022.
56. Uniswap. http s:// un iswa p.org/ , 2022.
57. Superrare. ht tps:// su perr are.co m/ , 2022.
58. Makerdao. http s:// ma kerdao .com / , 2022.
59. Request for proposals (rfp): Sign-in-with-ethereum. ht tps:// no tes.et here um.org
/@ dj rtwo /sig n-in -w it h-et hereum -RFP , 2022.
26
60. Hongyin Chen, Yukun Cheng, Xiaotie Deng, Wenhan Huang, and Linxuan Rong.
Absnft: Securitization and repurchase scheme for non-fungible tokens based on
game theoretical analysis. arXiv preprint arXiv:2202.02199, 2022.
61. Sam M Werner, Daniel Perez, Lewis Gudgeon, Ariah Klages-Mundt, Dominik
Harz, and William J Knottenbelt. Sok: Decentralized finance (defi). arXiv preprint
arXiv:2101.08778, 2021.
62. Brendan Eich. Brave browser. ht tps: // brav e.com/ , 2022.
63. Liping Deng, Huan Chen, Jing Zeng, and Liang-Jie Zhang. Research on cross-
chain technology based on sidechain and hash-locking. In International Confer-
ence on Edge Computing, pages 144–151. Springer, 2018.
64. Maurice Herlihy. Atomic cross-chain swaps. In Proceedings of the 2018 ACM Sym-
posium on Principles of Distributed Computing (PODC), pages 245–254, 2018.
65. Philipp Frauenthaler, Marten Sigwart, Christof Spanring, and Stefan Schulte.
Testimonium: A cost-efficient blockchain relay. arXiv preprint arXiv:2002.12837,
2020.
66. Gavin Wood. Polkadot: Vision for a heterogeneous multi-chain framework. White
Paper, 2016.
67. Wrapped bitcoin. ht tps: // wbtc.n etwo rk / , 2022.
68. Nicola Atzei, Massimo Bartoletti, and Tiziana Cimoli. A survey of attacks on
ethereum smart contracts (sok). In International Conference on Principles of
Security and Trust, pages 164–186. Springer, 2017.
69. Rujia Li et al. How do smart contracts benefit security protocols? arXiv preprint
arXiv:2202.08699, 2022.
70. Understanding the dao attack. ht tps: // www.coin desk .com /und er stan ding -d ao
-h ack- jour nali sts/ , 2016.
71. Understanding the dao attack. ht tps: // ww w.coinde sk .com /l ea rn /2 01 6/ 06 /2
5/ unde rsta nd ing-th e-da o-at tack / , 2016.
72. An in-depth look at the parity multisig bug. ht tps: // hack ingd istr ib uted.c om/2
01 7/ 07 /2 2/ de ep- div e- pari ty- bug / , 2016.
73. Lewis Gudgeon, Pedro Moreno-Sanchez, Stefanie Roos, Patrick McCorry, and
Arthur Gervais. Sok: Layer-two blockchain protocols. In International Conference
on Financial Cryptography and Data Security (FC), pages 201–226. Springer,
2020.
74. Wiki: Technical debt. http s:// ww w.wiki wand .com /en/ Tech nica l d ebt , 2022.
75. Aggelos Kiayias and Philip Lazos. Sok: Blockchain governance. arXiv preprint
arXiv:2201.07188, 2022.
76. Cirus. ht tps:// ci rusf ound at ion.com/ , 2022.
77. Kevin Werbach. The blockchain and the new architecture of trust. Mit Press,
2018.
78. Metamask. ht tps:// me tama sk.i o/ , 2022.
79. Tether. ht tps: // teth er.t o/ , 2022.
80. Dai. ht tps:// ma kerd ao.com / , 2022.
81. Chainlink. ht tps:// ch ain.li nk / , 2022.
82. Infura. ht tps:// in fura.i o/ , 2022.
83. Philipp Winter, Anna Harbluk Lorimer, Peter Snyder, and Benjamin Livshits.
What’s in your wallet? privacy and security issues in web 3.0. arXiv preprint
arXiv:2109.06836, 2021.
