Understand Waiting Time in EIP-1559 Transaction Fee Mechanisms

Project information

• Authors:
• Acknowledgments: This research is supported as an independent project supported by Ethereum Academic Grants Round.
• Project Summary: Blockchain enables peer-to-peer transactions in cyberspace without a trusted third party. The rapid growth of Ethereum and smart contract blockchains in general — calls for well-designed Transaction Fee Mechanisms (TFMs) to allocate limited storage and computation resources. However, existing research on TFMs needs to consider the waiting time of transactions, which is essential for both computer security and economic efficiency, though often mentioned in different words, such as latency or market congestion. Integrating data from the Ethereum blockchain and mempool, we explore how two types of shocks affect transaction latency. First, we apply regression discontinuity design (RDD) to study the causal inference of the Merge, the most recent significant upgrade of Ethereum. Our results show that the Merge significantly reduces the long waiting time and market congestion. Then, we apply time series analysis to research the interaction of significant NFT drops and market congestion using Facebook Prophet, an open-source algorithm for generating time-series models. Our analysis reveals that the models could significantly improve the forecast of market congestion by considering NFT drops as a shock. Our study contributes to the interdisciplinary research at the intersection of computer security, distributed systems, market design, causal inferences, and time series analysis. Our findings shed light on a new direction of TFMs that prevent market congestion by efficiently forecasting ex-ante other than the current ex-post treatments.

Table of Contents

• data
• code
• figs

data

data
├── processed_data
│   ├── merge_sanctioned.csv
│   ├── sanctioned_after.md
│   ├── sanctioned_before.md
│   ├── sanctioned_transaction_from.csv
│   └── sanctioned_transaction_to.csv
└── queried_data
    └── sanction.csv

The Merge Data Range

• the total number of blocks=140,000
• the total number of transactions= 23,871,253

type	from	to
date_time	2022-09-03 20:10:39	2022-09-25 02:27:11
unix_timestamp	1662235839.0	1664072831.0
block number	15467393	15607393

The NFT Data Range

• the total number of blocks = 181,075 |type | from | to | |-----------------|-----------------------|-------------------| | date_time |2021-08-04 00:00:00 |2021-09-21 00:00:00| | unix_timestamp |1629763200.0 |1632182400.0 | | block number |13084679 |13265754 |

code

code
├── Data_Analyze
│   ├── Analyze_Data_NFT.ipynb
│   └── Analyze_Data_The_Merge.ipynb
├── Data_Process
│   └── Process_Data.ipynb
└── Data_Query
    ├── Query_Data_EthereumETL_Public.ipynb
    └── Query_Data_MemopoolGuru_Public.ipynb

figs

figs
├── Result 1: the merge
│   ├── 1.1. The waiting time (delay)
│   │   ├── 1.1.1. Visualization
│   │   ├── 1.1.2. Moving Average Smoothing
│   │   └── 1.1.3. Regression Discontinuity
│   ├── 1.2. The market congestion (efficiency)
│   │   ├── 1.2.1. Visualization
│   │   ├── 1.2.2. Moving Average Smoothing
│   │   └── 1.2.3. Regression Discontinuity
│   └── 1.3. The gas used per second (security)
│       ├── 1.2.1. Visualization
│       ├── 1.2.2. Moving Average Smoothing
│       └── 1.2.3. Regression Discontinuity
├── Result 2: The NFT drops
│   ├── 2.1. Visualization
│   └── 2.2. The Time Series Forecast
└── Result 3: The Confounding Factors
    ├── 3.1. The Network for Sanctioned Transactions
    ├── 3.2. Sanctioned Transactions with Unobserved Delay (Only found after the merge)
    └── 3.3. Relevance Statistics for the Delay
        ├── 3.3.1. observed and unobserved delay before and after the merge
        ├── 3.3.2. sanctioned and unsanctioned transactions before and after the merge
        └── 3.3.3. statistics for the delay of the sanctioned and unsanctioned transactions before and after the merge

Result 1: the merge

1.1. The Waiting Time (delay)

1.1.1. Visualizations

• Figure 1

Click here to see the interactive figure

• Figure 2:

Click here to see the interactive figure

• Figure 3:

Click here to see the interactive figure

• Figure 4:

Click here to see the interactive figure

1.1.2. Moving Average Smoothing

• Figure 12:

• Figure 13:

• Figure 14:

• Figure 16:

1.1.3. Regression Discontinuity Design

Tables refer to the paper

1.2. The Market Congestion

1.2.1. Visualization

• Figure 10:

• Figure 11:

1.2.2. Moving Aveage Smoothing

• Figure 17:

• Figure 18:

1.2.3. Regression Discontinuity Design

Tables refer to the paper

1.3. The Gas Used Per Second

1.3.1. Visualization

• Figure 19

• Figure 20

1.3.2. Moving Aveage Smoothing

• Figure 21

1.3.3. Regression Discontinuity Design

Tables refer to the paper

Results 2: NFT Drops

2.1.Visualization

Figure 19:

Click here to see the interactive figure

Figure 20:

Click here to see the interactive figure

Figure 21:

Click here to see the interactive figure

Figure 22:

Click here to see the interactive figure

Figure 23:

Click here to see the interactive figure

Figure 24:

Click here to see the interactive figure

2.2. The Time Series Forecast

The congestion ratio

Figure 25:

Figure 26:

Figure 27:

Figure 28

The continued congestion ratio

Figure 29

Figure 30

Figure 31

Figure 32

Result 3: the Comfounding Factors

####3.1. The Network for Sanctioned Transactions

Figure 33

Figure 34

Figure 35

3.2 Sanctioned Tranctions with Unobserved Delay (Only After the Merge)

OFAC Sanctioned address Interacted	hash
0xd90e2f925da726b50c4ed8d0fb90ad053324f31b	0x0fb9a4e23d7c504571daed1394feea8bcb823a18a2887e7cab54f7fa4ffa2981
0xd90e2f925da726b50c4ed8d0fb90ad053324f31b	0xb37f8b00993271dd97ac9acdb4d84cb31cd210a037286bcc4bc7887f0008a81f
0xd90e2f925da726b50c4ed8d0fb90ad053324f31b	0xd5c80d358016a1fe77260fa9cab88a706337c6f41f9112eca8840e9233794e82
0xd90e2f925da726b50c4ed8d0fb90ad053324f31b	0x52cd1910b0710e73a9fb85f9e13861ef828a3244a1fa03944e47adf38ac9f851
0xd90e2f925da726b50c4ed8d0fb90ad053324f31b	0xe4a84c77748a627a093f79c6abb4b037c60ae9f58977bab2c46d17a657f2afcd
0xd90e2f925da726b50c4ed8d0fb90ad053324f31b	0x094cd7a9dc72cf6b3415fd18223a075e631d4ee5bfc900b9293dbd135db24334
0xd90e2f925da726b50c4ed8d0fb90ad053324f31b	0x78c426d5b25290aa03153c728383f2d422c7a14a40a0a5e11bea67c104e5e295
0xd90e2f925da726b50c4ed8d0fb90ad053324f31b	0x9d4fcf6315d8fbf2e07dc71a2c81df8c86a2dd518ac76c48eaeb397e28e95ca8
0xd90e2f925da726b50c4ed8d0fb90ad053324f31b	0x0b9613f988da92901816bc76af743974032985b5eeeab70d5940c65b9f7faca6

3.3. Relavant Statistics for the Delay

3.3.1. observed and unobserved delay before and after the merge

merge	delay observed	delay unobserved
before the merge	1.19478e+07	326561
after the merge	1.10589e+07	537997

3.3.2. sanctioned and unsanctioned transactions before and after the merge

sanctioned	before the merge	after the merge
no	9.71282e+06	9.12023e+06
yes	1007	466

3.3.3. statistics for the delay for sanctioned and unsanctioned transactions before and after the merge

	count	mean	std	min	25%	50%	75%	max
(non-sanctioned, before)	9.71282e+06	974.101	37825.7	2e-05	7.794	15.1918	29.6824	5.44163e+06
(non-sanctioned, after)	9.12023e+06	4868.07	152958	6.4e-05	9.14709	13.7926	18.6293	6.29244e+06
(sanctioned, before)	1007	6161.2	131471	0.115805	10.0729	19.6879	37.1808	2.95086e+06
(sanctioned, after)	466	47.262	257.401	1.03641	11.8761	17.1533	22.5716	3107.16

Appendix

The Transaction Count

• Figure 9: Click here to see the interactive figure

The Gas Used Percentage

Visualization

• Figure 5

Click here to see the interactive figure

• Figure 6:

Click here to see the interactive figure

• Figure 7:

Click here to see the interactive figure

Moving Aveage Smoothing

• Figure 8:

Click here to see the interactive figure

• Figure 15:

Regression Discontinuity Design

Tables refer to the paper