An Economic Analysis on EIP-1559

Ajian
9 min readMay 24, 2019

--

Intro

This post was written in Chinese at first in May 7th. My good friend, Elisa, helped to translate it into English a few later. I do not want to publish the English translation at the very start, as I thought that those theoretical thinking included in this post is so ordinary that someone in Ethereum community would point these ideas out and against EIP-1599, until I found that EIP-1559 was still proposed in the Istanbul update.

So, I publish this post, as I believe that EIP-1559 absolutely should not be accepted.

We could get some background from the link of Eric Conner’s blog post on EIP-1559.

This is a chart from Eric’s post, sourced from Etherscan.io, reflecting the historical transaction fees on Ethereum. Merely based on this chart, Eric stated that Ethereum transaction fees have been unstable, it’s been difficult for users to estimate the optimal gas price, and that the pending time has been longer than expected. As a result, he presented the fee system so that people can 1) predict the transaction fee, 2) pay as less as possible the fee.

I’m going to first go through the EIP briefly, following which I’ll provide an economic analysis on the proposal to prove that EIP-1559 cannot achieve what Eric claims to do and that it will introduce some other unnecessary properties to the network.

What’s proposed in EIP-1559

The EIP-1559 mainly consists of three parts:

  1. increase the block gas limit from 8 million to 16 million.
  2. Set a BASEFEE to each block, which will be the same to all transactions in one block. When the network is at >50% capacity, the BASEFEE increments up slightly and when capacity is at <50%, it decrements down slightly. However, miners don’t get to keep one penny of the BASEFEE, because it will be burned as suggested by Eric.
  3. Apart from the BASEFEE, if you want your transaction to be included sooner, you can always tip the miners.

As per Eric, “Burning this is important because it prevents miners from manipulating the fee in order to extract more fees from users” by sending more transactions.

The EIP would have looked like a price limiting policy without the burning part, and the tip is compensation like in Uber, where the price/km is set by the system but you can tip a surged price in the rush hours.

However, burning the BASEFEE makes the whole EIP into a sort of taxation policy. I’ll explain why.

Economic Analysis of EIP-1559

Transaction fee in the current market

The graph above shows how gas price works under the current transaction fee framework.

In this graph, axis-y represents gas price and axis-x represents the amount of gas. The blue line is the demand curve of Ethereum users, meaning the highest gas price users would be willing to pay for a certain amount of gas. The curve goes down as users make their most urgent transaction first, and thus every extra gas paid gives a decreasing utility for users, which results in the users being less and less willing to pay for gas. The yellow line shows the supply curve of Ethereum miners, which is the minimum gas price for a certain amount of gas for them to include transactions. Since it requires computation to earn the gas, and since the fact that miners spend more computation power to earn every extra gas. Thus, the gas price required by miners tends to go up. There is a part of the curve that is parallel to the axis-y. This is because the protocol regulates the maximum amount of gas spent in a certain time period, which means, however much miners are willing to provide computation resource, the protocol would not allow more than the limit. That is why miners will not be able to include more transactions however many users would pay for the gas price at this stage. (miners’ computational capacity are higher than the block gas limit because those with lower capacity are eliminated by the market.)

This means, in a certain time period, if the demand curve is D1, the amount of gas spent would be G1 and thus gas price P1; if the network demand is D2, which implies a full block, the gas price would be P2.

That said, economics serves only as a thinking tool. It is not the reality, but helps us to understand reality, like Classical Mechanics and the Evolutionism. We will never see the same gas price in all transactions included in a block in real life, nor will we see blocks continuously with the same gas usage. However, economics can tell us, while all external conditions maintain and the supply and demand remain unchanged, the gas price and the gas amount will approximate either (P1, G1) or (P2, G·B).

There’s always a cost for obtaining information, as the world is constantly changing. While understanding a static situation in the system can undoubtedly serve as an important step in understanding reality.

Now, what will happen after EIP-1559?

EIP-1559: a tax on transactions

The most dominant character in the fee market with EIP-1559 adoption would be that the cost users pay to the network starts to decouple with what miners earn.

In the original setting, miners receive however much users pay, while under EIP-1559, miners would only get the “tips”.

Let’s think this through. Does it change miners’ computational cost? No. And what about the correlation between the price and a certain supply, does it change? No. Well, if nothing here changes and you don’t pay the same as before for the same service, what happens? No deals, no service, no more transactions get included. It’s really that simple!

On the other hand, does it change users demand? No. And does it change the number of gas users are willing to buy under certain gas price? No.

So what the heck changes here?

The beautiful EIP makes users to pay an extra fee determined by the network apart from what they pay to the miners! It literally just uses the word “tips” to replace the original “fees”, and have the new “fees” to become a TAX!