84. Md Sadek Ferdous, Farida Chowdhury, et al. In search of self-sovereign identity
leveraging blockchain technology. IEEE Access, 7:103059–103079, 2019.
85. Consensys: Blockchain use cases: Blockchain in digital identity. ht tps:// co nsen
sy s.net/ bl oc kchain -u se- cases /d igit al- ide ntit y/ #how dd iwor ks , 2022.
27
86. Hyperledger aries. http s:// ww w.hyperl edge r.org/ us e/ arie s, 2022.
87. Ontology network, ont id. http s:// ont.id / , 2022.
88. uport - veramo. ht tps: // veramo .io/ , 2022.
89. Gavin Wood et al. Ethereum: A secure decentralised generalised transaction
ledger. Ethereum project yellow paper, 151(2014):1–32, 2014.
90. Yonatan Sompolinsky and Aviv Zohar. Secure high-rate transaction processing in
bitcoin. In International Conference on Financial Cryptography and Data Security
(FC), pages 507–527. Springer, 2015.
91. Ragib Hasan, Zahid Anwar, William Yurcik, Larry Brumbaugh, and Roy Camp-
bell. A survey of peer-to-peer storage techniques for distributed file systems.
In International Conference on Information Technology: Coding and Computing
(ITCC), volume 2, pages 205–213. IEEE, 2005.
92. Nazanin Zahed Benisi, Mehdi Aminian, and Bahman Javadi. Blockchain-based
decentralized storage networks: A survey. Journal of Network and Computer
Applications, 162:102656, 2020.
93. Xiaotao Feng et al. Bug searching in smart contract. arXiv preprint
arXiv:1905.00799, 2019.
94. Burak Benligiray, Saˇsa Milic, and Heikki anttinen. Decentralized apis for web
3.0. ht tps:// ap i3 .org/ .
95. Wei-Meng Lee. Using the web3. js apis. In Beginning ethereum smart contracts
programming, pages 169–198. Springer, 2019.
96. Elli Androulaki, Artem Barger, Vita Bortnikov, Christian Cachin, Konstantinos
Christidis, Angelo De Caro, David Enyeart, Christopher Ferris, Gennady Lavent-
man, Yacov Manevich, et al. Hyperledger fabric: a distributed operating system
for permissioned blockchains. In Proceedings of the thirteenth EuroSys Conference
(EuroSys), pages 1–15, 2018.
97. Christian Gorenflo, Stephen Lee, Lukasz Golab, and Srinivasan Keshav. Fastfab-
ric: Scaling hyperledger fabric to 20 000 transactions per second. International
Journal of Network Management, 30(5):e2099, 2020.
98. The algorand virtual machine (avm) and teal. ht tps:// de velo pe r.algo ra nd.o rg
/d oc s/ get-de tail s/ dapp s/ av m/ teal /speci ficat ion/ , 2022.
99. Yossi Gilad, Rotem Hemo, Silvio Micali, Georgios Vlachos, et al. Algorand: Scal-
ing byzantine agreements for cryptocurrencies. In Proceedings of the 26th Sym-
posium on Operating Systems Principles (SOSP), pages 51–68, 2017.
100. The pact programming language. ht tps: // gith ub.c om/k aden a-io /pact , 2022.
101. Kadena. ht tps:// ka dena .io/ , 2022.
102. Dune network. ht tps: // dune .net work / , 2022.
103. Sui blockchain platform. ht tps: // docs .sui .io/ le arn/ ab out- sui , 2022.
104. Ssc: Stellar smart contracts. http s:// gi thub .com/s tel lar -d epre ca ted/ do cs/blo
b/ mast er /gui des/ wa lk throug hs/s tell ar -s mart -con tr acts.m d , 2022.
105. Solidity. ht tps: // docs .s olid it yl ang.or g/ en /v 0.8.13 / , 2022.
106. Ethereum for rust developers. ht tps: // gith ub.com /ste ll ar- dep recate d/ docs /b
lob/ ma ster /g uide s/ walk throug hs /ste llar -s mart -c ontr acts .md , 2022.
107. Ethereum for javascript developers. ht tps:// et here um.org /en/ de velo pe rs/d oc
s/ pr og rammin g-la ngua ges/ ja vasc ript / , 2022.
