Original author: Nic Carter
source:
Compilation: Katie Gu @ Odaily星星日报
Watt invented the centrifugal governor in 1788, which is a small and ingenious device. With this device, the steam engine is feasible in the industrial field. The centrifugal governor obtains high-efficiency circulating input from the steam engine and applies it to the weighting ball. When they rotate, centrifugal force pushes them upward, moving the lever connected to the valve. When they spin faster, the valve closes. In this way, the governor gets input from the steam engine and mechanizes the steam flow, thereby regulating the speed of the engine.
This innovation makes the steam engine suitable for industrial requirements that require stability and predictable speed, such as mechanical looms. The point is that as the energy of the system increases, a negative feedback loop is formed, which limits its growth. For steam engines, this is part of its design. As we will see, a similar phenomenon exists in public chains, but the results are more complicated.
Strange case of periodic fees
As the market frenzy faded in 2017, I noticed that Bitcoin’s fee and transaction volume seemed to be repetitive. The block will be filled up, the fee will soar, the transaction volume will start to drop, and then the block will be filled up again. According to my analysis, this cycle was repeated six times in 2017.
Please note that I have smoothed the average handling fee and the number of transactions for the 7-day moving average value. People view the “fee crisis” of Bitcoin at the end of 2017 as a simple event, but in fact there were at least four periods where the fees rose sharply. If you count the smaller peaks, there are six periods. It’s just that most people calculate the fees in U.S. dollars instead of local units, so when the dollar value of Bitcoin also soars, they really notice the final soaring fees.
On average, the handling fee will peak approximately two weeks after the transaction is completed. Although it has been rising throughout the year, the entire cycle took two months. As the block fills up, a small number of new transactions pushes the fee to an unacceptable range for traders. Of course, the handling fee is just a phenomenon, and the potential problem is actually the block space.
You can visualize the development by checking the relationship between block filling and average commission.
Using this method, you can see that as the block fills and reaches the limit, the handling fee will gradually rise. As the fee reaches its peak, users choose to reduce transactions and clear blocks, resulting in uneven block sizes. But with the drop in fees, the block space looks more attractive to users, and users will almost return to the protocol again, causing the block to be filled up again. In the last soaring transaction fee around January 2018, whether it was the price of BTC or USD, the block reached its limit and lasted for several months, and the transaction fee reached an astonishing level.
SegWit was officially launched on July 23, 2017. Since it represents an increase in the effective block size (allowing blocks to carry up to 4MB of data), it can be seen that in the second half of that year, the second half of the block exceeded the 1MB limit. However, since SegWit is not immediately allowed by the processing program, the average block size will only grow slowly.
This is a process diagram. I call it a volatility graph, because it causes fluctuations in handling fees and blockchain resource utilization, just like a sine wave. But you can also describe it as a negative feedback loop, because when the fee reaches a certain threshold, the fee inhibits the transaction.
This is just out of my curiosity. I pointed this out at the time, but it has nothing to do with cycles. You can’t guess what the number of transactions will be in a week. The lesson I learned from this is simple: at some point, users are frustrated with fees and cancel the priority of on-chain transactions, especially if the fees are large relative to their transaction size. SegWit provided quite a bit of help. From a certain perspective, the handling fees are automatically adjusted because they encourage a large number of block space consumers to save a bit of consuming chain resources. However, all of this makes me more certain that the Bitcoin fork will require a large number of delayed transaction systems, which need to be added to BTC under a series of trusted models. For me, another gain is that users have proved to us that the transaction cycle is clear, and after a period of high fee pressure, it will take several weeks to return to the previous transaction level.
Ethereum’s 2020 “fee crisis”
This year, when the fee of Ethereum starts to climb and finally exceeds the fee of Bitcoin, I want to know if Ethereum will see the dynamic recurrence of Bitcoin’s fee. I am confused as to whether it will have the same impact or is more disruptive for Ethereum, because so much liquidity is “on-chain” (as opposed to exchanges where it is mostly off-chain). In my opinion, the supply of Ethereum’s block space is dynamic and can increase as usage increases. It turns out that the handling fee will eventually be more disruptive than I expected.
