Chapter 0 Introduction

Greg Maxwell and Gavin Andresen debated on reddit about the expansion path of Bitcoin. Maxwell used an analogy, saying that Bitcoin should be multi-layered, just like if you want to go to the moon, you need to use a multi-stage rocket. Gavin also used an analogy, saying that Bitcoin should be like a garden, where all kinds of flowers and plants can be planted, some die, some live, and compete freely. He also said that if a layer of the hierarchical structure like a rocket is wrong, the entire system will be in danger.

The post portal is here https://www.reddit.com/r/Bitcoin/comments/438hx0/a_trip_to_the_moon_requires_a_rocket_with/
It is particularly long, which shows how intense the discussion was.

This article discusses why I support Gavin's "garden theory".

Chapter 1 Let's first describe the "rocket theory" and the "garden theory" in detail

Maxwell saw that the number of Bitcoin users was growing, and the blocks were no longer sufficient to carry the growth of transaction volume. Therefore, he believed that Bitcoin needed to be expanded. The expansion idea he proposed was to first widen the main road through segregated witness, and then build a high-tech road on the main road - the lightning network. In addition, some racing tracks - side chain elements - should be built for people to race in various ways.

The entire Bitcoin network has evolved according to this idea into a network with the Bitcoin underlying protocol as the first layer, the Lightning Network and side chains as the second layer. Users trade and value transactions on the second layer, while the first layer provides computing power support to consolidate trust.

Gavin saw that the number of Bitcoin users was growing, and the blocks were no longer enough to carry the growth of transaction volume. Therefore, he believed that Bitcoin needed to be expanded, and the expansion idea he proposed was to directly increase the size of Bitcoin blocks. The Bitcoin network is like a garden. In the past, we had few customers, so everyone could play freely and plant some flowers and plants. Now that there are more customers, we will expand the garden and continue to allow everyone to plant trees and flowers freely without restrictions. Let the flowers, plants and trees grow freely, and you may grow a towering tree, and the death of some flower and grass seeds will not endanger the garden.

Maxwell believes that layered expansion is more efficient. He believes that Bitcoin needs to compete with payment institutions such as Visa, so the existing thinking should be changed. The Bitcoin network provides decentralized trust, but loses efficiency and capacity. Therefore, a more efficient payment system can be built on the basis of the Bitcoin network - the Lightning Network, which retains decentralized trust and improves payment efficiency and capacity.

Gavin believes that Maxwell's layered design requires each layer to be secure, and any problem in any layer will put the entire system in danger. A safer and more conservative approach should be adopted to first ensure the security and availability of the underlying construction, and then allow various innovations to develop freely on this safe soil. Whether such innovation is successful or not, it cannot affect the security of the underlying construction.

What these two masters said makes a lot of sense.

Chapter 2 A stable system should be a pyramid structure

It is very common for social structures to develop in a hierarchical manner. For example, the most familiar concept is the pyramid structure. A characteristic of a hierarchical structure is that the lower you go, the more adaptable it should be, and the more people (or activities) should be involved.

For example, the social classes in our society form a hierarchical structure. The bottom layer is composed of the majority of people who do manual labor. This layer is more adaptable to the natural environment. They can live without air conditioning, credit cards, and high technology. Going up one level, to the middle class, they need more external resources and division of labor, such as air conditioning, washing machines, banks, nannies, etc. The stability of this hierarchical structure is based on the fact that the number of people increases as you go down, such as a pyramid structure. In recent years, with the development of science and technology, the number of people at the bottom can also be reduced, and a large number of scientific and technological technologies can be used to solve things that could only be done before, but the entire social hierarchy still has to maintain a similar pyramid structure, called a spindle structure. This pyramid social hierarchy is stable. On the contrary, if there are too few people at the bottom, it will be troublesome. The entire social structure will collapse.

Let’s look at another layered design principle of computers.

Computers are designed to be used by users. The bottom layer is the hardware layer, followed by the operating system layer, and then the application layer with a very user-friendly user interface. This layering of computers also has obvious characteristics. The more efficient the bottom layer is, such as the extremely high computing efficiency of the CPU, while the higher the layer, the more efficiency needs to give way to humanization. For example, we can only use decimal to do calculations, rather than the efficient binary system.

If you want to design Bitcoin into a successful hierarchical system, you must have a solid underlying structure. The current bottom layer is the Bitcoin blockchain. How do you evaluate the solidity of this blockchain? I think it should be measured by computing power, number of users, and number of transactions. The greater the computing power, the more secure the Bitcoin blockchain is; the more users there are, the greater the value of Bitcoin; the more transactions there are, the more attention Bitcoin receives, and the more fees mining can earn, thus maintaining high computing power.

