Summit Review丨Wu Ming, Chief Technology Officer of Shanghai Shutu Blockchain Research Institute: Towards the large-scale adoption of Web3 - the perspective of user experience and infrastructure construction

Article published under Wanxiang Blockchain stream

Original article link:

Source link:

Machine translated English PDF:

Machine translated English text:
Summit Review丨Wu Ming, Chief Technology Officer of Shanghai
Shutu Blockchain Research Institute: Towards the mass adoption of
Web3 - the perspective of user experience and infrastructure

This article is compiled based on the shorthand of Mr. Wu Ming, Chief Technology
Officer of Shanghai Shutu Blockchain Research Institute, at the 9th Global Blockchain
Summit. There are slight deletions that do not affect the original meaning. It only
represents the views of the guests and does not represent the views of Wanxiang
Block. chain position.

Hello everyone, the topic I want to share today is mass adoption of Web3.0. I will discuss it
from two aspects: user experience and infrastructure construction. I will also share with you
Conflux’s thinking, progress and future layout in this direction.

Wanxiang Blockchain 14
Wanxiang Blockchain 2023-10-09 23:26 Posted onShanghai

First, let’s review the development history of blockchain technology and Web3.0
technology. Blockchain technology has gone through a very long development process. The basic
theories of blockchain in the 1980s, namely public keys and distributed systems, have become
increasingly mature. In the next twenty or thirty years, chain structures, multi-party
supervision evidence storage systems, anti-Sibyl attacks, and distributed electronics and other
technologies have emerged. In 2009, due to the development of the Internet and computing power,
all technologies were integrated together, eventually forming a blockchain system that could be
implemented - the Bitcoin network. In 2009, the emergence of smart contracts gave the
blockchain programming capabilities and ushered in the prosperity of decentralized
applications. At that time, everyone began to realize and pay attention to the performance of
the blockchain, because performance would become a bottleneck.
By 2018, there were a series of research teams designing high-performance blockchain systems to
solve performance problems, and Conflux was one of the representatives. Nowadays, the future
development trend of blockchain is mainly interface standardization, functional modularization
and user-friendly experience, as well as scale expansion/These are all driven by Web3.0 mass
adoption. Conflux has also done some work. For example, we led the formulation of the IEEE
P3217 application and system interface standard, and jointly launched the BSIM card with China
Telecom, thus lowering the threshold for users to enter Web3.0. At the same time, we are
constantly improving the system’s expansion capabilities. With the emergence of blockchain
technology, one of the key factors in the development of the Internet from Web2.0 to Web3.0 is
the information distribution and benefit distribution model. In the era of Web 1.0, efficient
information distribution was achieved and it was a read-only Internet. Web 2.0 benefits from
the centralized large-scale distributed system, which greatly improves the capabilities of the
Internet’s information processing platform. This processing capability also promotes the antiinternationalization of user-generated data models, making it a readable and writable Internet.

In Web3.0, the emergence of blockchain allows us to construct an information processing platform
that does not rely on centralized companies, so that users have autonomy over the data they
create, and at the same time create digital identities on it that do not rely on centralized
entities. , digital assets, digital rights and digital contracts, thereby realizing open
economic activities in the digital world and improving its efficiency. Therefore, the emergence
of Web3.0 is spurred by blockchain technology.
There is a big gap between the scale of Web3.0 users and the scale of Web1.0 users. Assuming
that starting from when the number of users reaches 5 million, in the next eight years, the
number of Web2.0 users will grow to 400 million, but the number of Web3.0 users has not yet
exceeded 100 million. And it can be seen that after 2018, the user scale of Web3.0 has
encountered a bottleneck. There are many bottlenecks in mass adoption of Web3.0. First of all,
the user experience of Web3.0 is different from that of Web2.0, which brings a high threshold
for Web2.0 users. Secondly, Web3.0 application types are not broad enough. Currently, the most
mainstream applications are Defi and GameFi. These applications are applied to pure assets and
are more likely to breed Ponzi financial games. There is a very good narrative now, which is
RWA, which maps traditional financial assets onto the chain and improves the efficiency of its
economic activities through reasonable tokenization. I personally think that RWA will become
the next application growth point of Web3.0.

Another issue is compliance supervision, because the decentralized nature of Web3.0 brings great
difficulties to regulatory compliance, which is also a factor that hinders the mass adoption of
Web3.0. Another factor is the interoperability of multi-chains. The existence of multi-chains
is inevitable because different chains have different characteristics, different ecologies,
different smart contract development environments, different asset deposits, and different
performance and security. When users use Web3.0 applications, they will inevitably use
different chains at different time periods. This gives users the feeling that Web3.0 APPs are
very fragmented and fragmented. Without a smooth, safe and reliable cross-chain infrastructure,
it will also cause big problems for the mass adoption of Web3.0.

