The Ethereum Foundation laid down a challenge to the community to analyze the performance and behavior of the chain before and after the Merge. So, here is our possibly naive analysis, performed without having done any prior work in the Ethereum ecosystem.

The primary tool we used was the Alchemy API to retrieve blocks and transactions, as well as some additional libraries to examine miner rewards.

Our first impression was that much remained the same:

Aside from a slight decreasing trend in the baseFeePerGas, the primary metrics associated with a block appear remarkably stable across the Sept 15, 2022 merge boundary line.

Diving in a bit further, we noticed one striking difference: the number of unique miners increased dramatically immediately after the merge.

10k Blocks 8/27/22 generated by 54 miners (1696802 transactions)

10k Blocks 9/16/22 generated by 679 miners (1887472 transactions)

Of the 54 miners sampled from 8/27, two-thirds (36) were participating on 9/16.

Miner Differences

In order to investigate this difference, we retrieved a larger sample of 519059 blocks from before and after the merge, skipping over the actual merge day. The increase in number of miners holds up over time, examined here in chunks of 1000 blocks.

It was suggested by some other writers that miners who were working before the merge are no longer participating. We also tested this proposition, and in fact although for a short period after the merge some “pre-merge miners” participated, this dropped off rapidly:

Obviously the method of gaining rewards changed after the merge, but we were curious if the amount of reward changed as well. This was a more time-consuming calculation, and only samples were taken.

From a small number of samples, the reward per block seems roughly stable at around 2.04 ETH before and after the merge

So likely, the reason for the non-intersection of miners before and after the merge is due to the change in technology and not a reduction in rewards.

However one possible exception is blocks that have uncles. While we are not deep enough into the ETH ecosystem to be entirely sure, it seems probable that after the merge the lack of uncles is a positive thing for the chain, it might reduce rewards due to minors. However, in our small sample inspection, uncle blocks had only a slightly larger reward of about 2.07ETH.

In any case, it also seemed significant to note that

No More Uncles

We also wanted to examine the velocity of blocks and transactions, as this is one of the primary advantages promised by advances made possible by the merge. So far, the speedup is slight - 12.9 transactions per second since the merge, vs 11.7 transactions per second before.

(Pdb) dfa.transactionCount.sum()/(dfa.secs.max() - dfa.secs.min())
12.890979125248508
(Pdb) dfb.transactionCount.sum()/(dfb.secs.max() - dfb.secs.min())
11.712863266492299

Summary

In summary, the Ethereum blockchain has been stable across the merge boundary, with statistically insignificant changes to most block variables, with the exception of dramatic changes to uncles and miners.

Transaction Count per Block: essentially unchanged

Gas Used per Block: essentially unchanged

Base Fee Per Gas: slight downward trend

Number of Unique Miners: dramatic increase sustained over time

Uncles: Not seen since shortly after the merge

Rewards per Block: essentially unchanged

Overall Transaction Speed: slight increase

We are looking forward to the further changes unlocked by the change to proof of stake that should enable further improvements to the chain.

Reference Code:

The code in this repo was hacked rapidly and some of the figures above were generated on the fly in an interactive session. However, the basic functionality required to generate them is all present here.

https://github.com/Whats-Cookin/eth-merge-data

I prefer to deal with people I know and trust, to work on things I care about, to buy products that were made in fair working conditions. I strongly want never to buy things that murderers profited from, yet today I often do so. An anonymized marketplace does not allow me to easily act “rationally” based on my actual values. A richer ecosystem of information would provide much higher utility, than the current common model of anonymized transactions.

In math, the shape of the universe follows from the axioms chosen. An assumption of scalar rationality in which all values are transactional and anonymizable is not wrong mathematically, it simply determines the shape of the universe described.

In science, we try to describe the observed shape of the world. Models and theories are tested by their ability to predict behavior in the real world. Hand-waving away large swaths of human motivation and observed behavior as "irrational" is a religious, not scientific approach.

In engineering, we try to shape the world. What shape do we want? For very clearly, tools that only allow anonymous transactions lead to a world in which everything is for sale.

The exciting thing in our time is not that we can automate markets, but that we can actualize language and intention. SBTs are a good starting point, that allow us to efficiently encode the values and relationships, the specific judgements and experiences that in fact drive our behavior.

I wrote more about this some time back: Can’t Buy Me Love: An Argument for Implementing Illiquidity which discusses encoding trust, changing cooperative structures and using earned tokens for governance, and offers computational models of love and corruption, among other things.