FLIP: Revisiting Flow Transaction fee

Status	Proposed
FLIP #	74
Author(s)	Kshitij Chaudhary (kshitij.chaudhary@dapperlabs.com)
Sponsor	Kshitij Chaudhary (kshitij.chaudhary@dapperlabs.com)
Updated	2023-03-23

Introduction

In October 2021, with FLIP-660 it was determined that the transaction fees on Flow should be proportional to the cost of effort to the network imposed by deployment of resources such as CPU and bandwidth. Fees were broken down into three components - ‘Inclusion Fee’ that accounts for the resources required to process a transaction due to the transaction’s core properties (byte size, number of signatures); ‘Execution Fee’ that accounts for the operational cost of running the transaction script, processing the results, sending results for verification, generating verification receipt, etc; and a Surge factor as a multiplicative factor to dynamically account for network pressure and market conditions.

Despite the trifurcation however, transaction fees on Flow continue to remain incredibly low, posing threats of network attack, and potentially dis-balancing demand for transaction processing against network-capacity in future. This FLIP proposes an increase in transaction fee and seeks community feedback on the proposed changes.

Problem Statement

Low transaction fee poses threats of attack, necessitates continued token-issuance by the protocol (inflation), and makes FLOW use-cases seem peripheral vis-a-vis other networks.

In early days of the Flow network, when the number of transactions were fewer, transaction fees were intentionally kept low to encourage network usage. Today however, with Flow’s purpose-built compute platform and web3 utility for the masses driving a growing user-base (18M+ accounts) on the network consistently sending 6M+ transactions in an epoch (week), there is a need to revisit the fees.

Today, the cost of attacking Flow with an inordinate number of transactions is incredibly low, making the network susceptible to attack by malicious actors. Additionally, as exhibited in the table below, despite ~900K daily transactions (sometimes exceeding 1M), total weekly fee collection (~50 FLOW) accounts for merely 0.004% of epoch rewards (1.3M+ FLOW tokens) paid out to node operators and delegators. During rewards pay-out, the protocol first draws from the transaction fee pool, thereafter issuing new tokens to achieve the target reward rate for each epoch (explained here). Low fee collection thus remains a key factor behind Flow’s 5% annual inflation. Note that a detailed discussion on Flow inflation is out-of-scope for this FLIP and will be comprehensively addressed in a separate proposal.

Finally, Flow’s low transaction fee unfortunately makes FLOW use-cases less perceptible to many users. Competing networks such as Solana, NEAR and Avalanche for instance have 300x, 450x, and 1,300x average fee per transaction respectively vis-à-vis Flow, while on ETH, the average fee is ~480,000x (see table below). Given FLOW’s multiple use-cases including staking, delegating, paying transaction fees, paying storage fees, trading assets and participating in experiences built on Flow, it is crucial that FLOW token’s utility is priced appropriately to reflect and surface its value-proposition to users.

Network	Daily total txns (est.)	Daily total fee (~$)	Avg fee/ txn (~$)	V/s Flow txn fee
Avalanche	2,000,000	$ 22,375	$ 1.12E-02	1,324 times
Bitcoin	305,700	$ 296,060	$ 9.68E-01	114,612 times
BNB	2,800,000	$ 551,696	$ 1.97E-01	23,318 times
Cardano	77,300	$ 9,599	$ 1.24E-01	14,696 times
ETH	1,000,000	$ 4,063,415	$ 4.06E+00	480,882 times
NEAR Protocol	420,000	$ 1,594	$ 3.80E-03	449 times
Solana	17,690,000	$ 44,206	$ 2.50E-03	296 times
Flow	947,558	$ 8.01	$ 8.45E-06	N/A

Market prices used are Year-to-date (YTD) moving averages
Average fee per transaction is calculated by daily total fee/ daily total transactions
The last column simply divides average transaction fee for a network to the average transaction fee on Flow

Proposal

Today, the functional specification of transaction fee on Flow looks as follows.

Transaction fee = {inclusion fee + (execution effort * unit cost)} x surge

Inclusion fee = 1E-6 FLOW (currently constant, but dynamic inclusion fee FLIP was proposed last year);
Execution Effort Unit Cost = 4.99E-08 FLOW (currently constant);
Surge = 1.0 (currently constant);
Execution Effort is variable based on transaction type, as shown here.