Some people have suggested that the system can be kept running without increasing the number of transactions, for example, by using the current block size plus the maximum transaction volume after the segregated witness, while increasing transaction fees to maintain computing power benefits. I personally think this is completely unfeasible. You can refer to my article last week, "Why I Hope Bitcoin Keeps Low Mining Fees".

If Maxwell wants to develop the Bitcoin system into a hierarchical structure, the first thing he should do is to strengthen the underlying structure of Bitcoin. What he should do now is to expand the capacity instead of hastily launching the Lightning Network.

Chapter 3: Hierarchical structure needs to consider too much compatibility and loses neatness

Let me first analyze two methods of file management in Windows, one is the hierarchical structure of folders, and the other is the library.

If you are a Windows user, you will find that Windows is now cultivating users' new file management habits if you pay close attention.

Here is a common file path in Windows, C:\ProgramData\Microsoft\User Account Pictures\Default Pictures. You can click on your folders one by one on the left side of the Windows window. If you have used your computer for many years, you may find that your files are in a mess and you often can't find where the files are. In order to deal with this complex file level management, I use a software Total Commander to manage Windows files. I used to think that Windows engineers were sick. In order to store a few desktop pictures, they created more than ten layers of folders. This must be something that idiots do.

But there is an additional structure called "Library" on the left side of the Windows window, as shown below.

Windows intends to let users manage files by categorizing similar things. If you are a Mac OS user, you must know this. It is called Finder in Mac. Mac does not have the concept of hard disk. All files are categorized and managed by Finder.

If you have used both Mac OS and Windows, you will definitely accept that it is neater to manage files by categorizing them in Finder than by managing them in layers in Windows. Now Windows is also starting to use libraries to train and guide users to use libraries to manage files.

Now let's go back to the section title. Why does the hierarchical structure make the system complex and not concise? This is because of the cost of compatibility. To innovate on an existing program, it must be backward compatible. For example, in order to be compatible with old applications, a program must allow the usage habits of users of the previous program. Programs that develop in a hierarchical manner must face the problem of common usage habits of new and old users.

Back to the expansion design of Bitcoin, if we use the hierarchical approach to develop and hang a lightning network on the main chain, then the lightning network must use a backward compatible method to be compatible with the user habits on the Bitcoin main chain. This will constrain the design of the lightning network.

If you have used an offchain wallet, you will find that the user experience of almost all offchain wallets is different from that of QT wallets, and is better and more efficient. Because offchain wallets do not need to consider compatibility with users on the Bitcoin main chain, they are completely creating a brand new user experience.

But can the Lightning Network completely ignore the feelings of QT wallet users like the offchain wallet? I think it won’t work if the Lightning Network is designed according to the layered expansion idea. Designers must design the Lightning Network to be an evolution of the Bitcoin main chain, so that users feel that the two are the same thing. Such a design will make the Lightning Network very complex and not concise.

Just like Windows is now training users to use libraries to replace hierarchical folder management, from Windows 7 to the current Windows 10, Windows dare not completely cancel the folder management because it has to be compatible with old users and old programs. I used to wonder why Windows didn't change and learn the file management system from OS. Now I know that the hierarchical folder design of Windows was originally designed, and then designed generation by generation, each generation had to consider the users and programs of the previous generation, resulting in an increasingly complex folder hierarchy.

Mac OS's Finder uses simple concepts to manage files, such as "documents", "downloads", "videos"... These concepts are common in everyone's mind, and you know what they are when you see the name. This design idea is built on a solid cultural foundation, and every user can understand it. The folder management mode of Windows requires users to remember the meaning of each folder. This design is based on the habits of previous old versions.

At this point, the logic is a bit complicated, so I will briefly summarize it. Building a system in layers will constrain the design of the next layer of software because it must be compatible with the behaviors of users and programs in the previous layer, which will make the next layer of software extremely complex.

If we have a solid foundation and then innovate on this basis, the innovation can be completely innovative without considering the user habits on this basis, that is another world.

If we build the Bitcoin blockchain solidly enough, with sufficient computing power, more and more users, and more and more transactions, then this foundation will become more and more solid. On this solid foundation, we can design new applications and develop new user needs, rather than snatching users from the main chain. Then the new design can be freer, more innovative, and simpler.

On the contrary, if the main chain block capacity is insufficient, we expand by layers, which is similar to the underground layer sharing users with the bottom layer (if we look at it from the perspective of transaction fees, it is to snatch users). This requires the design of the next layer to be fully compatible with the user habits and program habits of the bottom layer, and such a design will become more and more complicated. The more complicated it is, the more dangerous it is.