Just like Bitcoin in 2017, with the growth of transaction volume, the utilization rate of Ethereum has also soared this year, thereby gradually increasing the handling fee. The token launched by Compound in mid-June has increased the utilization rate of the blockchain, thereby increasing the pressure on handling fees. Since mid-August, many other products have been launched to increase handling fees. The dual issuance of SushiSwap and Uniswap tokens are several notable events. On September 2, the average transaction cost of Ethereum exceeded $14. In total, the ETH fee of USD 16.7 million was paid that day, far exceeding the newly issued miner income of USD 5.98 million. As these fees increased, some users chose to postpone transactions, and the number of transactions began to decline. As we entered a new period of high fees in mid-August, the daily transaction volume of Ethereum began to decline steadily. Handling fees peaked late on September 2, but they also started to fall last month.
Two days later, on September 4, I predicted in the On The Brink podcast that the high transaction fee of Ethereum will not only affect the utility of the chain, but also affect the liquidity of the decentralized exchange. Here is a short conversation from the podcast:
Nic: We have seen the phenomenon of transaction fee-tx count fluctuations in Bitcoin in 2017, and I predict the same phenomenon will occur in Ethereum. Obviously, the handling fee increases with the increase in the utilization of the block space, and at a certain critical threshold, users start to be unacceptable and stop trading for a period of time. This is an economically unfavorable deal for them. As a result, the number of transactions and fees decrease, and then the fees become cheaper, so people can trade again and keep the loop.
And I think if you look at the sell-off, it may also be relevant, because some of the retail investors who use on-chain exchanges are pricing in these transactions. Moreover, if there are no retail investors, smart traders also have a large amount of unknowing trading liquidity to play against. Therefore, I think that under normal circumstances, this is a blow to liquidity.
Matt: Without retail investors, there would be no collectives.
Moreover, although it is difficult to determine the ETH price dynamics, I am not saying that they are completely caused by this phenomenon, but ETH and USD peaked on September 1 and then fell from there, while the handling fee is for the rest of the month Keep going up.
So what happened?
First of all, Ethereum has entered a period of volatility in what I think will be its first major fee chain usage. The fee reaches its peak about three weeks after the transaction is counted, and then both decline. What’s more interesting is that the average size of Ethereum transactions and various stablecoin transactions soared with the increase in fees. This makes sense. Users can see how much commission they are willing to pay (as a percentage of the transaction). As the commission increases, they will stop making small transactions and large traders begin to dominate.
Not only that, but basic Ethereum transactions also show this fee sensitivity. With the increase in ETH fees, tokens like Tether have also proved the increase in transaction size.
This shows that traders have a fee threshold, which is the percentage of the transaction amount they are willing to pay, and as the fee increases, they become reluctant to make smaller transactions unless they really have to.
The above can be explained by a third explanatory variable, such as the growth of liquid mining, which not only blocks the chain, but also leads to the entry of larger traders. When I investigated the composition of ether transactions, I understood better. I intuitively felt that traders with fewer assets were pricing in fees, but I didn’t know until I compiled this chart.
Initially, this chart may not be obvious. As we have seen above, as the fee increases, it shows a decrease in the number of transactions. But what is more interesting is the dynamics of changes between ETH and USDT transfers, which I divided into the minimum threshold of $500. You will see that as the fees increase, transactions that account for more than $500 in ETH have steadily increased their share relative to smaller transactions. Let’s ignore the content of the contract and focus on the two types of ETH transfers found above:
For non-contract bullish transactions, about 85% of ETH transfers below $500 remained unchanged in June, but when the fees soared, they fell to 40%. Finally, there is some solid evidence that proves my intuition that the fees benefit smaller users.
It can be predicted that the same phenomenon exists in the use of the largest ERC20 token, Tether. As the median fee began to climb in mid-August, smaller USDT transactions began to fall. Larger Tether transactions remain firm, but still show a downward trend in the overall trend, as transactions gradually stabilize during periods of high fees.
Will high fees affect liquidity?
Therefore, we have determined that in Bitcoin and Ethereum, there is an obvious negative feedback loop between handling fees and blockchain resource utilization. In Bitcoin, we know that this cycle has historically taken two months to complete. We have not yet seen what Ethereum will look like. In addition, it is obvious that traders have to consider the maximum acceptable fee threshold relative to the size of the transaction they conduct, and smaller traders will postpone the transaction during periods of high fee. This has caused the average transaction size of ETH and other tokens to gradually increase during periods of high fees.