It is proposed that

Inclusion fee is maintained as a constant, but increased 100x, from 1E-6 FLOW to 1E-4 FLOW for all transaction types
Unit Cost of Execution Effort is increased 100x, from 4.99E-8 FLOW to 4.99E-06 FLOW

Transaction type	Effort	Current fee (FLOW)	Proposed fee (FLOW)	Proposed increase
Empty Transaction (inclusion fee only)	0	1.00E-06	1.00E-04	100 times
Create 1 account	43	3.15E-06	3.15E-04	100 times
Create 10 accounts	433	2.26E-05	2.26E-03	100 times
Add key to account	0	1.00E-06	1.00E-04	100 times
FT transfer	17	1.85E-06	1.85E-04	100 times
Deploying a 50kB contract	574	2.96E-05	2.96E-03	100 times
Mint a small NFT	18	1.90E-06	1.90E-04	100 times

Impact

First, if the proposal is accepted and implemented, users would pay a 100x transaction fee vis-à-vis today. Simply put, minting 1 million small NFTs on Flow costs ~1.9 FLOW today; with increased fee however, users would have to pay ~190 FLOW for such volume. By generally increasing the cost of mass attack by 100x, the proposal is aimed at disincentivizing malicious actors and enhancing network’s security and resilience against attacks.

Second, while this proposal by itself will only marginally reduce inflation (from ~5% to ~4.98%), it is a crucial step in the direction of addressing inflation on Flow in the long-term. As an example, the graph below exhibits the inflation rate with status-quo vis-a-vis with proposed fee increase, demonstrating a meaningful reduction in inflation if number of transactions grow multifold (ceteris paribus). Besides increased TPS however, a pragmatic approach that revisits other factors such as minimum staking requirements and annual yields would help guide lower inflation on Flow, to be explained and addressed in a separate FLIP soon.

Third, this change would make FLOW’s use-cases more perceptible and established, further accentuating its stature as a token of value. The table below shows the average transaction fees of various networks vis-a-vis Flow in the current and proposed scenarios. Simply put, for every dollar that users would pay to transact on the Flow network, they currently pay $450 on NEAR, $300 on Solana, and $480,000 on ETH - for the same transaction(s). This differential would reduce 100x with the approval and implementation of this proposal, yet even with the implementation of the new fee structure, Flow would continue to remain an economically attractive network to transact on. Note that given FLOW market price’s high correlation with the listed tokens’ prices (”R” averaging 0.8+), even a change in FLOW token price or stochastic market volatility would not substantially alter the competitiveness of the proposed Flow fees compared to other networks, as can be seen in the last two columns of the table below.

Network	Current Avg txn fee	Current Multiple v/s Flow	Proposed avg txn fee	Proposed Multiple v/s Flow	Expected Multiple if FLOW/$ changes
ETH	$ 4.06E+00	480,882 times	$ 4.06E+00	4,808 times	4,040 times
BSC	$ 1.97E-01	23,318 times	$ 1.97E-01	233 times	182 times
Bitcoin	$ 9.68E-01	114,612 times	$ 9.68E-01	115 times	940 times
Solana	$ 2.50E-03	296 times	$ 2.50E-03	3 times	2.4 times
Avalanche	$ 1.12E-02	1,324 times	$ 1.12E-02	13 times	10.6 times
Cardano	$ 1.24E-01	14,696 times	$ 1.24E-01	147 times	120.5 times
NEAR	$ 3.80E-03	449 times	$ 3.80E-03	4.5 times	3.4 times
FLOW	$ 1.21E-05	N/A	$ 1.21E-03	N/A	N/A

Final Word

Transaction fees offer many benefits in Flow network’s economic design, including compensating the node operators for the resources deployed to process transactions and secure the network, protecting the network against spamming attacks, and generating long-term stability in the Flow economy via reduced inflation. In a mature state, transaction fees should allow the Flow blockchain to self-regulate transaction throughput in a way where it would always approach optimal throughput. This proposal however is meant to be a starting point for community discussions on how an increase in inclusion fees and unit cost of execution effort could lead to network-wide benefits including economically deterring network attacks and strengthening the overall FLOW economy. The community is invited to share feedback on this post and help shape the economics of FLOW.

Appendix 1

Comparison of Flow v/s Solana transaction fees (current scenario) for specific transaction types

The table below demonstrates how Solana fees associated with different transaction types fare against Flow today, ranging from ~100x for empty transactions to 3x for an average contract deployment.

	Flow txn fee (US$)	Solana txn fee (US$)	SOL : Flow fee ratio
Empty Transaction	1.11E-06	1.09E-04	98 X
Create 1 account	3.49E-06	1.09E-04	31 X
Create 10 accounts	2.51E-05	1.09E-03	44 X
Add key to account	1.11E-06	1.09E-04	98 X
FT transfer	2.05E-06	1.09E-04	53 X
Deploying a 50kB contract	3.29E-05	1.09E-04	3 X
Mint a small NFT	2.11E-06	1.09E-04	52 X

Note that Solana also has a prioritization fees which is not considered in the analysis below but is paid by ~48% of non-voting transactions.
Assumed 1 signature per transaction & Solana’s fee of 5000 Lamports per signature, where 1 Lamport = 1E-9 SOL; and 1SOL = US$ 21.07 (YTD moving average of SOL/US$ market price)
To create 10 accounts, 10 signatures for creation of 10 accounts were assumed
Solana-Flow multiple is calculated by dividing Solana txn fee for a transaction type by Flow txn fee for the same type