Let’s take a look at the graph above. Let alone the block gas limit and the dynamic BASEFEE mechanism. We now only consider the result from the fact of having a BASEFEE. As miners don’t get to keep any of the BASEFEE, users need to pay tips for acquiring miners’ service, and pay the BASEFEE along with it. Given that the BASEFEE is above 0, the gas supply from miners will match the users gas demand at some point, where the total tips miners receive for them to provide the exact same amount of gas equals the total cost of users minus the BASEFEE. As shown in the graph above, they match at G2. In this case, miners get a tip of P2, and users need to spend P2 + BASEFEE for the transaction to be included. And the green area is BASEFEE * Gas used, which is the amount of ETH burned in a certain time period — — the total amount of transaction tax paid by users. What about the triangle beside the green rectangle? **It is the value dissipated through this mechanism. Users would have been able to spend G1 number of gas, but only spent G2. This is because there are users only willing to pay less than the BASEFEE for the transaction, resulting in them not being able to send it out at all.

That’s only the simplified version of the problem. Because Eric wanted a dynamic BASEFEE according to the gas usage, the situation gets worse changes a bit.

This DYNAMIC graph above demonstrates the fact that when the total gas used is below 8 million (today’s gas limit per block), BASEFEE will go down. Therefore, if G1, the amount of gas used before this EIP, is below 8M, a BASEFEE above 0 will result in the actual amount of gas spent lower than G1 and 8M. Therefore, BASEFEE will keep decreasing till 0, when there’s no more tax in the network, and the gas used will reach G1.

It can be proved that the above conclusion applies to all situation when G1 is below 8M.

While it gets interesting when G1 is higher than 8M

As shown in the graph above, if the G1 starts higher than 8M, as long as the BASEFEE is not so high that it keeps the gas consumption above 8M, the BASEFEE will keep increasing, until the point when the gas consumption decrease to 8M and stabilize there.

Therefore, when the demand is high enough (I consider here G1 higher than 8M), the BASEFEE is essentially a dynamic tax regime to lock the system throughput.

This is what my economic analysis on EIP-1559. Obviously, there’s something more than economics.

Other issues with EIP-1559

Currency value

I know you are probably still wondering something else: after burning all that much ETH, which lower the supply, will we… go to MOON?!

Well, I would say, it will be lucky if it doesn’t go to HELL. The truth is, currency value going up with lower supply ONLY happens when it’s trading value decrease less proportionally, remains or even increase during the same period. In the case of EIP-1559, who guarantees you that the trading value won’t decrease?

I’ll bet my 1 ETH on trading value decrease more than the supply.

Think about it another way, if burning does more good than harm, why do we need to implement such burning requirement? People would have just been happily burning their ETH!

Limited state increase

Serious thinker might ask another question here: the EIP-1559 limited gas consumption, which leads to a less state increase, but does it worth it with the tradeoff to tolerate the dissipated value?

I need to admit that it is a side effect that the state increase would be limited, and that I cannot answer the question myself. However, we can think of the question from another perspective, which is, will this limit be any different if we simply lock the block gas limit forever? In this case, except for the taxation part, all the effect — — limiting transaction numbers, state increase — — are effectively the same.

Conclusion

I’m pretty confident that after this post, you will change your mind a bit if you were supporting the EIP-1559. As stated, it wouldn’t make users’ life easier, wouldn’t make transactions cost more predictable, because what was proposed is basically changing a name to what was not predictable — — renamed it “tips”; it wouldn’t improve the UX, because you need to explain to users where all the fees go, why are we burning it; moreover, most people supporting the fee market reformation consider today’s fee too high, but EIP-1559 will make it even higher, as it’s not only the miner you’re paying, the protocol will start to tax you now!!

Before I finish this post, I would like to write something a bit off topic.

It is sad that some people might never understand that some rules in a peaceful and civilized society that will not change with human wills, or better, might change but it will come with serious consequences. They probably don’t consider things like cooperation and think that they can have the society to function according to their will. If someone refuses to do so, that guy is either retarded or immoral. They might even hold the opinion that they enforce their will so that the society can be organized unitedly by this same concept, and that with this kind of unification, the society will become extraordinarily strong!

People that hold this kind of opinion don’t understand the basic social principles, nor are they interested in respecting other or the diversity itself. What they want is merely the redistribution of benefits, not able to take into account how to CREATE value, because most of them think that all trading are zero-sum games. The worst part is, they do whatever they can to extract benefits.

They often share their “opinion” to the public, keep repeating the same, meaningless nonsense without any critical thinking.

Back to the topic, and to the first graph you saw in the post. Of course the “fee” volatility will be much less after EIP-1559, as at the same time you’ve got another volatile “tip” graph! Well, maybe Eric thinks that one elegant graph is better than zero.

--

--

Ajian

website editor of EthFans.org, which is a Chinese Ethereum fans community. My email: hongji@ethfans.org