Chapter 4: Solving a single goal is suitable for using a hierarchical structure

I am a mechanical designer. On the premise of accurately understanding user needs, I often use hierarchical structures to meet needs. But the prerequisite is that the demand is clear, not scalable. For example, I have designed a large number of environmental protection equipment, such as a water purifier that filters water. A hierarchical structure is used. The first level of filtration only filters millimeter-level particles, the second level filters 0.1 mm debris, and the third level filters smaller particles. The advantage of this design is that it is a low-cost way to gradually meet needs. If only one-level filtration design is used, then a layer of the densest filter is needed, so that filtering large particles and small particles together will make this layer of filter net scrapped faster, and the cost of replacing the filter net is high. If multi-stage filtration is used, a three-layer filter net is used, and large particles are blocked by the first-level filter net, and the pressure of the subsequent filter net is small. The coarser the filter net, the cheaper it is. In this way, the overall cost of the design will be reduced.

The same is true of Maxwell's rocket theory. In order to overcome the earth's gravity, we need rockets as power, but the rocket itself has weight, and the ascent process also consumes energy. Therefore, a multi-stage rocket is used. First, the first-stage rocket pushes the satellite to a certain height, then the first-stage rocket is unloaded, and the second-stage rocket is ignited to push the satellite up again. This cost is lower than using a single-stage rocket. Because after the first-stage rocket is unloaded, the entire system does not need to bear the weight of the first-stage rocket itself.

However, this hierarchical structure is designed with clear design goals. For example, we have a clear requirement to filter out all particles of a certain size in the water, or we have a clear goal to send a satellite into Earth orbit. With a clear goal, the designer can break down the goal into small goals, and then divide the small goals into layers of design to solve them. If the goal is not clear, it is necessary to observe the development later to determine how to improve it, and then the design idea must be changed. And if the user needs are misunderstood at the beginning, the designed hierarchical structure must be completely overturned and redesigned. This is very risky.

The best way to solve this complex and ambiguous task is an evolutionary one.

Chapter 5 Solving complex tasks using biological growth methods

In the research of artificial intelligence, there is a research method called genetic algorithm. Simply put, it is to use a method that can try various solutions by itself, analyze the returned results, change its own algorithm, and then further explore the answer, and then further analyze... This is an analysis method that infinitely recursively searches for the target.

The genetic algorithm was proposed because the task to be solved is too complex. We humans don’t know what the goal is, so we have to let the algorithm try it by itself and then show us the results. This algorithm requires the construction of a solid foundation that can be controlled by humans. Although we humans don’t know what the final result will be, or even what each step of solving the problem is, we know that each answer is better than the previous one. We move closer to a better result little by little. In order to achieve this evolution in a better direction, we must provide the program with better computing resources before each progress, and we must never be stingy with this algorithm. We give enough of what we can provide, and this algorithm has the conditions to explore continuously.

We all have to admit that we don’t know the future of the Bitcoin ecosystem. Just like the designers of the Internet, they didn’t know that we would use the Internet to hail taxis and hook up for sex…

If we follow Maxwell's current idea, we will use a hierarchical structure to design Bitcoin's expansion, and lock the Bitcoin main chain at 1M, and achieve expansion through the lightning network and side chain elements. Regardless of whether this approach will successfully promote Bitcoin, this approach essentially limits innovation on the Bitcoin blockchain. When Bitcoin's 7 transactions per second are all occupied by side chains and lightning networks, if you want to use Bitcoin, you need a higher fee, which limits the possibility of Bitcoin evolving in a higher direction.

And what if, after building the Bitcoin network in layers, it is discovered that Maxwell’s understanding is wrong, then everything will be ruined, and Bitcoin cannot be restarted.

Ideally, we should put down our pride and admit that we don’t know the future of Bitcoin. What we can do is to give Bitcoin enough nutrients, like genetic algorithms, to expand the block capacity, and allow people to use Bitcoin more freely (rather than requiring high fees and long queue confirmation times). Let the world’s innovations freely innovate on this blockchain, compete on it, and then evolve to find better applications for Bitcoin.

If you understand it the way Gavin does, if you have enough garden soil, it doesn't matter if some fruit or dead trees grow on it. What matters is that the soil still has enough nutrients to support new great applications.

Chapter 6 Conclusion

We cannot afford a failed hierarchically designed rocket system; any mistake could cost us everything.

We need a soil for innovation to grow freely. Now the soil is not nutritious enough, so we need to expand it and turn the Bitcoin blockchain into a garden.

Thanks for reading.

If you find the article useful, please give me some Bitcoin to encourage me to continue writing.

Author: tan90d (Weibo @LightningHSL WeChat tan90d)
My BTC address: 14mhzjkJ71oMAMkKu3dy98dnUpkyQBHL1r
Source: Babbitt Information