108. Yul docs. ht tps: // docs .sol idit yl ang.or g/ en/v 0.5.3/ yul.ht ml , 2022.
109. Vyper docs. ht tps: // vy per.read thed ocs.io/e n/ v0 .1 .0 -b et a.12 /ind ex .h tml ,
2022.
110. Marta Lokhava, Giuliano Losa, David Mazi`eres, Graydon Hoare, Nicolas Barry,
Eli Gafni, Jonathan Jove, Rafa l Malinowsky, and Jed McCaleb. Fast and secure
28
global payments with stellar. In Proceedings of the 27th ACM Symposium on
Operating Systems Principles (SOSP), pages 80–96, 2019.
111. Move docs. ht tps: // docs .sui .io/ bu il d/ move, 2022.
112. Diem. ht tps:// de velo pe rs.d iem.com/ do cs/t echnic al- paper s/ th e-di em-b lock
chai n-pa pe r/ , 2022.
113. Ocean protocol. http s:// oceanpro tocol.com/ , 2022.
114. Syndicate: Turn any wallet into a web3-native investing dao. ht tps:// sy nd icat e.
io/ , 2022.
115. Utopia: Collaborative payroll and expense management for daos. ht tps: // www.
ut opiala bs.c om/ , 2022.
116. Arbol. ht tps:// ww w.ar bolmar ke t.co m/ , 2022.
117. Etherisc. ht tps:// et heri sc.c om/ , 2022.
118. Royal. http s:// roya l.io/ , 2022.
119. Mirror: Create and connect your world on web3. http s:// mi rr or .x yz / , 2022.
120. Creaton. ht tps:// ap p.crea ton.io/# / , 2022.
121. Gari network. ht tps: // gari .net work / , 2022.
122. Gitcoin: Build and fund the open web together. ht tps: // gitc oin.co/ , 2022.
123. Linkdrop. ht tps:// li nk drop .io/ , 2022.
124. Cointraffic. http s:// coin traf fic.i o/ , 2022.
125. Manifold. ht tps:// ww w.mani fold .x yz / , 2022.
126. Clout.art. ht tps:// cl out.ar t/ , 2022.
127. nifty.ink. ht tps: // nift y.ink/ , 2022.
128. steem. ht tps:// st eem.com/ , 2022.
129. Akasha. http s:// ak asha .org/ , 2022.
130. Cyberconnect. ht tps:// cy bercon nect .me/ , 2022.
131. Coinvise. http s:// ww w.coinvi se.c o/ , 2022.
132. Rally. ht tps: // ra ll y.io/ , 2022.
133. Mynfteam. ht tps:// ww w.my nf .tea m/ , 2022.
134. Status. ht tps:// st atus .i m/ , 2022.
135. Violet. ht tps:// vi ol et.co/ , 2022.
136. Litentry. ht tps: // www.li tent ry.c om/ , 2022.
137. Spruceid. ht tps:// ww w.spru ceid .com/s pr ucei d, 2022.
138. Crypto.com. ht tps:// cr ypto .com / , 2022.
139. Zapper. ht tps:// za pper .f i/ , 2022.
140. Rainbow. http s:// rain bow.me/ , 2022.
141. Zerion. ht tps:// ze rion .io/ , 2022.
142. Haskell web3 documentation. ht tps:// hs -w eb3.read th ed ocs.io/e n/ late st/i ndex
.htm l , 2021.
143. Anchor protocol. ht tps: // docs.a ncho rproto co l.co m/ develo pers- ear n/ anch or
-e ar n-sdk , 2022.
144. Solana-web3.js. http s:// so lana -lab s.gith ub.i o/ sola na- web 3.js/ , 2022.
145. Civic. ht tps:// ww w.ci vic.com/ , 2022.
146. Flux protocol. http s:// ww w.fluxproto col.or g/ , 2022.
147. Polygon. http s:// poly gon.te chnolo gy/ , 2022.
148. zksync. ht tps:// zk sy nc.io/ , 2022.