In 2017, Bitcoin transactions were conducted on centralized exchanges, and blockchain was used for settlement between exchanges and user deposits and withdrawals. The actual market is off-chain. Users can fund exchange accounts with fiat currency and hold (and then sell) Bitcoins without actually touching the blockchain. Therefore, when a fee crisis occurs, it will definitely hinder the Bitcoin economy, but if there are already funds on the platform or want to remit US dollars, then many users can trade on the exchange.
In contrast, the series of transactions of Ethereum transactions and related tokens in 2020 is largely an on-chain phenomenon. Today, centralized exchanges are still very important for price formation, but some DEXs like Uniswap sometimes even surpass the largest centralized exchanges. Since automated market makers DEXes do not require KYC, custody of tokens with third parties or lengthy startup processes, they are much more convenient for end users. The orderless AMM model is also very easy to use. And certain types of assets and risk exposures (such as smaller DeFi tokens or liquid mining) can only be done on the chain. Eventually, a vibrant on-chain liquidity industry emerged. Since almost everything traded on these DEXs is Ethereum or Ethereum tokens, everything needs to be paid. Unlike centralized transactions, every transaction you make on DEX must be settled on the chain. Therefore, on-chain fees have always been a consideration.
Therefore, compared to the ordinary fee-tx count fluctuation model we saw in Bitcoin in 2017, this is a modified version that explains the new dynamics seen in Ethereum in 2020.
So, where does the second loop come from? Why should we single out the heavy chain that requires DEX for special treatment?
First of all, it must be understood that, as Maya Zehavi said, the high handling fee is a decreasing tax on users. The tax rate is decreasing because the handling fee is not proportional to your assets, but whether you are transferring $100 worth of ETH or a handling fee of $10,000, the charging method is roughly the same (the handling fee is a function of the transaction calculation volume, not The dollar value of the transaction). This is similar to the diminishing approach of business taxes, because groceries account for a larger share of working-class income than the rich. Therefore, for wealthy families, the fixed 5% business tax is actually a large part of their income.
Give an analogy. Imagine a private poker game where the commission is fixed in U.S. dollars instead of a percentage of the original value. (Commission is the fee charged by the operator for running the game). Players can choose to play a given hand and pay a fixed fee, or they can choose to be out, in which case they do not need to pay a fee. There are various players at the table, there are a few professionals, some semi-professional players and some gamblers. They are passionate but not particularly good at poker. You can place bets wherever you like. To win, you not only need to contend with your opponents, but you also need to profit from crowdfunding. It is not enough to win, you must bear the transaction costs to win.
If the draw is low, then everyone will actively participate. But when the table operators become greedy and increase rake, players with fewer chips start to be helpless and have more and more cards in their hands. If they need to raise more than 1/10 of their chips just to play a game, they will opt out (unless they have a trump card). As the absolute value of commissions rises, players gradually set prices from the game, starting with the player with the least funds.
When smaller, less skilled players start to quit the game, the profit of the game for everyone else is greatly reduced. After all, semi-pros rely on unskilled opponents to win money. Every poker player knows that you might not want to sit at the table with durable players and semi-professional opponents on Tuesday morning. You want to show off your skills on Friday night, play cards with lazy amateur gamblers, make quick decisions, and then drink in the dark.
In analogy with Ethereum, fees are like bets. They make retail prices high. Until recently, they spent a lot of time on Uniswap and other DEXs. However, as the handling fee rises to an average of $14 (for DeFi transactions, such as the exchange of ETH to DAI on Uniswap, the handling fee is much higher), for individuals with less funds to participate in the transaction or liquidity For mining, it becomes uneconomical. The entire market is controlled by them. Moreover, because retail investors are professionals who trade (and profit from it), they belong to people who don’t understand the situation. Therefore, if retail investors do not directly reveal their identity to participate, then this game is meaningless. All of this indicates that liquidity has declined and transactions have deteriorated.
In summary, this theory is difficult to prove by experience. A detailed analysis of wallets that are active in DeFi under different thresholds and their response to fees, or comparing the difference with ETH fees, will help to prove these assumptions. However, there are some indicative data for reference. This is the sum of the total amount of DEX provided by Friedrik Haga’s Dune Analytics Information Center and the comparison with the average transaction fee of Ethereum.
You can see that the quantity and handling fee have changed in the past few months. This is consistent with the prohibition of the use of handling fees (especially in the retail sector). The reason is difficult to infer. It is probably because the high number of DEX is the reason for the increase in the commission (and vice versa), rather than the high commission is the driving force to reduce the number of DEX. More interestingly, this graph shows the daily new addresses participating in various DeFi protocols. This is the data of the Dune dashboard built by Richard Chen, compared with the average ETH fee.
