Original Author: Solaire, YBB Capital

Introduction

"The only true voyage of discovery, the only fountain of Eternal Youth, would be not to visit strange lands but to possess other eyes."

This is a famous quote from the French writer Marcel Proust, taken from his novel "In Search of Lost Time". The Chinese meaning of this sentence is that the true journey of discovery lies not in seeking new continents, but in seeing things from a new perspective.

At the Polkadot Decoded conference on June 28th, Gavin Wood, the founder of Polkadot, used this quote as the core idea of his speech to present a new perspective on Polkadot. He suggests viewing Polkadot as a multi-core computer, focusing on providing more underlying resources, namely computing cores, for the blockchain, rather than just parallel chains and relay chains.

In this article, we will interpret the new paradigm of Polkadot in a simple and understandable way, based on Gavin Wood's latest speech.

Polkadot in the Traditional Paradigm

Before understanding Gavin Wood's new direction for Polkadot, let's review the current network structure of Polkadot and slot auctions.

The network structure of Polkadot consists of several main components:

1. Relay Chain: The heart of Polkadot, responsible for ensuring the security of the entire network, processing cross-chain transactions, and sharing security;

2. Parachains: Multiple independent chains connected to the relay chain. Each parachain has its own on-chain logic and functionality, which can be designed to perform any specific purpose, such as data storage, identity authentication, or financial transactions;

Polkadot allows communication with external blockchains (such as Bitcoin and Ethereum) to achieve cross-chain interoperability.

The structure can be understood as shown in the following diagram:

In the Polkadot network, different blockchains (referred to as parallel chains or Parachains) can connect to a unified relay chain. This relay chain is responsible for ensuring the security of the entire network and processing cross-chain transactions. This means that different parallel chains can communicate and interact with each other, achieving cross-chain interoperability.

In the Polkadot network, the resources of the relay chain are limited, which means that only a limited number of parallel chains can connect to the relay chain at the same time. These available connection spots are called "slots." To determine fairly which parallel chains can obtain these slots, Polkadot introduces a mechanism called "slot auction." In this auction, parallel chains wishing to obtain slots need to bid, and the parallel chain with the highest bid will get the slot. Bidding is done with the native token of Polkadot, DOT. Once a parallel chain wins the auction, it can use the slot for a certain period of time (such as two years). During this period, the parallel chain can perform its operations and interact with the relay chain and other parallel chains. When this period ends, the parallel chain needs to participate in the auction again to retain its slot, or transfer it to other bidders.

To put it simply, the Polkadot network is like a bunch of LEGO blocks. Each block is like a small network ("parallel chain") with its own tasks and functions. For example, some may be used specifically to record people's names, while others may be used to store game scores, and so on. These small networks can work independently and do their own things. However, these small networks sometimes need to communicate with each other, such as when one network needs to know the game score of another network. This is where a big network ("relay chain") comes in to help them communicate with each other. This big network is like a super LEGO connector that connects all the small LEGO blocks (small networks) together, allowing them to exchange information. The so-called parallel chain slot auction refers to the fact that the interfaces of this super LEGO connector are limited, and in order to fairly determine who can use these interfaces (i.e. slots), the interfaces need to be auctioned for lease.

Although this composition method is safer and more interoperable than Cosmos's IBC standard, the high threshold for slot auctions is a pressure for both the community and developers. This means that the Polkadot ecosystem is not as diverse as Cosmos and the main use cases for Polkadot tokens are only participating in slot auctions, governance, or security staking. In these use cases, DOT is only staked and not irretrievable, so DOT currently has almost no consumption scenarios and only offers the product form of parallel chain slot auctions. There are also issues with the economic system. Gavin Wood's latest speech suggests that we can look at Polkadot from a new perspective and how to solve the pain points of Polkadot in this way.

