The heart of the Bitcoin scaling debate

The core of the debate over Bitcoin expansion is actually a dispute between two factions: whether Bitcoin is结算系统or电子现金系统.

Settlement System

The first is that Bitcoin (with a capital letter, representing the network) is a global settlement system, and Bitcoin (with a lowercase letter, the thing we hoard) is functionally equivalent to gold.

Bitcoin payments on the blockchain are used to settle retail/commercial transactions that are managed by offline networks or third parties. These settlement numbers can be high because they represent thousands of transactions that occur outside the blockchain and are completed by a large number of users. This system is very similar to today's financial system, where financial institutions and companies use unique domestic and international networks to settle payments.

Proponents of this school of thought reject the long-term use of blockchain for retail purposes (such as buying a cup of coffee), and instead view secondary networks (such as the Lightning Network, or LN) as a more flexible and low-cost environment for conducting such activities.

Electronic cash system

The second school of thought is that Bitcoin is both a settlement system and an electronic cash system. According to this school of thought, the neutrality of Bitcoin payments on the blockchain is the most important. In other words, on the blockchain, retail/commercial transactions should occur in parallel with settlement payments.

This view opposes the following scenario: due to the constraints on the transaction capacity of the Bitcoin network, transaction fees increase, and a. only financial institutions or enterprises can afford the transaction fees; b. transaction fees no longer have an advantage over traditional legal currency payment service providers, so ordinary traders are forced out of the blockchain.

Security and Management

The debate over capacity expansion has largely focused on the security impact of capacity expansion and the trade-offs between the two views. On the surface, the significant differences come from people's subjective evaluation of acceptable consequences. In fact, the core of the debate is the sword-qi dispute I mentioned above. This has led to a management crisis for Huashan Bitcoin.

The purpose of this article is not to express support or opposition to either view. Our focus is on Bitcoin's transaction capabilities and its future impact on the network while serving either view.

In this post, I build upon an excellent BitcoinTalk thread. Below, I compare Bitcoin to popular settlement networks, remittance tools, and retail/merchant payment providers, and find out where Bitcoin actually stands out compared to any of these systems.

Bitcoin, Settlement Networks and Payment Providers

To compare transaction capabilities, I selected representative retail/commercial payment service providers, settlement networks, and clearing and settlement systems.

Assuming that an average Bitcoin transaction is 520 bytes (by the way, Bitcoin transactions are around this size, and slightly larger if you include Lightning Network or OpenBazaar multisig), I calculated the minimum size that a Bitcoin block would need to be in order to support the following systems.

It is more intuitive to change to a picture:

Notes and Observations

1. I was surprised to see how close Bitcoin's transaction capabilities are to those of Fedwire Funds and TARGET2. Interestingly, both of these systems serve small, proprietary networks of financial institutions and, unlike Bitcoin, are not really accessible to consumers.

2. The Automated Clearing House (ACH), in the United States, is a typical settlement and clearing network between depository institutions and central banks. I estimate that 90% of payments have nothing to do with buying coffee or something. From the perspective of "Bitcoin is a settlement system", achieving such transaction capacity requires increasing the block size by more than 200 times its current size.

3. ACH just stands for the United States!

4. SWIFT (Society for Worldwide Interbank Financial Telecommunication) is not a settlement network, it just sends payment orders, which are used to execute settlement transactions.

5. Western Union has a 15% share of the global remittance market.

6. Currently, none of the accepted Bitcoin Improvement Proposals (BIPs) would increase the block size to Visa-level transaction capacity (nor does BIP101).

7. It is conceivable that people will use Bitcoin for all of the above applications until transaction fees rise to the point where a particular application becomes economically irrational (due to protocol changes or restrictions on miners holding a majority of hashrate). This effect means that there will always be pressure to make transactions directly on-chain whenever possible.

Influence

If Bitcoin is to fulfill its aspirations of becoming a global settlement network, regardless of whether that network captures any meaningful market share, Bitcoin’s transaction capacity needs to improve.