In addition to the two significant increases in handling fees for the issuance of SushiSwap and UNI tokens, after the ETH handling fee reached a new plateau in mid-August, the number of new daily participating users of DeFi seems to be decreasing.
The conclusion drawn here has more reference value. Obviously, users with a small amount of assets will postpone transactions when the commission is high, but the direct impact on chain liquidity is difficult to measure.
The future prospects of Ethereum fees
We have not yet seen the complete fee chain resource cycle play a role in Ethereum, so it is difficult to estimate the length of the cycle. But I predict: I hope that we will see Ethereum maintain a continuous, positively correlated fluctuation between transaction fees, transactions, and the liquidity and transaction volume of transactions on the crucial chain.
It will look like this:
If you think that the trading volume of DEX and the liquidity environment on the chain will generally drive up the price of ETH, then you don’t need to spend too much imagination to make an educated guess about the potential price impact of this phenomenon. In the need for the value of Ethereum to be used as a debt-free collateral on the basis of the DeFi system, the restricted handling fee is likely to reduce the total locked value in the system and increase the pressure on the price of ETH.
In other words, there are some counter-cyclical features that may reduce volatility and protect Ethereum from the impact of transaction fee fluctuations.
Dynamic block space supply
Unlike Bitcoin, Bitcoin only officially increases the upper limit of block size once it exists. Ethereum is more willing to generate more block space when necessary, although the speed of increase is limited, in theory, flexible block space can be used To stabilize the fee, but this is at the cost of increasing verification requirements, which is already quite high for Ethereum. As the utilization rate of Ethereum increases, miners are rewarding traders with higher and higher rewards.
However, because the increase in the handling fee limit will increase the uncle block rate, make the network more vulnerable to DOS attacks, and make the full node operating cost higher, the Ethereum community has divided discussions on further increasing the handling fee. As far as node operation costs are concerned, although Ethereum holders are more willing to accept greater compromises than Bitcoin holders, both parties generally believe that merely increasing block space is not the final solution. Therefore, although the block space has a certain degree of flexibility to enable it to cope with increasing demand, there is almost no support from the community to pursue a completely loose block space policy.
ETH 2.0/fork
When I first wrote about the prospects of ETH fees, God V responded to my article (I claimed that Ethereum may suffer high fees for a long time, which will affect the viability of non-financial applications), God V clarified His views on handling fees:
In fact, he pointed out that he still believes that Ethereum has a great influence on the future of low-cost blockchains, where various non-financial and financial applications can be used. The solution he proposed is the recent Rollups and the long-term ETH 2.0/fork. V God said that he expects the capacity of ETH 2.0 to increase by 100 times, although he expects that due to demand, the handling fee may eventually rise to the same level.
I even agree that if your production increases, the world will find more uses for the commodity and the price will rise. Therefore, if you create more block space, it is not surprising to see the average fee increase. In any case, ETH 2.0 looks very far away, so it is difficult to predict. But at least, the handling fee will not be adjusted in the short term.
Rollup
Ethereum currently believes that Rollups is the main way to alleviate the current fee dilemma of Ethereum. There are two main types of Rollups: ZK and Optimistic, but they usually involve bundling many payments together and greatly increase the economic density of transactions. In theory, while preserving the guarantee of basic layer transactions, it also greatly improves TPS. In short, traders rely on intermediaries, which match up a large number of transactions and broadcast them.
ZK Rollup involves broadcasting obviously truncated transaction stubs and proof that the transaction bundle is valid for the ledger. Optimistic involves a semi-trusted operator that matches transactions, and most traders believe that the operator has no malicious behavior. Theoretically, deterrence is achieved through a combination of fraud evidence and financial penalties for misconduct. Currently, ZK Rollups are mostly limited to simple transfers, while certain types of Optimistic Rollups (OR) promise to open all transactions currently possible in Ethereum. For a comprehensive analysis of OR status, see Daniel Goldman’s comprehensive report.
Rollups effectively moved the transaction data out of the chain and placed the proof of validity on the chain. Bitcoin developers had similar data conservation visions a few years ago and pursued Lightning (which can reduce hundreds of thousands of payments to a few on-chain transactions), side chains, and measures to promote efficiency, such as batch processing and the use of SegWit. Thanks to the popularity of Rollups and the obvious speed of progress, V God is supportive and regards it as the best near-term method to expand Ethereum. Although ETH 2.0 may not work yet, the future of Ethereum’s Rollups is close at hand.