These are application problems I mentioned above, and I believe these application problems will
be well solved in the future. Once this time comes, will our system be able to support a large
number of Web2.0 users entering the Web3.0 world? At this time, are the system’s capacity and
expansion capabilities ready to handle such a large number of users? Therefore, the expansion
capability of the system is also an issue that we have been paying close attention to. Due to
time constraints, in this speech, I will mainly introduce the work done by Conflux in terms of
user experience and system scalability.

The user experience of Web2.0 is very simple. Users only need to interact with a Web2.0
service. In the Web3.0 world, a user has to interact with different Web3.0 components and
chains. Every time you interact, you need to use the private key to sign and authorize the
operation. However, it is very difficult to manage the private key to be both safe and
convenient. Web3.0 should give users autonomy, which means that users must have management
responsibility for their Web3.0 accounts. A Web3.0 account is a private key. The private key is
an irregular combination of numbers and letters, which is difficult to remember. So the usual
approach is to use a mnemonic, which is 24 English words. It is also difficult to remember, but
it also requires a sense of security. We cannot take screenshots, save to files, or upload to
the Internet. You can only copy it, and you cannot let others see it when copying it. Once the
mnemonic phrase and private key are lost, the assets will never be recovered. This matter puts
a very heavy psychological burden on users. 90% of users forget the mnemonic phrase. This is a
very complicated Web3.0 user to manage. It requires multiple uses to deal with Web3.0, and when
the user signs, , many users don’t know what they are signing. Therefore, the Web3.0 account
management system seems to give users greater autonomy, but in fact it is neither safe nor
convenient. We are thinking about how to improve the account system of Web3.0. You can refer to
the account iteration of Web2.0.

As early as the software era, Web2.0 users only needed to log in with a password. In the era of
web services, users can log in to the platform using their email addresses. In the era of
mobile services, users can log in to the platform using mobile phones and SMS verification
codes. The user experience is becoming more and more convenient. When it comes to Web3.0, some
projects in the industry use Web3.0 wallets or platforms to host user private keys to give users
the same experience as Web2.0. However, this approach introduces a centralized mechanism and
brings new Security Question. In response to such problems, Conflux and China Telecom jointly
launched the blockchain-based BSIM card, which embodies the logic of the hardware wallet in the
SIM card. After inserting the card into the mobile phone, users naturally gain the convenience,
convenience and convenience of the software wallet. The security of a hardware wallet, because
it is decentralized, does not introduce the risk of centralization.

At the same time, the SIM card can naturally bind the mobile phone number and the private key
account. On the one hand, it is regulatory friendly, and on the other hand, it makes it easier
for users to participate in Web3.0 applications directly through the mobile phone number. We
believe that the logic of BSIM card is a big step forward for mass adoption of Web3.0.
The blockchain system is a complex system composed of many components, including consensus
network, execution and storage. The impact of these components on the performance of the entire
system follows the wooden barrel principle, that is, the system capability is determined by the
shortest board. When the Conflux project was first created, the consensus system was the
bottleneck of the entire blockchain system. Let’s start with consensus. Classic consensus
algorithms rely on the basic structure of a chain. After a block is generated, there is a huge
delay in broadcasting on the blockchain network. If we need to maintain the ledger into a chain
structure, we must reduce the block production rate, so the system throughput is greatly
reduced. If we want to improve the system’s block production rate, a forked ledger structure
will inevitably appear, and then the consensus algorithm must be adjusted. Therefore, Conflux
proposed an innovative algorithm based on the tree graph ledger structure, which enables nodes
in the entire network to reach consensus on the logical order of blocks in the tree graph
ledger, so that without sacrificing security and decentralization, The system’s block production
rate is greatly increased, thereby improving the system’s throughput.

When the consensus bottleneck is removed, the network will become the next platform. One of the
bottlenecks of the network is network bandwidth utilization, that is, how many transactions the
network system can convert per second, and the other is the delay in transactions reaching the
entire network. Let’s first look at the utilization of network bandwidth. We have a job called
Shrec. The Shrec job is a transaction forwarding protocol. Its core idea is to solve a problem
when doing transactions in the network, which is how to avoid a node receiving duplicate
transactions multiple times. Generally, when a node forwards a transaction to another node, it
first sends a transaction ID and asks whether this ID has been received before. If it has not
been received, I will send the complete ID. This involves how to set and structure the ID. If
the ID is too long, repeated forwarding of the ID will also put a lot of pressure on the network
bandwidth. If the ID is too short, it will cause conflicts, that is, different transactions are
mapped to the same ID. Once this occurs, many transactions cannot be forwarded well. Therefore,
Shrec’s job is to innovate and invent a transaction ID construction method, so that we can well
balance the ID length and conflict rate, thereby maximizing the use of network bandwidth.
The second job is how to reduce the delay in transactions reaching the entire network. We use a
distributed automatic learning method to construct virtual coordinates in the regional space for
each node in the network. The distance expressed by the virtual coordinates between different
coordinates describes the delay in transaction forwarding between the two nodes. With this
information, the nodes in the entire network are aggregated and classified, and an optimal path
for transaction forwarding is found.

