Remove OP_Return: What impact does it have on the Bitcoin ecosystem?

Recently, there has been a proposal to remove the size limit on OP_Return outputs in the Bitcoin Core software code repository. OP_Return is an output type designed to store arbitrary (non-transaction) data on the Bitcoin blockchain. Under the existing size limit, Bitcoin Core nodes will not relay transactions that exceed this limit. However, this is not a consensus rule, and Bitcoin Core nodes will always regard blocks containing such transaction outputs as valid blocks and accept them.

Removing this restriction has sparked some controversy, with some believing that the restriction is necessary to prevent or stop spam. Many who advocate for keeping the restriction hope that Bitcoin is used solely for "financial transactions" and wish to consider data related to images and the like as non-standard transactions.

Ordinals Frenzy

However, the craze of storing images on the Bitcoin blockchain has long since emerged. This craze began in early 2023 and is known as "Ordinals". Unlike using OP_Return outputs, images are stored in the input scripts of Taproot spends. The vast majority of these Ordinals transactions are already standard transactions, which are relayed by Bitcoin core nodes. In some cases, this Taproot method is cheaper than OP_Return, as it benefits from the 75% witness data discount brought by the Segregated Witness (SegWit) upgrade.

According to a dashboard on Dune.com, the usage of Ordinals has remained high, with over 88 million inscriptions and more than 7,000 Bitcoin spent in transaction fees. At current Bitcoin prices, this exceeds 660 million dollars. Many businesses have emerged in the Ordinals space, looking to capitalize on this growth trend, with millions of dollars already invested in Ordinals-related tools such as wallets, inscription trading systems, and methods for creating Ordinals and submitting them to the network.

Many people view these images as spam. In our view, in this context, we tend to look at spam from the intent of the person creating the transaction. Are they trying to deliberately harm others, or are they trying to profit from it? Keeping this in mind, we believe that storing images on the blockchain is not always spam, as those doing so seem primarily to be seeking personal entertainment or speculation and trying to profit. However, if someone puts images on the blockchain to maliciously attack others (which does exist to some extent), then yes, that is spam.

Although images on the blockchain may seem like junk information to many, we agree with the theory of subjective value:

The value of various consumer goods and services does not objectively and intrinsically arise from the things themselves, but rather depends on the individual conducting the assessment. His valuation is a subjective matter, and even he cannot reduce it to objective terms or measurement standards.

Some people seem to like images on the blockchain and have already paid over 600 million dollars for them. Since the value of these items is subjective, what right do we have to question it? We can only say that we do not value these images and would not pay for them. We believe that those businesses and individuals hoping to profit by investing in this field will likely end up losing money. But let's let the market decide!

Our view is that the horse has already left the barn, and people are already using blockchain on a large scale for image storage, and retaining the OP_Return restriction does not change this. Systems that use part of the Taproot input script to store images already exist, and on a per-byte basis, this method is four times cheaper than OP_Return.

Bitcoin Mining

We have been关注 Bitcoin mining for over 14 years. We have witnessed Bitcoin mining evolve from an amateur hobby into an industry with publicly listed companies. We closely monitor 10 of these listed companies and have read nearly all of their public disclosure documents. These miners report their financial statements quarterly and update the market with their production data monthly. We frequently communicate with the investor relations officials and management teams of these companies. These management teams are very fond of Ordinals, viewing it as a potential revenue driver, which is a crucial income driver in a competitive industry. The idea that Ordinals is spam and should be filtered out does not resonate with these professional management teams now or in the future. Some people may dislike it, but this is the reality of business. This is also the reality that some of us have always anticipated. Bitcoin has grown, it is a business, and businesses need to maximize returns and shareholder value.

Bitcoin is about incentives and the pursuit of incentive compatibility. The operation of Bitcoin is not because this space is composed of a group of altruistic, well-meaning, and goal-aligned individuals. The Bitcoin space consists of a diverse range of people with different views and philosophies. The operation of Bitcoin is not because we are all on the same team, but because it is robust and the incentive mechanisms are aligned. We believe it is time to remove the OP_Return restrictions, which are paternalistic protective measures, and embrace the economic reality of the block space market.

If larger 'OP_Return' outputs are still considered non-standard transactions, but people still want to use them, miners will only set up businesses that receive these transactions directly, bypassing the public mempool. Marathon [ticker:MARA US], the largest publicly traded miner, has already done just that. However, as far as we know, this service offered by Marathon is not popular at the moment. Still, if miners start receiving transactions off-chain, this has a lot of negative effects for Bitcoin. This will mean that the difference between the transactions in the block produced by the miner and the transactions that users expect to see will increase. This could break techniques such as compact blocks, which help blocks travel faster through the network by eliminating the need for nodes to download transactions twice (once for the mempool and once after the transaction enters the block). It may be wise for Bitcoin Core to remove this restriction upfront to ensure that compact blocks are not broken. If it is corrupted and the block propagation delay increases, then this could benefit larger miners and mining pools at the expense of smaller miners, exacerbating the centralization of mining.

