Scaling Bitcoin and what some will do to stop this
We at nChain have been doing research into many aspects of Bitcoin. One of these is scaling. A consequence of this includes the following paper:
“Moore’s Law, Innovation, and the Scaling of Bitcoin” https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3065857
In this research, we tested the propagation of transactions and blocks over the Bitcoin network.
In this, I am going to list some of the objections to this paper and research. What we are finding is that there is a lot of pessimism and open condemnation of the idea that Bitcoin can scale. To put it simply, many academics seem to hate Bitcoin. As a result, let us pull apart some of the criticisms publicly. We start with the first.
Defining sustainability to simply meet the processing rate of credit card companies is too narrow of a definition and fails to address the energy, infrastructure cost associated with running bitcoin mining farms. As the mining difficulty level increases, this cost which is already unsustainable, will only going to increase. It is a fallacy to assume that new process/device technologies or software breakthroughs may somehow solve the problem — because any gain in processing speed will be automatically offset by the increased difficulty level. The energy efficiency of bitcoin mining is measured in GHash/watt and suggesting off the shelf general purpose CPU or GPU (such as Xeon Phi or CUDA on a Nvidia GPU) can beat the efficiency of custom application specific processors (ASP) is incorrect. The paper fails to justify why public bitcoin is the right technology instead of private application specific blockchain technology that can achieve the same goals at orders of magnitude lower energy requirement. Especially, with the consolidation of bitcoin mining nodes to fewer and fewer players, the argument of open and decentralized bitcoin is already falling apart. Claim that Moore’s law will survive for next 600 years and hence predicting Bitcoin mining will survive up to year 2600 seems to be far fetched, to put it mildly.
In our paper, we did not define the ideal as Visa, rather, we state on page 13 that, “This is more than 1,000 times the processing capacity of the existing VISA network.”
This is the classical argument that is promoted by Core and their followers. The result is that corporations are bad. Bitcoin was and cannot be a mesh, it was a complete graph and works when many high powered machines communicate. The issue always comes to Bitcoin is too inefficient, it uses too much power.
“The paper fails to address the real issue of massive energy requirements for running bitcoin mining farms.”
The academics then claim of the paper:
“There is hope expressed in algorithm and software breakthroughs, but no concrete example is shown how this will be applied.”
Which demonstrates they did not read it. We display how we scale more than 1 million times the existing transaction limit. This is not the issue, what is the issue is that we did not need any boondoggle changes to Bitcoin. It works and it works without any new protocols or changes. It works without segwit or Lightning. People seem to hate this.
It becomes even clearer what the agenda is later in the reviews.
“The paper reads like a one-sided opinion piece in the Bitcoin scaling debate. It uses the original Bitcoin whitepaper and comments in Bitcoin’s original source code to derive something like “Satoshi’s interpretation of how Bitcoin should work”, which is an ideological interpretation by the author’s rather than a scientific argument (the paper does not even try to interpret these statements in the context of that time).”
In fact or point, they clearly reject the entire need to scale:
“Most of the arguments made in the paper are unfounded:
— Why does Bitcoin need to scale to VISA level? Clearly there must be tradeoffs between centralized and decentralized systems?
— Where do the necessary and sufficient conditions come from? Why is this a complete enumeration?
— Full nodes that do not mine still provide value to the system by serving SPV clients. If the number of these nodes does not grow (as the auther’s state the number of nodes converges), how can the nodes serve an increasing amount of SPV clients?
— Evaluation of scaling proposals cannot be limited to the availability of sufficient processing power. For example, larger blocks can lead to more centralization as (with the current Bitcoin protocol) they hurt fairness (see, e.g., Bitcoin-NG by Eyal et al.)
— The proposal for a block size is just a proposal, it lacks any economic arguments, measurements or simulations.”
One small thing that the review did get correct (and there are not many insights it would seem):
“The paper furthermore completely neglects the academic research on blockchain scaling solutions.”
Yes, we reject this. The idea that we have scaled and run a test with 340 GB blocks would seem to be an anathema to the reviewers. They state:
“The paper lacks any kind of economic arguments, measurements or simulations to support the proposed block size increase.”
Which of course is false. As we note on Page 14, “The simulations proved that it is possible to scale the Bitcoin protocol to support a block size exceeding 300 GB”.
So, we see that the issue for some is that when it comes to large blocks, “they hurt fairness”. Whatever that means. Rather, we show that Bitcoin scales and can do so efficiently. Today, we see increasing propagation delays as the network sends large blocks. The integration of small systems that do not mine ensures this.
In our models we demonstrate that removing small nodes and allowing these to be culled is beneficial to the network. Bitcoin is and was never a home user system, it was designed to end on specialised nodes in datacentres. This was stated in 2010:
The current system where every user is a network node is not the intended configuration for large scale. That would be like every Usenet user runs their own NNTP server. The design supports letting users just be users. The more burden it is to run a node, the fewer nodes there will be. Those few nodes will be big server farms. The rest will be client nodes that only do transactions and don’t generate.
When this is used as the model for the network, we see that we can expend the level of power and propagional capacity many times. The result is to be able to scale the network to more than 20,000 transactions per second with a delay and latency of between 2 and 3.5 seconds.
Why does Bitcoin need to scale to VISA level?
Well, this was asked, and the answer is simple, because it can. Bitcoin is better than Visa. You can say that this is just my view, but like it or not, we are going to scale Bitcoin further than any group opposing scaling wants to see and then some.
The answer for these people is that if you do not want to have a large scale financial system, stick to BTC and not BCH as we will help ensure that the various Bitcoin Cash groups have the means to make this happen.
Well, it is very simple. Bitcoin Cash can and will scale. nChain has funded the Bitcoin Unlimited 1GB test. In this, it is definitively demonstrated that Bitcoin can scale. Not just a little, but beyond Visa. I really do not care if a bunch of academics think scaling Bitcoin is wrong. The result is that we are going to do this. Bitcoin is not a system for the elite or the few, it is a global payment system.
nChain are in the position (unlike many academics) where we do not need to seek funding and we do not need to seek the generalised pessimism that comes from academia. The result is we will not be asking permission, we will empower others in Bitcoin Cash to ensure that we do grow. Not in small steps, but such that Bitcoin scales to be able to replace the current global financial system in the coming decades.
Whist others enjoy telling us what we cannot do, we plan to do this and show it can. Bitcoin can scale and we will demonstrate just how much.