If Bitcoin is to break into国际汇款市场, its transaction capabilities will need to be improved.

If Bitcoin is to replace the settlement networks used by central banks and financial institutions, a small increase in Bitcoin’s transaction capacity will be enough, coupled with significantly higher transaction fees to exclude low-value payments from the blockchain.

The table above shows the transaction capacity required and the resulting block size at different transaction frequencies. There are 3 models used:

1. Estimated number of Bitcoin users by 2019

2. Approximate population of the United States

3. World population

Note that these calculations do not account for or attempt to estimate transaction volumes by commercial, financial, or government entities.

With an effective secondary network (such as Lightning Network LG), users may only need to open and close payment channels twice a year and conduct two on-chain transactions on average (optimistically this is possible). If current estimates of users in 2019 are valid and the Lightning Network is already in use by then, transaction volume and block size may be significantly reduced, even lower than current levels. If this is the case, miners may have to accept lower transaction fees.

However, as Bitcoin becomes more widely accepted, the pressure to upgrade the blockchain will always exist in the long run. Even under the conservative view of using Bitcoin as a global settlement system, the number of transactions per person per year is extremely low, and the transaction capacity required by the system will push the block size to the upper limit recommended by most block proposals.

This means that larger blocks are inevitable. Unfortunately, this means that over time, mining and full node operation will inevitably depend on the level of hardware and network technology infrastructure. The risk of centralization will depend on the acceptance of Bitcoin, the degree to which secondary networks migrate transactions off-chain, and the amount of on-chain transactions that are used regardless of the existence of secondary networks.

Failure to successfully increase transaction capacity will make the roles entering the blockchain more centralized, replicating an existing financial system. Ironically, the network can run on anyone's computer, but it is possible that no one can afford to use it. This would be a failure scenario.

Lies, damned lies, and linear trends

When does block size become an issue?

If we look at the linear trend over the past two years (starting today, September 14, 2015):

First, R^2 is far from perfect, so don’t take the analysis above as 100% fact:

1. A rough estimate is that the average value of blocks will reach the upper limit in 1067 days, that is, three years from now. That is, the purple line in the figure reaches the 1M node. In fact, many blocks will be fully loaded before then.

2. And there is no reason to assume that this trend will run linearly, especially as new projects come online (such as OpenBazaar) giving people more reasons to adopt Bitcoin transactions, similarly, expectations of rising Bitcoin prices will drive up Bitcoin's spread and transaction volume.

3. If we calculate the standard deviation of the average daily block size over the past two years, it is around 0.12MB, which means we would see a clear problem after 810 days.

4. If we use a soft cap of 0.75MB and add the standard deviation of block size, we will see a surge in blocks after 310 days if most miners do not raise the soft cap. Now some mining pools like BitFury are already mining 900KB blocks, and we have reason to believe that other miners will follow suit.

Final Thoughts

Decentralists are to be commended for their brilliant performance in alerting the community to the centralization and security risks that come with increasing block sizes, and those who favor larger blocks would be wise not to ignore their warnings.

On the other hand, I have observed that decentralists tend to ignore the importance of transaction capacity and insist that Bitcoin can be used for settlement, perhaps not realizing that any meaningful settlement network would require a significant increase in Bitcoin's transaction capacity.

People have specifically highlighted the Lightning Network as a way to secure the ecosystem, and that the ability to trade can be handled by the Lightning Network and similar secondary networks. Maybe that will be the case, but I think there will be a lot of technical challenges. I really want to see them get to work. Will they be ready, and when will they be tested? And of course, they still have to pass the most important test, whether consumers will accept them.

Likewise, those who are keen on bigger blocks shouldn’t forget that improvements to blockchain efficiency, such as sharding and reversible Bloom filter lookup tables, are not yet ready for deployment.

Let us work hard and listen to each other's concerns.