There are two reasons why Rollups may not be a panacea for the Ethereum fee issue. First of all, it will be a challenge to urge all block space users to become the person in charge of the system, especially if they are service providers and can transfer fees to end users instead of internal personnel. We learned from Bitcoin that if intermediaries can transfer fees to end users, they have no continuous incentive to invest in more sustainable infrastructure.
In addition, because Ethereum has repeatedly raised the fee limit to effectively help block space consumers (at the cost of validators), a large number of users may spend more time lobbying resources to increase the fee limit instead of spending a lot of time on it. Matching transactions. Similarly, the upcoming mega block space of ETH 2.0 may weaken the enthusiasm of large traders for Rollups. On the contrary, the current main contribution of ETH2 is likely to reduce the enthusiasm of large traders to improve its efficiency.
Second, the global user base of blockchains like Ethereum is not entirely developers and advocates, but can actually attract users directly. Ethereum is open to everyone and does not exclude participation. This is part of the reason why it is more popular than Ponzi schemes and other niche projects. The coordinator of these schemes (usually the main users of some block space) does not necessarily have to plan long-term or try to optimize their traces on the chain, but is more focused on making money quickly.
In terms of technology, the settlement quality of Rollups (especially Optimistic) is different from that of basic layer transactions. Vanilla Ethereum transactions are almost final transactions and do not bear the risk of refund or settlement. This attribute allows “atomicity”, which means that chain transactions either all happen or not. This enables users to safely interconnect multiple systems without having to risk chain failure because a payment in the chain fails to clear. This is a very desirable attribute, depending on its status as a digital carrier asset. Ethereum has “atomicity”, which belongs to composability, allowing smart contracts to safely reference each other, so that more complex systems can be built without having to evaluate each module.
The introduction of more complex systems like Rollups will raise questions about these assumptions about atomicity and composability. Since some fraud proofs in “Optimistic Rollups” challenge the trust model, the deadline can be extended in case of dissatisfaction. Matter Labs estimates that the final determination time for OR is about 1-2 weeks. (Please note: this estimate was made in November 2019, so the state of the art may have changed since then.) A suggested solution’withdrawal period’ involves intermediaries, which provide users with Temporary access right to exit from Rollups. Allow users to use frozen tokens under special circumstances. As far as I know, this effectively creates a tiered system with longer certainty compared to the “free version” of Rollup’s exit channel, and shorter certainty in the paid accelerated version.
Simply put, Optimistic Rollups introduces the potential for delayed settlement. Delayed settlement is the core of all efficient modern payment systems, but this is a different settlement model from the basic layer of Ethereum transactions (similar to cash). If you are accustomed to only using bank wire transfers for transactions, switching to ACH and credit payments will be more effective, but it also means that you have to bear the guarantee losses surrounding settlement. This is why credit card payments can be refunded. In fact, such instructions were not implemented immediately. So it is good for consumers but bad for merchants, and these funds have been designated to pay bills due.
In addition to the issue of determinism, impaired composability is another issue that may prevent an immediate transition to OR. I am not an expert in Rollups, but I have seen that the problem of composability is the biggest disadvantage of OR.
in conclusion
The difference between the centrifugal governor and the on-chain commission is that the adjustment of commission is more of a side effect of a functioning system than a key design consideration. In the public chain, the existence of the fee to ensure non-wasteful consumption of network resources and provide income for verifiers. Strictly speaking, they are not going to check the usage of the system. But in fact, they do work, and this braking tends to be sharp rather than gradual.
We have seen fluctuations in resource usage, transactions, and even the liquidity of products on the chain, rather than slow deceleration. These dynamics have already appeared in Bitcoin, and have just begun to appear in Ethereum. However, due to the advantages of on-chain exchanges in Ethereum, it can be said that they are more disruptive to today’s networks. As Kyle Samani said, the limited throughput affected by fee pressure may be an “invisible danger” for DeFi.
The Ethereum community should consider the impact of repeated charging cycles and fluctuations on the network, and at the same time realize that some applications may not be able to operate due to long-term high fees. Finally, for Rollups, compared with the attractive fork technology, it may not be the best method from the perspective of fees, but it does play a role in settlement guarantees.