Polkadot Multi-Core Computer

As mentioned above, the current relay chain of Polkadot is like a super building block connector. Its primary responsibility is to ensure the security and interoperability of parallel chains. In this perspective, Polkadot is more like a blockchain hosting platform. In Gavin Wood's new perspective, Polkadot can be seen as a long-running multi-core computer. Developers can build applications on this computer, and users can use applications on this computer. In this computer, each core can run simultaneously and perform different tasks. A blockchain running on a core is a parallel chain, which runs continuously on a reserved core. This is similar to our computers, where different programs can run on different processors without affecting each other. In this new paradigm, the concept of the relay chain disappears, and it is replaced by cores and parallel chains.

Performance of the Multi-Core Computer

According to Wood's description, the Polkadot computer currently has about 50 cores running continuously, and they can operate in parallel. Based on benchmark tests and some optimization ideas by Wood, in the next few years, the number of cores is expected to reach several hundred (500-1000). For these cores, we can imagine them as multi-core CPUs, which have bandwidth (the total amount of data in and out of the core) and computing power. Currently, the performance bandwidth is 1 Mb/s, and the computing power scores 380 in the Geekbench 5 test (a popular cross-platform benchmarking tool that can test the performance of a computer's central processing unit (CPU) and graphics processing unit (GPU)), with a latency (the time interval between executing two consecutive work units) of 6 seconds. As hardware develops, both bandwidth and computing power will continue to improve.

Imagination in the New Paradigm

These cores are not only capable of running parallel chains. By changing perspectives and shifting our thinking paradigm, we can imagine a future where smart contracts run directly on the cores. Compared to running on smart contract chains like Ethereum, multi-core computers can perform better in terms of cost and computational power. They are highly versatile and, as a continuously running global computer, Polkadot has greater potential for imagination compared to chains.

From Blockchain to Block Space - Core Time

Let's first understand what cores and core time are with the help of the following image.

As shown in the image, there are five colored parallel blocks, each representing a core. Each individual block represents core time (the evolution from chain to space), while the colors on each row represent different parallel chains, such as blue parallel chains and green parallel chains. The image shows a total of 5 parallel chains, with each one utilizing a core. This is the current usage pattern in Polkadot, but cores can actually be used in multiple ways.

For example, parallel chains can be randomly distributed across any available cores, without affecting performance. Based on this feature, cores have multiple ways of utilization, which Wood refers to as exotic scheduling.

Range Partitioning

In the image, each core has 11 core time slots (assuming), which can be segmented into ranges. For instance, the first row of cores runs the orange parallel chain for six core time slots. When it no longer needs to process transactions, it can allocate the remaining five core time slots to the blue parallel chain. The fourth row demonstrates three parallel chains running on a single core, but it can also be more complex, with five or six parallel chains running on a single core.

Range Layering

Here, Wood calls it layering. Our understanding can be simplified: it is a way to change the ordering of core time usage. The first and second lines demonstrate the case of two parallel chains taking turns on one core, while the third line shows the case of 2/3 of the time running the light blue parallel chain and 1/3 of the time running the yellow parallel chain. The fourth line demonstrates the case of three parallel chains evenly sharing the time on one core.

Core compression

Core compression means processing multiple or verifying multiple blocks simultaneously on the same core. In other words, it is like a super-efficient factory that produces multiple products simultaneously on one production line, in order to improve production efficiency and reduce waiting time.

Multi-core allocation

The multi-core allocation is somewhat similar to a combination of elastic servers and fixed servers, or parallel computing of CPUs. It is used to deal with complex situations (Wood gives an example here: the case where the same paraID and the same task are assigned to multiple cores). The blue or orange parallel chains in the diagram have one core that is used for a long term and intermittent cores, in order to process two blocks in one time period. The pink combination includes intermittent core usage and additional allocated cores, which can be used to handle high transaction throughput.

Multi-chain on one core

Unlike layering, in the future, multi-chain on one core means putting two or three parallel chains on one core for full usage, in order to share the cost of one core.

Combination

You can combine all the above methods, just like assembling Lego bricks, to combine different forms of nuclei. Parallel chains with different needs can form countless uses, creating an extremely flexible and ubiquitous computing power.

Nuclear Time Economy under Polkadot

By understanding the use of nuclei, we also understand that nuclei have multiple elastic uses. According to the needs of different parallel chains, they can be freely combined, so the previously high-threshold slot auction on Polkadot can be transformed into a nucleus auction. This method is similar to selecting and configuring servers on Amazon Web Services today, where you can adjust the rental period and number of servers according to your needs. By flexibly choosing, you can better utilize the performance of Polkadot.