149. Starkware. http s:// st arkw ar e.co /sta rkne t/ , 2022.
150. Zk-rollups. http s:// do cs.e thhu b.io/eth er eum-road ma p/ laye r-2-sc alin g/ zk- r
ol lups / , 2022.
151. Optimism. ht tps:// ww w.opti mi sm.i o/ , 2022.
152. Arbitrum. ht tps:// ar bi trum .io/ , 2022.
29
153. Optimistic rollups. ht tps: // ethe re um.o rg /en/ de velo pe rs/d ocs/ scal ing/ op ti
mi stic -r ollu ps/ , 2022.
154. Threadb. ht tps:// do cs.t ex tile .io/ th re ads/ , 2022.
155. Gundb. ht tps:// gu n.eco/ , 2022.
156. Bittorrent. http s:// ww w.bitt or re nt.c om/ , 2022.
157. Hardhat. ht tps:// ha rdha t.org/ , 2022.
158. Blockchain foundry. ht tps: // blockc hain foun dry.com/ , 2022.
159. Brownie. http s:// et h-br owni e.read thed oc s.io/en/ st able / , 2022.
160. Settlemint. http s:// ww w.sett lemi nt .com/ , 2022.
161. Deepdao. ht tps:// de epda o.io/o rg aniz at ions , 2022.
162. Snapshot. ht tps:// sn apsh ot.o rg /#/ , 2022.
163. tally. ht tps: // www.ta ll y.xyz/ , 2022.
164. Nansen. ht tps:// ww w.nans en.a i/ , 2022.
165. Token terminal. http s:// toke nter mi nal.com/ , 2022.
166. Messari. ht tps:// me ssar i.io/ , 2022.
167. The block data dashboard. ht tps: // www.th eblo ckcr ypto .com /dat a/ nf t- non-fu
ng ible- tok ens/ nf t-ov er view , 2022.
168. Neon evm. ht tps: // docs .neon-la bs.o rg/d oc s/ gett ing sta rted / , 2022.
169. Ens: Ethereum name service. ht tps: // ens.do main s/ , 2022.
170. Bonfida: Solana name service. ht tps: // nami ng.b onfid a.or g/ #/ , 2022.
171. Api3: The web api economy. ht tp s:// ap i3 .org/ , 2022.
172. Pocket network. http s:// ww w.po kt.n etwo rk/ , 2022.
173. Datahub: The web 3 gateway. ht tps: // data hub.fig ment .io/ , 2022.
174. Getblock: Superior node infrastructure for building dapps. ht tps:// ge tblo ck.i o/ ,
2022.
175. Moralis. ht tps:// mo ra lis.io / , 2022.
176. Quicknode. ht tps:// ww w.quic knod e.co m/ , 2022.
177. Figment learn. ht tps: // lear n.figme nt.i o/ , 2022.
Appendix A. Web3 Primitives
This section recalls several concepts used in Web3. We provide two main parts:
one for the web-related knowledge, including its meaning, evolution, and archi-
tecture, while the others for basic primitives used to build a Web3 service.
A.1 Web and Architecture
We first provide the common sense of so-called Web1/Web2, and abstract the
backbone of their architectural designs. Based on that, we provide comparisons
with Web3 from the perspective of its architecture.
Web1/Web2. The concepts of Web1/Web2 have become common knowledge
for most internet users. We conservatively show their core properties from a
high-level view during their evolution. The earliest version of the Web (Web1)
is featured by its static sites, which consist of components such as images and
text. Users access the targeted items by first finding a browser and then clicking
what has been presented on the page. In this sense, Web1 is deemed as the
read-only web. Web2 extends Web1 by importing more complex designs (e.g.
30
front/back-end architecture), enabling interactive actions from users by dynamic
HTML. Users can both read and write the content presented on sites, and can
also upload or download files stored in databases. Users have their customized
options of choosing which services are supported by providers (Facebook, Google,
Amazon, etc.). This directly paves the way for various applications that require
interactive web services, including marketplaces, user-generated content, and
social media platforms. However, these centralized service providers gradually
become the oligarch in their single fields because the big company has controlled
huge amounts of data from users, some of which are even privacy-sensitive, like
users’ passwords or financial history. The advent of Web3 mitigates such issues
by replacing the centralized back-end server with distributed ledgers. User can
hold their personal accounts (containing digital assets) safely rather than relying
on centralized banks. The services from each website, if interacting with users,
have to be authenticated through the way of connecting the wallet (cf. Sec.2).