After the network bottleneck is resolved, the next step is execution. Execution is partly
calculation and partly storage access. For computing, it is easy to think of parallel
processing, but Conflux has not yet implemented a parallelized virtual machine, but we have done
preliminary research. At the application layer, the application itself has many inherent
conflicts. In this case, even if there is a virtual machine that can be scheduled and processed
in parallel, the parallelism of the application is not fully utilized because the application
does not have much parallelism. For example, we analyzed the trace of Ethereum and found that
the parallelism degree of transactions in most blocks does not exceed 4. The reasons for this
problem are one is the airdrop of Token, and the other is Defi and GameFi. The operation that
causes this phenomenon is called a global counter. There is a global variable, and everyone adds
1 to it. We have come up with a solution for this matter. One is to use some methods that
change the programming paradigm, such as airdrops. We hope to distribute it to multiple senders
first, and then send it to users from multiple senders respectively, so that the degree of
parallelism will be higher. For the global technology device, a sharding technology device is
used. Different shards are independent technologies, and the value of the entire global
technology device is the sum of all shard values. We then propose a better way to exchange the
bytecode of the operation. We hope to pass exchangeable attributes from instructions to the
virtual machine through exchangeable operation bytecode, such as exchangeable addition
operations, so that the virtual machine knows how to parallelize exchangeable operations. For
example, if you have two parallel transactions, they add 1 to the same variable, but there is no
conflict when submitting, because the +1 operation can be exchanged. In the end, as long as the
+1 operation is written as +2, there will be no problem with this transaction.
These solutions all require modifying the user’s programming model, which raises the question of
why developers should use a new programming paradigm to do this, because an incentive mechanism
is needed. Once incentive mechanisms are introduced into parallel scheduling, parallel
scheduling will bring new challenges. Because we not only require the result of scheduling
execution to be certain, but also the Abort behavior of a certain transaction in the middle,
because due to the incentive mechanism, some accounts will be punished and the Token will
change, and this status will also have an impact. That is to say, abort will affect the status,
so the abort must also be determined, so we innovatively designed a parallel scheduling
algorithm that can determine the abort.

Finally, there is the storage issue. The storage system of the blockchain is a verifiable
storage system. The usual approach is Merkle Tree. Merkle Tree will bring about read and write
amplification of storage access, log(n ) read and write amplification. That is, when you
update a value, all nodes from the leaf node to the root must be updated. There are some new
verifiable structures in academia, such as AMT. AMT has a good feature. When you update a value,
the update of the commitment value is O(1). When you apply such a theoretically good algorithm
to an actual storage system, you will face many challenges. For example, the advantage in
complexity is not necessarily fast to implement. Because every operation of AMT involves very
slow cryptographic elliptic curve operations. Second, the capacity of AMT will bring some precalculated cryptographic parameters and construct them, and the scale of this cryptographic
parameter is related to the number of sets.

If we want to update a value in AMT, such as updating from a to a’, the commitment update will
be updated. We noticed that there is a’-a multiplied by Gi, which is a very complicated
elliptic curve multiplication. If a’-a multiplied by Gi equals 1, this multiplication can be
removed, because multiplying by 1 equals nothing. take.

In response to this observation, we came up with an ingenious design, that is to say, because in
the blockchain storage system, this is a multi-version storage system, which can prove that each
value appears in the latest version, and What is the latest version. That is to say, every
value update will be accompanied by a version update, and version updates are all +1
operations. In this way, we can store the version in AMT, and AMT can prove what the latest
version of a Key is. We store the real Key value in a set of append-only merkle trees. After
each block is processed, this merkle tree constructs a merkle tree at once from all the values
updated in the block. . This construction process is linear, without amplification. Through
this method, we removed the AMT elliptic curve.

The second is the non-scalability of AMT. To address this problem, we designed a multi-level
AMT structure, dividing a Key into multiple segments, and different segments are indexed to
different AMTs. For example, 10, 01 and 11 are divided into three segments - 10, 01 and 11,
which are indexed into AMT levels 1, 2 and 3 respectively. At the third level, the version
number corresponding to this key can be indexed. In this way, we have built a complete
verifiable storage system, which we call LVMT. Our experimental results show that it brings
nearly 3 times performance improvement compared to a merkle tree-based storage system.
Finally, I will briefly introduce the ecological situation of Conflux. Conflux’s ecology is
divided into two categories, one for mainland users and one for global users. For mainland
users, it mainly focuses on compliance applications of digital collections and metaverse. There
will be some Defi, NFT and GameFi for global users. Facing the global ecosystem relies on the
injection of large-scale financial assets such as stable coins. Now as Hong Kong embraces Web
3.0, all walks of life are calling on the Hong Kong government and companies to issue their own
stable coins. We are also communicating and cooperating with these related stablecoin issuance
projects. In addition, I am personally optimistic about RWA. I personally think that RWA is
definitely a catalyst for mass adoption of Web3.0. We are also currently working closely with
some good RWA-related projects and look forward to a large number of very high-quality RWA
applications appearing in Conflux in the near future.
That’s all my sharing, thank you all.

1 Like

Thanks for sharing with the community.

1 Like