Based on this, Gavin Wood proposes two possible models: bulk purchase and instant purchase, which also leads to some new concepts: nuclear time assets and Axiom (broker).

Nuclear time assets

• Do not need to deploy or allocate directly

• Nuclear time is essentially homogeneous but can be divided into different non-homogeneous assets (comparable to NFT)

• Can be traded and assigned to one or more parallel chains

Axiom (broker)

• A proprietary broker parallel chain system

• The broker parallel chain system can purchase large blocks of nuclear time and divide them into smaller units

• The broker parallel chain can publicly trade these non-homogeneous assets on other parallel chains

• The purchased small units of nuclear time are consumed on the broker parallel chain to allow the owner to allocate computing resources on the core of Polkadot

Understanding the above two concepts, let's take a look at bulk procurement and real-time procurement. The form of bulk procurement is to sell once a month, selling a month's worth of core time assets at a unified price. The sales target is set at 75% of available core time, with possible fluctuations. The price will be adjusted up or down based on deviation from the target. Unrented cores will enter the real-time market, with special consideration for existing parallel chain tenants. Any remaining cores after bulk procurement will then enter the real-time market for sale through brokers. The goal is to achieve 100% utilization of core time. Small blocks of real-time core time can be used to increase transaction throughput and reduce latency (currently, parallel chains have a block every 12 seconds, but with multi-core allocation, it can be compressed to 6 seconds). Real-time core time can also be used for things like core contracts. For those who want to use a core in the long term, brokers will keep records of past purchasing prices for reference in the following month. Buyers can choose to purchase core time at the same or similar prices, or more core time. This allows for better budgeting of cyclical costs and risks.

As for the impact on existing parallel chains, the leases will remain in place and the pricing for core time procurement will be determined by governance. Wood suggests starting with a very low price to lower the barrier of entry, and existing tenants will have priority in purchasing, setting the main price at the minimum (floor price). This will ensure long-term availability. Wood also mentions that parallel chains have more flexible block generation times. To better understand this, I have moved Wood's use of core time to the middle part of the article. Now that we understand the flexible use of core time, it is easy to grasp the concept of more flexible block generation times.

Currently, Polkadot has a fixed block generation time of around 12 seconds, which may be further optimized to 6 seconds. In the future, by combining flexible block generation times with the use of core time, the following scenarios may arise:

• Multiple chains with one core: Multiple parallel chains share one core, generating a block every 12 seconds or 18 seconds. The benefit is cost sharing.

• Multi-core allocation: In cases requiring multitasking or high transaction throughput, parallel chains can automatically enter the real-time market to purchase additional core time.

• Core compression: Combining multiple parallel chain blocks into a core, allowing the same core to simultaneously process or validate multiple blocks. Compression can reduce latency but may increase bandwidth costs. Payment fees are required to open and close a block.

• Combination: Combination has multiple cases. Wood gives an example of two cores performing simultaneous operations, which can double the reduction in latency. For example, reducing from 12 seconds to 6 seconds, and from 6 seconds to 3 seconds. Essentially, this is a usage of multi-core allocation.

The Era of Core-Centricity

Many aspects of Polkadot have been controversial in the past. In the new paradigm described by Gavin Wood in the first part of his speech, the multi-core computer solves the problems that Polkadot used to have in a new way, such as fixed resource allocation and lease periods for slots. Cores provide choices for different parallel chains with different demands. The heavily criticized slot auction threshold can also be significantly reduced, bringing about ecosystem diversification. By using different gameplay composed of essential assets such as cutting core time, it can bring more vitality to the Polkadot token and its economic system. The different uses of cores and the multi-core computer generated by the combination give us a considerable space for imagination. Perhaps all the disputes stem from looking at the problem from only one perspective. In fact, some problems can be solved by simply changing the perspective. Gavin Wood has given a perfect demonstration. We look forward to the new era of Polkadot centered on cores.