Web Architecture. We briefly summarize the potential architectural layouts
and components of a web application. Building a typical web application relies on
a client-server model. The client means the ends browsed by users through com-
puters, or smartphones, while a web server serves the data and requests. Here, the
webserver architecture is relatively complex that covers many fundamental as-
pects, including application tiers, operating systems (Windows, Linux, Solaris),
platforms (.Net, LAMP), performance/quality of service (latency, throughput,
low memory utilization), and physical capacity (computing power, storage, and
memory). Receiving and responding to requests is the most basic action in suc-
cessfully performing a web application. Firstly, a user visits a website by in-
putting a URL in the browser on the front-end. The browser parses the URL
and sends the request to find the IP address via HTTPs. Then, the web server
catches the request and processes it following the business logic (also called do-
main logic and application logic) in the back-end. The business logic manages
the ways in which each piece of data is being accessed and determines the cor-
responding workflow, especially for each application. Last, the user receives the
response on the web page sent from back-end servers.
A.2 Fundamental Components
Then, we summarise the basic components that are used to establish Web3 from
the front user-end to the back server-end, covering light client (wallet), VM-
engined blockchains (computation) and decentralised storage systems (storage).
Besides, we also introduce a close concept decentralized authentication, which is
important for individuals who physically connect themselves with online virtual
identities (often in the form of anonymous addresses).
Decentralized Authentication. Different from traditional ways of authentica-
tion, decentralized authentication removes, or at least weakens the dependency
[69] of trusted third parties (TTPs) during the procedures of verification and
identifications. Each user accordingly has a unique identifier under the W3C
31
commendations for decentralized identifiers (DIDs) and verifiable credentials
(VC) [49]. DID can be identified by the DID’s controller, who might be a single
person or an organization (also known as the self-sovereign identity [84]). VCs
are the proofs that follow an open standard for digital credentials, such as a
passport or a license, or an ownership certificate of bank accounts. Each DID
will be associated with a series of attestations generated by paired DIDs, usually
in the form of VCs, to attest to its characteristics [85]. Blockchain, in such cases,
has two typical roles that either directly replace TTPs or assist existing ones
by recording and managing the issuers’ digital certificates [69]. Several proof-of-
concept projects have been proposed to highlight their targets to reshape DID
files, including the open-source platforms developed by Hyperfabric Aries [86],
Ontology [87], uPort (now rebranded as Veramo [88]), etc. Beyond that, if user
privacy is an essential requirement, more complex security-related primitives
(e.g., zero-knowledge proof, homomorphic encryption, or commitments) have to
be introduced in the scheme.
Light Client in Blockchain. The term light (equiv. lightweight)client shares
a similar meaning of its usage for both Web2 and Web3: it merely receives
the requests from users and forwards them to back-end servers (or blockchains)
without participating in any logic processes. A slight difference is that a client
in Web2 is often instantiated as a browser (covering both web browser, mobile
browser, or an App), whereas in the context of Web3 or blockchain [21], it is
typically represented as a wallet [22], which is supported by locally running
light nodes that connect to full nodes for information synchronization. A wallet
interacts with the online blockchain system by sending a transaction to the
Txpool (transaction pool) and broadcasting them to peers via the gossip protocol
[14]. Using such a client can pose more compatibility to resource-constrained
environments such as different hardware devices, as well as reduce the costs
of performing complex computations on-chain. The light client is an essential
component in building Web3 applications, as it is the first entry to access the
decentralized web environment. A shred of explicit evidence is that every Web3
page enforces users to connect the wallet (a button in the upper right position)
when users want to conduct interactive actions on this website.