The establishment of enterprises that accept non-standard transactions by miners and mining pools can also bring other negative consequences. Establishing such enterprises requires costs, such as technical costs and marketing costs. This business model may also have monopolistic characteristics, as users may prefer to use a simple platform to submit their non-standard transactions. This increases the barriers to entry for mining and mining pools, making it more difficult for small participants to mine. This again leads to more centralization pressure. Once these systems develop, it becomes difficult to stop these businesses even if the strategies of Bitcoin Core are relaxed later on. For example, once the infrastructure is built, lazy customers may continue to use these services instead of the public mempool.

We believe that Bitcoin developers should strive to keep the software competitive. Making the open-source trading choice algorithm competitive in maximizing revenue, preventing miners from building their own proprietary algorithms, and making the public mempool competitive to stop miners from establishing private mempool businesses. We understand that not everyone thinks this way, but this is the current economic reality in the mining field. We hope the mempool can work effectively, and we believe that removing the OP_Return restriction is a better choice than burying our heads in the sand pretending that garbage transactions aren't being mined. The choice is between having an effective mempool or an ineffective mempool.

Node Operator

If we assume that the blockchain is full, then the increase in the usage of OP_Return actually makes running a full node easier. Keep in mind that OP_Return does not enjoy witness discounts, so the maximum size of blocks containing only OP_Return outputs is 1MB, far less than the maximum of 4MB. Meanwhile, OP_Return outputs do not inflate the UTXO set. Other protocols use alternative systems (such as fake addresses) to store arbitrary data, and this approach has serious negative consequences for those seeking to verify all Bitcoin transactions. OP_Return is simply data that does not need to be verified and can be ignored. Those concerned about cheaper node operating costs do not need to worry about the problems that removing the OP_Return restriction would bring.

How do we stop spam?

At the beginning of this section, we will quote Eric Voskuil's work "The Economics of Cryptography." In the book, Eric writes:

Resistance to censorship is the result of transaction fees.

The core objective of Bitcoin is to resist censorship, and transaction fees are a fundamental component of the security model designed to achieve that goal. Attackers hoping to censor transactions cannot succeed by encouraging node operators to filter out certain transactions from their memory pools. If that were possible, Bitcoin would not be particularly strong in preventing censorship. Instead, everything revolves around fees, fee rates, and miners attempting to maximize revenue from each block. This is true regardless of whether a transaction is spam, whether it uses OP_Return, or whether it employs Segregated Witness. To enter the blockchain, one must bid higher than other users, which is the only viable spam prevention model. As Satoshi Nakamoto said:

As long as you're willing to bid higher than other users, you can almost always get in at a certain price.

This garbage information prevention model has been understood for many years. As we mentioned in our article about Segregated Witness in September 2017:

Indeed, spam senders can generate a 4MB block with the same fee as a 1MB block. This is a potential issue. However, it does not change the security properties of the system, as a 4MB block is not cheaper than a 1MB block; it simply costs the same. Regardless of whether Segregated Witness is used, spam senders can always bid higher than legitimate users. In fact, attackers can simply generate 1MB of non-witness data to compete with 'legitimate' users, and the cost of this attack method remains the same as before. Segregated Witness does not and cannot change such security dynamics: if attackers want to bid higher than users with spam data, they can do so.

To ensure that your transactions are confirmed in a timely manner, you need to have good information about what miners may include in the next block so that you can set appropriate transaction fees. This is also another reason why node operators may want their mempool strategy to align as closely as possible with the actual operations of miners. If Bitcoin Core does not remove the relay restriction on OP_Return, users will have to run alternative software or use third-party websites to obtain information on what transaction fees to use.

Indeed, the OP_Return output has always had limitations, and Bitcoin has been running well for over a decade under these constraints. So why remove it now? In our view, this reasoning resembles the discussions often held during the block size debate from 2015 to 2017. The "big block camp" frequently argued that Bitcoin had been functioning well for years without full blocks, so why introduce full blocks now? The answer is that it is merely a matter of successful economic reality. As former Bitcoin developer Gregory Maxwell stated in an important email in 2015:

The demand for cheap, highly replicable, and permanent storage is infinite.

In our view, these changes are always inevitable. The demand to store images on the blockchain is infinite, and the only way to prevent this demand is through transaction fees. Some people have always believed this, while others do not, which may lead to debates.

We are fortunate that the small block faction won the block size debate. If the large block faction had won and adopted something like Bitcoin XT, the block size limit might currently be around 250MB. Blocks could be filled with images, with 250MB of images every 10 minutes. This could make it impossible for ordinary users to run nodes and may have already stifled Bitcoin.

Conclusion

We have a certain level of support for removing the OP_Return restriction. Now is the time to face economic realities and remain competitive. We hope that the local memory pool can operate effectively and that the public p2p transaction broadcasting system becomes the ultimate winner. If attackers or spam senders want to bid higher than other users, they can do so, and we should embrace this reality. The budget for spam won't last forever, and many investors in blockchain imagery are likely to collectively lose millions of dollars. People will learn painful lessons, and Bitcoin will become stronger as a result.