VM-engined Blockchain. The most significant difference between Web3 and
previous web versions lies in the usage of blockchain. From the architecture
perspective, blockchain replaces the traditional centralized back-end servers with
distributed ledger systems like Ethereum [89]. A blockchain-based system takes
over the tasks of processing business logic and responding to users (under Web2
semantics). This depends on well-functioned on-chain virtual machine (VM) that
enables state transitions. VM, in the context of blockchain, can be equivalently
regarded as smart contracts that are automatically operated following the coded
rules. The rules contain the logic requested from clients. Smart contracts, in
this sense, play an essential role in enabling Web3 and DApps. Meanwhile, to
ensure consistency across distributed nodes, a consensus protocol is required in
each specific blockchain system. The consensus mechanisms solve the conflicts
32
and maintain the chain stability by initiating a set of predefined principles (the
longest-chain rule in PoW [14], the weightiest-chain rule in GHOST [90], and
more [18]). Further, all these nodes mutually communicate in a P2P network.
These components make up a typical blockchain architecture, which acts just
like a fully functional back-end server from an external view.
Distributed Storage towards Blockchain. File storage systems in distributed
networks are fundamental for sharing and storing sensitive content across dif-
ferent nodes. Two major ways of distributed storage are either increasing data
availability (replication) or reducing data loss (erasure code). Adopting a dis-
tributed file system can obtain benefits including fault tolerance, scalability,
availability, and performance [91]. In traditional ways, a lot of servers (on a
scale of hundreds/thousands) have to cooperate to execute tasks requested from
clients and applications, including service providers like Hadoop File System
(HDFS), CernVM File System (CVFMS), and Andrew File System (AFS). Even
though these providers deploy many machines in different areas, services are still
controlled by providers, resulting in partial centralization. Differing from them,
blockchain-based storage systems remove trusted central parties, which are the-
oretically more secure than centralized storage [92].
User Wallet Application Blockchain
Identity Creation
Login
Use
calculation/storage
token
LoopLoop
token reward/payment
Fig. 3: Workflow of A Web3 System
Blockchain storage mainly relies on redundant replications for data security in
case of loss. Storj [35] operates on Ethereum and replicates the metadata in
multiple locations. Sia [36], as a distributed storage system, uploads a list of
publicly verifiable root hashes of submitted files to get verification on-chain.
InterPlanetary File System (IPFS) [37], a peer-to-peer file system, leverage the
proof-of-spacetime and proof-of-replication to guarantee that (i) the data is being
33
stored during a specific duration of time and (ii) the data has been distributed
in multiple hardware, rather than a single physical storage location. Based on
such investigations, we abstract the distributed storage components, which can
take most of the workloads when on-chain capacity is not enough.
A3.Web3 Workflow
In a Web3 system (cf. Fig.3), user’s data is processed and stored in a decentral-
ized and community-driven network, using open protocols instead of centralized
TTP. An important feature of Web3 lies in its instant rewards, enabling users
to obtain a fair share of revenues when they contribute to the network.
Appendix B. Token Standards
Tokens in Web3 play an essential role in incentivizing users and developers. The
token standard, as the subsidiary of the smart contract standard, defines the
methods to create, deploy and issue new tokens. We have investigated existing
token standards from different blockchain platforms and summarised them in
Tab.2. Most token standards are issued through Ethereum, which has the biggest
and most mature on-chain virtual machine and smart contract. The standards
in the table are not separate, and many of them have close relations. As a result,
these standards set the baseline of the entire ecosystem, even for the token
standards in other competitive blockchain systems. For instance, Binance smart
chain and Avalanche follow very similar principles. We give a brief guide in Fig.4.
Table 2: Summary of Token Standards
Standard Date Platform Feature Application Notes
ERC20 2015 Ethereum Token API / Fungible Token Vote/ICO
ERC721 2018 Ethereum Non-Fungible Token Artwork/IP
ERC777 2018 Ethereum Token Approval Improving ERC20
ERC1155 2018 Ethereum Semi-Fungible Token Game Bundling of ERC20
ERC223 2017 Ethereum Token Recovery
ERC998 2018 Ethereum Comp osable Non-Fungible Token Game/Ownership
ERC1238 2018 Ethereum Non-Transferrable Non-Fungible Token Badge
ERC1594 2018 Ethereum Core Security Token Standard Securities (Financial)
ERC1400 2018 Ethereum Security Token Standard